<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[GreenInfo Network - Medium]]></title>
        <description><![CDATA[We create, analyze, visualize and communicate information in the public interest. Learn more at www.greeninfo.org - Medium]]></description>
        <link>https://medium.com/greeninfo-network?source=rss----7b016ec69299---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>GreenInfo Network - Medium</title>
            <link>https://medium.com/greeninfo-network?source=rss----7b016ec69299---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Mon, 16 Nov 2026 20:30:08 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/greeninfo-network" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Introducing GreenInfo’s New Executive Director]]></title>
            <link>https://medium.com/greeninfo-network/introducing-greeninfos-new-executive-director-845883422615?source=rss----7b016ec69299---4</link>
            <guid isPermaLink="false">https://medium.com/p/845883422615</guid>
            <category><![CDATA[leadership]]></category>
            <category><![CDATA[maps]]></category>
            <dc:creator><![CDATA[GreenInfo Network]]></dc:creator>
            <pubDate>Wed, 29 Oct 2025 21:39:32 GMT</pubDate>
            <atom:updated>2025-10-29T21:55:19.128Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/832/1*WeJfEO5HPbzrdk9uOmKHDw.png" /><figcaption>Matt Lamborn</figcaption></figure><p>On December 1, 2025, Matt Lamborn will become GreenInfo Network’s third executive director, succeeding Dan Rademacher, who has led the organization since December 1, 2015.</p><p>Matt brings over 20 years of diverse geospatial experience spanning urban planning, agriculture, emergency operations, and geospatial intelligence. Throughout his career, he has excelled at translating complex technical concepts into accessible solutions that resonate with stakeholders at all levels.</p><p>He is no stranger to parks and trails — he helped design many parks, trails, and open spaces that are now documented in GreenInfo’s California Protected Areas Database (CPAD) and Conservation Easements Database (CCED).</p><p>Most recently, Matt served as GIS Departmental Coordinator for Napa County’s Planning, Building, and Environmental Services Department, where he designed, built, and managed a highly-respected GIS program.</p><p>His nonprofit leadership experience is equally impressive, including roles as president of the California Chapter of URISA (now Cal Geospatial Professionals Network), board member for Angwin Disaster Airlift Response Team, and Geospatial Intelligence Program Manager for Night Owl Reconnaissance, a nonprofit dedicated to locating missing persons.</p><p>Whether engaging audiences at conferences, advocating for his organization, or managing day-to-day operations, Matt will combine vision with tactical expertise to drive GreenInfo forward.</p><p>“I am beyond excited and incredibly honored to serve as GreenInfo Network’s next executive director,” Matt Lamborn said. “ I owe a tremendous amount of gratitude to the executive directors who came before me: Larry Orman, who had the vision to create this incredible nonprofit to practice GIS in the public interest, and his successor Dan Rademacher, who doubled GreenInfo Network’s reach and capacity. During this transition, my focus will be on honoring the long-standing relationships that have made GreenInfo special, starting with our staff, board of directors and the many clients who have put their trust in us over the years. I’m eager to build on this strong foundation to explore new ways GreenInfo Network can continue to use information and GIS to serve communities and organizations in the public interest.”</p><p>This leadership transition marks an exciting milestone as GreenInfo begins its 30th year of operation. “Matt is the ideal person to build on the solid foundation that GreenInfo’s founder Larry and I built over the past three decades,” said Dan Rademacher. “His experience in government, nonprofit, and consulting spaces will help GreenInfo and all of our clients continue to create maps, tools, and datasets that make our communities more equitable and sustainable.”</p><p>GreenInfo looks forward to extending our record of long-term, stable leadership. “Dan Rademacher has been an outstanding leader of GreenInfo Network for the past 10 years. The Board and staff are grateful for his vision and guidance, and we wish him the best,” said Michael McLaughlin, Greeninfo’s board chair. “Looking forward, Matt Lamborn brings tremendous skill and experience to his new role as Executive Director. We couldn’t be more excited about the future of Greeninfo under his leadership.“</p><p>Starting December 1, you can reach Matt at matt [at] greeninfo.org as he settles into his new role and connects with the many people who put the “network” in GreenInfo Network.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=845883422615" width="1" height="1" alt=""><hr><p><a href="https://medium.com/greeninfo-network/introducing-greeninfos-new-executive-director-845883422615">Introducing GreenInfo’s New Executive Director</a> was originally published in <a href="https://medium.com/greeninfo-network">GreenInfo Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Reflections on 10 years of leading GreenInfo]]></title>
            <link>https://medium.com/greeninfo-network/reflections-on-10-years-of-leading-greeninfo-338eee6ea1b3?source=rss----7b016ec69299---4</link>
            <guid isPermaLink="false">https://medium.com/p/338eee6ea1b3</guid>
            <category><![CDATA[maps]]></category>
            <category><![CDATA[leadership]]></category>
            <dc:creator><![CDATA[GreenInfo Network]]></dc:creator>
            <pubDate>Wed, 29 Oct 2025 21:34:27 GMT</pubDate>
            <atom:updated>2025-11-02T23:25:34.985Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sWWi3aoWS0YYlgznawaL9A.jpeg" /><figcaption>Most of the GreenInfo team at our annual staff retreat in spring 2025.</figcaption></figure><p>On December 1, 2015, I became the second executive director of GreenInfo Network, succeeding founder Larry Orman. Exactly 10 years later, on December 1, 2025, GreenInfo’s third executive director, <a href="https://blog.greeninfo.org/introducing-greeninfos-new-executive-director-845883422615">Matt Lamborn</a>, will take my place. I’m headed to iNaturalist to become their Head of Product, where I’ll be hanging up my Executive Director hat and diving deep on a single product.</p><p>My work over the last decade was the perfect confluence for me: Leading a great team, working with a diverse group of smart clients, and building things that make the world a bit better, a bit more equitable, one step at a time.</p><p>A lot has changed since 2015 for GreenInfo as an organization: We’ve grown the budget and team, adding new design, design research, and project management roles on the interactive team, and creating new GIS developer roles. We’ve broadened our web design and development practices, introduced new data automation methods, and vastly increased our focus on making our work fully accessible.</p><p>What hasn’t changed is the wide array of projects created by an amazing group of people — our staff, close contractors, and our clients. I’m going to shine a light on some of those folks who often work in the background.</p><p>GreenInfo has an incredible history of bringing people together over the long haul. Larry Orman led GreenInfo for 19 years and mentored me in my first few years as executive director. The median tenure of our staff is over 10 years, and we have several people at or above 20 years. We have clients who have worked with us for most or all of the 29 years since Larry and Louis Jaffé founded the organization in 1996.</p><p>Indeed, more than half of GreenInfo’s current staff were here when I arrived in 2015. A big thanks to Anne Katzenberg, Amanda Kochanek, Maegan Leslie Torres, Tim Sinnott, Stephanie Ding, Gregor Allensworth, Tom Allnutt, and Dariya Draganova for their collective 138 years of experience at GreenInfo Network. The role of executive director is so much easier when you have staff with that kind of history with each other, our clients, and the work.</p><p>Which is not to take away from the remarkable folks who’ve joined in the last decade and are key parts of the staff now: team managers Stephen Smith and Kristina Yamamoto, design researcher María del Carmen Lamadrid, GIS analysts Joseph Stout and Andrew Nham, and interactive project manager Robel Mengistu.</p><p>Along the way, associate directors Maianna Voge, Anjuli Jain Figueroa, and Vanessa Knoppke-Wetzel were thought partners for me in the often hard work of leading a small nonprofit organization. Folks like Gina Rosa Cova, Yessenia Chiau Zhang, Chris Henrick, Saba Gebreamlak, Amanda Damewood, and Kimberly Becerril blazed through with brilliance and excellent projects we’re still building on today. Jayita Bhojwani came to us through the Conservation Lands Network and proceeded to build a new code-based analysis and automation practice with Amanda Kochanek and Dariya Draganova that fundamentally changed how we manage public lands data.</p><p>Almost a dozen board members have provided key support for me and the organization since I took over in 2015. Michael McLaughlin, our longtime board chair, helped me see the importance of data, not just in the work we do, but in the management of that work. Mike Migurski, Karin Tuxen-Bettman, Maggi Kelly, Vyki Englert, Travis White, Madelyn Glickfeld, Antwi Akom, Aekta Shah, and Dick Cameron all helped me stay connected to various parts of the geospatial, cartographic, and civic tech worlds. Jaymee Go has been my “Sacramento whisperer,” helping us keep up with the ebbs and flows of policy and funding out of the capital. And Louis Jaffé has always been there to make sure I kept the long view in mind.</p><p>Beyond staff and board, I’ve had deep and meaningful collaborations with so many of our clients.</p><p>Jon Christensen of UCLA has been a collaborator since before day 1, and we’ve developed innovative data-driven methods to improve equitable access to parks and nature, culminating in an award-winning park needs assessment for Los Angeles this year.</p><p>It was a joy to see Annie Burke manage the rebirth and reimagining of TOGETHER Bay Area, and to work with her and Tom Robinson on the Conservation Lands Network, now beginning its third iteration.</p><p>The Resources Legacy Fund continued to bring us amazing opportunities, as they have since 1996. Among the projects supported by RLF, Matt Armsby and Alfredo Gonzalez helped bring critical funding to our work on California’s 30x30 campaign and innovative park equity research with The Wilderness Society and Jon Christensen.</p><p>We continued to work for park equity through our long partnership with the Office of Grants and Local Services within California State Parks. Sam Cadeaux, Natalie Bee, Richard Rendon, Viktor Patiño, Lee Butterfield, and Sedrick Mitchell were a constant source of inspiration and deep collaboration as we developed new ways to think about and measure the park and recreation needs of Californians. Two incredible contractors, print designer Annabelle Ison and accessibility expert Sarah Fossheim, joined us in this work.</p><p>Our work with California State Parks was made possible by our longtime stewardship of the California Protected Areas Database (CPAD) and Conservation Easements Database (CCED). Amanda Kochanek has developed and led this public lands work for nearly 25 years. CPAD and CCED became even more important when the governor and legislature set the goal of conserving 30% of the state’s land and waters by 2030. That 30x30 goal led us to work closely with the California Natural Resources Agency, first with Jennifer Norris, and now with Meghan Hertel and Madeline Drake.</p><p>On my first or second day at GreenInfo in 2015, I started working on the Bay Area Greenprint. The core team of The Nature Conservancy’s Elizabeth O’Donoghue and Carrie Schloss, TOGETHER Bay Area’s Tom Robinson, and Adam Garcia (through several job changes) continue to meet and iterate on that tool 10 years later. And we’ve built similar tools for the northern Sierra Nevada and Osceola County, Florida.</p><p>There are so many more — Niles and Allison at the East Coast Greenway; Jeff, Minh, and many others at OpenHistoricalMap; Justin at All About Redistricting; Christine at NYC Crashmapper, and more, too many to list them all.</p><p>It’s been an incredible honor to lead this organization for 10 years. I look forward to seeing GreenInfo and its clients thrive for many years to come.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=338eee6ea1b3" width="1" height="1" alt=""><hr><p><a href="https://medium.com/greeninfo-network/reflections-on-10-years-of-leading-greeninfo-338eee6ea1b3">Reflections on 10 years of leading GreenInfo</a> was originally published in <a href="https://medium.com/greeninfo-network">GreenInfo Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Accelerating Legal Document Review with AI]]></title>
            <link>https://medium.com/greeninfo-network/accelerating-legal-document-review-with-ai-f62ab0a9dc46?source=rss----7b016ec69299---4</link>
            <guid isPermaLink="false">https://medium.com/p/f62ab0a9dc46</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[elections]]></category>
            <category><![CDATA[law]]></category>
            <category><![CDATA[analysis]]></category>
            <dc:creator><![CDATA[Stephen R. Smith]]></dc:creator>
            <pubDate>Tue, 27 Aug 2024 15:06:09 GMT</pubDate>
            <atom:updated>2024-08-27T15:03:36.535Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*q0OgrtrIjP4Ne-5a" /><figcaption>Photo by <a href="https://unsplash.com/@homajob?utm_source=medium&amp;utm_medium=referral">Scott Graham</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>The College of William &amp; Mary Law School has developed a website called “eBenchbook” with the goal of making all election-related legal, regulatory, and statutory documents readily available without needing expensive legal subscriptions to services like Westlaw or Lexis Nexus. The initial site focused on five states, intending to expand later.</p><p>In 2021, GreenInfo began developing the full eBenchbook website that would expand to all 50 states. With legal documents — statutes, regulations, cases, and advisory opinions — provided by FastCase (now vLex), we created a processing system to convert the data from XML to HTML and load them onto the website for easy searching and reading. However, once on the site, users needed a way to easily browse the data by topic or tag.</p><p>W&amp;M developed a tagging system that allows users to find relevant documents quickly. Each semester, they assigned a handful of law students to read, review, classify, and tag the 50,000+ documents provided by FastCase. This was a tedious task for law students, but with a new website launch looming, there was an urgent need for a faster solution.</p><p>Enter ChatGPT. We wanted to know whether this advanced language model would offer a faster review process, reduce the workload on humans, or enhance consistency through precise prompts. In this context, achieving efficiency and accuracy is critical, so we developed a set of careful procedures to ensure only accurate information is published on the site.</p><p>In this blog post, we’ll explore ChatGPT’s transformative potential in document review, the hurdles faced, and the outcomes of leveraging this innovative technology.</p><h3>Testing the Hypothesis</h3><p>To expedite the review process of over 50,000 legal documents, we formulated a hypothesis centered on the capabilities and limitations of large language models. Our theory was built on several key points, each aimed at addressing the specific challenges of manual review and enhancing overall efficiency and accuracy.</p><p><strong>Accelerating the Review Process with ChatGPT</strong></p><p>Our primary hypothesis posited that ChatGPT, or another similar language model, could drastically reduce the time required for initial document review and classification, allowing law students to focus on reviewing and enhancing the AI’s answers with their expert human judgment.</p><p><strong>Achieving Consistency Through Detailed Prompting</strong></p><p>Another cornerstone of our hypothesis was the potential for ChatGPT to deliver more consistent results when guided by detailed and specific instructions. By crafting precise prompts, we aimed to harness the AI’s ability to follow set guidelines, thereby reducing variability.</p><p><strong>Pre-screening for Election Applicability</strong></p><p>To further streamline the process, we theorized that pre-screening documents for their relevance to elections could substantially reduce the number of documents needing detailed review. This step would involve using ChatGPT to filter out clearly irrelevant documents, allowing the remaining potentially relevant documents to undergo a more thorough examination.</p><p><strong>Achieving Consensus Through Multiple Reviews</strong></p><p>We also speculated that having ChatGPT review each document multiple times could lead to a consensus, thereby improving the reliability of the AI’s classifications. We chose to evaluate each document 10 times each, allowing the model to make separate, independent judgments. This iterative approach mimics a peer-review process, ensuring higher accuracy before human review.</p><p><strong>Ensuring Quality with Human Oversight</strong></p><p>Despite ChatGPT’s promising capabilities, we recognized the importance of human oversight to ensure the highest-quality final product. Law students and legal experts still play a critical role in the review process, particularly for complex or ambiguous documents. The combination of AI-driven initial review and human quality control aimed to blend the best of both worlds, leveraging AI for speed and consistency while maintaining the nuanced understanding and judgment that only humans can provide.</p><p>By testing this hypothesis, we aimed to uncover a robust and efficient method for legal document review that could be applied to our current project and potentially to other large-scale data tasks. The rest of this blog post will explore the implementation, challenges, and results of integrating ChatGPT into our document review process.</p><h3>LLM Challenges</h3><p>Several issues and challenges emerged as we used ChatGPT for our document review process. These obstacles highlighted the complexities of integrating an AI solution into a workflow requiring deep technical human knowledge and underscored the importance of careful planning to ensure accurate outcomes.</p><p><strong>Cost of API Credits</strong></p><p>One of the primary concerns was the cost associated with API credits. ChatGPT operates on a token-based system, charging per million tokens processed. Each word is roughly one token. Early estimates of our API costs were reasonable but did not align with the costs billed, so we had to look closely at the billing details. Upon closer inspection, we realized that the pre-instruction prompt required approximately 7,000 tokens for each document reviewed, greatly increasing the total tokens evaluated. Given that documents could be up to 80,000 tokens in length, these costs quickly added up. Given this scale factor and the number of tokens remaining in the pool of documents, our total estimate for evaluating all of the documents was around $14,000.</p><p><strong>API Limits and Token Management</strong></p><p>Early on, the API limits also posed a constraint, extending processing time. OpenAI uses a tier system based on the amount of money you’ve spent on the platform. For initial testing, we were operating at Tier 1 (&lt; $50 spent) with a limit of 30,000 tokens per minute (TPM). To handle the volume of our document review project, we had to upgrade to Tier 3, which allows 800,000 TPM but becomes available only after investing significantly more in the platform.</p><p><strong>LLM Hallucinations and Inaccuracies</strong></p><p>A notable challenge with using language models like ChatGPT is the phenomenon of “hallucinations,” where the AI generates incorrect or nonsensical information. In our case, this resulted in several issues:</p><ul><li><strong>Wrong Tags</strong>: ChatGPT sometimes assigns tags to documents that don’t exist in the list of tags.</li><li><strong>Wrong Hierarchy</strong>: The tags were hierarchical, where a “parent” tag like “Voting” might have several sub-tags like “Vote by Mail.” These were frequently misapplied, with the most common issue being hallucinated tags under a parent tag.</li><li><strong>Typos and Formatting Errors</strong>: Despite being given specific instructions on returning a short, structured reply, the AI occasionally introduced typos and formatting inconsistencies, such as unnecessary spaces or misplaced dashes.</li></ul><p>Despite these challenges, our experience with ChatGPT underscored both the potential and the limitations of using AI for legal document review. While issues such as cost, API constraints, and hallucinations require careful management, they do not diminish the value that AI can bring to the process. With continued refinement and strategic implementation, AI can complement human expertise rather than replace it.</p><h3>Round 1 Results: Surprisingly Good!</h3><p>In the initial phase of our ChatGPT integration, we reviewed 80 documents, investing $50 in API credits. This preliminary round was crucial for evaluating the effectiveness and accuracy of AI in handling the complex task of reviewing legal documents.</p><p>This analysis phase was followed by a human review by seasoned law professors, who evaluated the results for accuracy.</p><p><strong>Initial Document Review</strong></p><p>Out of the 80 documents reviewed, ChatGPT identified 61 as “not applicable” to our criteria. Upon legal review, 59 of these determinations were correctly tagged. Interestingly, one document was an unusual case that even a seasoned law student would have flagged as peculiar. Only one document was incorrectly classified as not applicable.</p><p>A few discrepancies arose among the 19 applicable documents, which were subsequently tagged. One of these documents was not applicable, and four were not legal case documents but judicial orders, which are excluded from our site. This highlighted an area where further refinement in AI instructions was needed.</p><p><strong>Tag Selection Quality</strong></p><p>Among the correctly classified documents, the quality of the tag selection varied. ChatGPT’s tagging accuracy was as follows:</p><ul><li>Good Quality Tags: 7 documents</li><li>Fair Quality Tags: 7 documents</li></ul><p>Even the tags classified as “fair” were sufficiently close to guide law students toward the best possible selections. Overall, 75% of the tags returned by ChatGPT were valid and aligned with our instruction set. While not perfect, the remaining tags were close enough to assist law students in their quality control process.</p><p><strong>Insights and Adjustments</strong></p><p>These initial results offered valuable insights. The AI demonstrated a reasonable ability to discern applicable documents and provided accurate primary tagging. However, the misclassification of judicial orders and the occasional incorrect applicability judgment indicated areas for improvement.</p><p>Despite the challenges, the first round of results was promising. It confirmed that ChatGPT could significantly accelerate the document review process, provided proper checks and balances were in place. Moving forward, our focus will be on fine-tuning the AI instructions and improving the efficiency of the review process to handle larger volumes of documents more accurately.</p><h3>Testing GPT-4o-mini: A Cost-Effective Alternative?</h3><p>In July 2024, we tested a more cost-effective version of the AI, GPT-4o-mini, to see if it could match or exceed the performance of its predecessor at a fraction of the cost. With a significantly lower price point of $0.15 per million tokens (compared to $5 per million tokens for GPT-4o), this test aimed to determine if budget savings could be achieved without sacrificing too much accuracy.</p><p><strong>Document Review with GPT-4o-mini</strong></p><p>We reviewed the same set of 80 documents using GPT-4o-mini. Despite the cost advantage, the results were mixed:</p><ul><li><em>Tagging Issues</em>: GPT-4o-mini occasionally returned more than the five requested tags. Additionally, sometimes the AI returned “No Applicable Tags” as a tag value rather than replying “No” to the “applicable” question.</li><li><em>Typos</em>: The model suffered the same issues as 4o, occasionally returning values that needed to be cleaned up later. Leading or trailing spaces were occasionally an issue, as well as answering “Applicable” instead of “Yes” as instructed.</li></ul><p><strong>Comparing Concurrence on Applicability</strong></p><p>The concurrence rate on document applicability was a key metric for evaluating the cheaper AI’s performance. We felt more confident when the AI analyzed the document similarly over multiple runs. With 4o, only two documents didn’t have complete agreement across all runs. However, GPT-4o-mini showed a decline in accuracy:</p><p><em>100% Self-Concurrence</em>:</p><ul><li>GPT-4o: 2 out of 80 didn’t have 100% agreement.</li><li>GPT-4o-mini: 13 out of 80 didn’t have 100% agreement.</li></ul><p><em>Inter-model Concurrence:</em></p><ul><li>Only 58 out of 80 documents perfectly aligned between 4o and Mini.</li><li>If we consider “majority” concurrence, 69 out of 80 are aligned between the two models.</li></ul><p><strong>Comparing Tagging Performance</strong></p><p>The differences between the two models became evident when comparing tag selections:</p><ul><li><strong>GPT-4o </strong>generated 82 unique tags, of which 21 were invalid. Despite these inaccuracies, the inaccurate tags were still very close to other tags in the list, which still benefits law students in their review.</li><li><strong>GPT-4o-mini </strong>generated 208 unique tags, but 133 were invalid. Only 27% of the tags were perfectly matched, significantly increasing the burden on law students to filter out the erroneous tags.</li></ul><p><strong>Insights and Next Steps</strong></p><p>The cost savings offered by GPT-4o-mini came with a notable trade-off in accuracy and reliability. The excessive and incorrect tags and the misclassification of document applicability highlighted the challenges of using a less sophisticated model.</p><p>These findings reinforced the need for a balance between cost and quality. While GPT-4o-mini provided an affordable alternative, the additional workload required to correct its errors negated some financial benefits. Moving forward, our approach would involve optimizing AI prompts and refining pre-screening methods to improve the accuracy of more economical models.</p><h3>Future Work: Refining Integration</h3><p>Our initial trials with ChatGPT and GPT-4o-mini have provided valuable insights and highlighted areas for improvement. To fully harness AI’s potential in legal document review, we plan to implement several enhancements and refinements to our process.</p><p><strong>Tailored AI Instructions</strong></p><p>One key area for future work involves crafting more precise AI instructions. We will give a shorter instruction set for the initial assessment, including a more precise definition of relevant documents and election law.</p><p><strong>Detailed Legal Instructions</strong></p><p>To enhance accuracy, we will incorporate more detailed legal instructions about including or excluding specific document types. This will involve closer collaboration with legal experts to refine the criteria and ensure the AI can accurately distinguish between relevant and irrelevant documents, eliminating irrelevant data such as “orders” or other non-applicable documents.</p><p><strong>Efficient Token Usage</strong></p><p>Another area of focus will be optimizing token usage. By evaluating documents fewer times — potentially only five times — we can maintain high accuracy while reducing the costs associated with token consumption. This approach will involve balancing the need for a thorough review with the financial constraints of using AI at scale. We may also use GPT-4o-Mini for applicability review, then switch to GPT-4o for tagging.</p><p><strong>Process Speed Up</strong></p><p>To speed up the review process, we will refine our workflow to prioritize critical tasks:</p><ul><li><em>Evaluate Applicability</em>: Conduct a preliminary assessment to filter out non-relevant documents.</li><li><em>Evaluate Tags</em>: Apply tags based on the refined criteria and tailored instructions.</li></ul><p><strong>Human Oversight and Final Review</strong></p><p>Despite the advancements in AI technology, human oversight remains crucial. After the AI has completed its initial review and tagging, law students and legal experts will conduct a detailed review to ensure the final product meets the highest standards. This combination of AI efficiency and human expertise will provide a robust and reliable document review process.</p><h3>Conclusion</h3><p>Experimenting with ChatGPT for legal document review has revealed significant potential benefits and numerous insights. One of the standout realizations is AI’s valuable role in a human-in-the-loop workflow. Even in highly technical and detailed tasks like legal document review and classification, AI can significantly enhance efficiency and consistency when paired with human oversight.</p><p>However, the cost of using the newest and most advanced AI models at scale remains a considerable challenge. The high expenses associated with large language models like GPT-4o underscore the need for careful financial planning. Nevertheless, as we’ve seen over the past few years, these costs will decrease, making AI solutions more accessible and practical for a broader range of applications.</p><p>While smaller language models like GPT-4o-mini offer a more cost-effective alternative, they are not without limitations. These models can still provide value in specific, limited use cases where the complexity of tasks is manageable. However, larger models consistently outperform their smaller counterparts for more intricate and demanding tasks, offering higher accuracy and reliability.</p><p>Ultimately, the key to successful AI integration lies in meticulous planning, continuous evaluation, and iterative refinement. Establishing clear guidelines, refining AI instructions, and maintaining rigorous human oversight are essential to ensure the AI’s outputs are helpful, reliable, and trustworthy. By adopting this approach, organizations can effectively leverage AI to enhance their workflows, reduce manual workload, and achieve greater efficiency without compromising quality.</p><p>As we move forward, our focus will remain on optimizing AI integration, balancing cost and performance, and ensuring that our AI solutions are robust, accurate, and reliable.</p><h4>Post Script</h4><p><em>This blog post was written with assistance from ChatGPT-4o. The author planned and outlined the post and had the AI write a first draft. Subsequently, several staff members at GreenInfo reviewed and edited the article extensively for accuracy, tone, clarity, and flow.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f62ab0a9dc46" width="1" height="1" alt=""><hr><p><a href="https://medium.com/greeninfo-network/accelerating-legal-document-review-with-ai-f62ab0a9dc46">Accelerating Legal Document Review with AI</a> was originally published in <a href="https://medium.com/greeninfo-network">GreenInfo Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Building a Spatial Analysis Tool for Bird Conservation in Colombia]]></title>
            <link>https://medium.com/greeninfo-network/building-a-spatial-analysis-tool-for-bird-conservation-in-colombia-c0f8abc814d5?source=rss----7b016ec69299---4</link>
            <guid isPermaLink="false">https://medium.com/p/c0f8abc814d5</guid>
            <dc:creator><![CDATA[Stephen R. Smith]]></dc:creator>
            <pubDate>Thu, 15 Aug 2024 20:10:03 GMT</pubDate>
            <atom:updated>2024-08-15T20:11:08.248Z</atom:updated>
            <content:encoded><![CDATA[<p>At GreenInfo Network, we have been working with Esri products and tools since the beginning. We keep up with all the developments in ArcPro and ArcGIS Online as Esri updates and provides new options and tools, so we were excited for the opportunity to work in an Enterprise environment with some of the latest Esri releases on a recent project with the Audubon Society. We had the chance to dive deeper into Esri’s Enterprise stack, including ArcServer and Portal for ArcGIS.</p><p>The National Audubon Society needed a tool to support bird conservation efforts in Colombia. They knew that providing a tool for spatial data analysis could aid in various conservation-related decisions. With the help of an extensive set of vector and raster data from NASA, we created <strong>“</strong><a href="https://storymaps.arcgis.com/collections/7f42edb5698e4900bd37d3b616fe112f"><strong>Alas Seguras</strong></a><strong>”</strong> (“Safe Wings”), a tool to support bird advocates in identifying key areas for bird conservation in Colombia. It leverages Esri’s <a href="https://doc.arcgis.com/en/experience-builder/latest/configure-widgets/suitability-modeler-widget.htm">Suitability Modeler</a> widget as part of an Experience Builder in Audubon’s ArcGIS Online platform. Data layers are hosted in Audubon’s Enterprise Portal, and ArcServer Image Services is used to make the <a href="https://github.com/Esri/weighted-raster-overlay-service-toolbox">Weighted Raster Overlays</a> available.</p><h3>Uncovering Key Use Cases with User Research</h3><p>Our user research played a crucial role in uncovering the best uses for the tool by deeply understanding the needs and motivations of the users. We conducted a self-reflection exercise with Audubon’s oversight team to help them articulate their goals and the metrics they use to measure success. This exercise was instrumental in identifying the specific questions users needed to answer and the initiatives they were pursuing to achieve their desired outcomes. Through this research, we discovered that users are not merely looking to view data; rather, they seek to leverage the tool to drive impactful decisions by answering key questions that guide their actions. By focusing on these user-defined case uses, we ensured that the tool was tailored to support meaningful decision-making processes.</p><p>With input from the users, we developed a list of user scenarios and use cases, which helped determine that they needed a tool to help tool users identify areas of high ecological value. Based on the user groups and the available data, we developed the following modules for different use cases — each with practical applications that directly contribute to bird conservation efforts:</p><ul><li>The <strong>Agricultural Areas Module</strong> aims to implement regenerative agriculture practices like reforestation and soil reconstruction. It helps identify areas of high ecological value within agricultural ecosystems, particularly beneficial for wetland birds.</li><li>The <strong>Livestock Areas Module </strong>promotes sustainable and environmentally friendly livestock practices. It helps locate areas where ranchers can be encouraged to plant trees, enhance habitats for lowland forest birds, and connect national parks for avian movement.</li><li>The <strong>Bird-Friendly Communities Module</strong> engages communities and interest groups in strong conservation movements. It helps convince local governments about the ecological value of their areas for migratory forest birds and supports the implementation of Colombia’s ENCA goals.</li><li>The <strong>Protected Areas Module</strong> is vital for identifying and strengthening existing protected areas and prioritizing new ones. It can guide decisions on where tree planting can enhance bird habitats in montane forests and contribute to carbon sequestration. It also focuses on increasing landscape connectivity between protected areas.</li></ul><p>Each module addresses specific conservation needs and provides tools for mapping and analyzing areas critical for bird conservation. By collaborating with farmers, communities, and conservationists, “Alas Seguras” aims to create a sustainable environment for birds, particularly migratory species, while benefiting local ecosystems and communities. This tool underscores the importance of integrated, data-driven wildlife conservation and habitat management approaches.</p><h3>Esri’s Suitability Modeler Widget</h3><p>We needed a flexible, detailed, and user-friendly tool to provide this data-driven support for Colombian decision-makers and bird advocates to achieve these goals. To this end, we turned to Esri’s Experience Builder to create a platform for configuring the final application.</p><p>The Suitability Modeler widget enhances decision-making processes for conservation activities with accurate data. This powerful analytical tool allows users to evaluate multiple factors simultaneously by combining and weighing different data layers. For example, bird advocates could combine different data layers to answer:</p><blockquote>Where can surface water sources work as natural corridors along an agricultural landscape through riparian vegetation?</blockquote><p>To answer this question, a user would select and rank the following layers:</p><ul><li>Surface water</li><li>Land Use: Agriculture</li><li>Bird Friendliness: Wetland</li></ul><p>Users could even prioritize sub-layers within the weighting, such as the types of water (permanent, seasonal, ephemeral), to further refine the analysis.</p><figure><img alt="Screenshot of the configuration panel of the widget showing the selected layers and their model prioritization scores" src="https://cdn-images-1.medium.com/max/1024/1*acdFFQ70SumcNMa3VBQqwg.png" /><figcaption>The Suitability Modeler provides fine-grained tools for tuning an analysis to your specific needs.</figcaption></figure><p>Once users receive their analysis results, they can view the data on the map, summarize data by department or city, and see a pie chart showing the distribution of conservation values.</p><figure><img alt="A map and pie chart displaying the results of the analysis. The map has colored squares with different hues representing high or low analysis values. The pie chart shows the distribution of those values in a selected department" src="https://cdn-images-1.medium.com/max/1024/1*2wyi3gFdJB-GgQdKc15sZQ.png" /><figcaption>Analysis results show a spatial distribution and the proportional values in a selected area.</figcaption></figure><p>By integrating suitability modeling into our web application, we’ve enabled users to interact with and experiment with various scenarios, making it easier to engage with and contribute to the conservation planning process.</p><p>We needed a large set of raster and vector datasets to power the Suitability Modeler. The rest of this post describes our experience preparing and publishing this data for use in the final tool.</p><h3>Raster Data Preparation</h3><p>The data creation process was very involved and presented several challenges. The source datasets were in wildly different formats but needed to be combined into a single <a href="https://github.com/Esri/weighted-raster-overlay-service-toolbox">Weighted Raster Overlay</a>, a specialized mosaic raster dataset. To transform the 22 data layers, we generated a large set of ArcPy scripts with the help of OpenAI’s ChatGPT. These allowed us to iterate and refine a repeatable process to integrate future updates more easily.</p><p>Each dataset began as either a vector or raster dataset. Raster layers eventually combined into a Weighted Raster Overlay, and the vector layers acted as selection areas for summarizing the raster layers in the Suitability Modeler widget.</p><p>ChatGPT was pivotal in helping us develop, troubleshoot, and refine the Python scripts. It provided a conversational interface to discuss and refine our code, making the development process more efficient and collaborative.</p><p>Within ArcPro, we used Python Jupyter notebooks to create a modular and reusable system. We evaluated each dataset and developed steps to align it to the final mosaic dataset format — for example, some required reclassification, some required reprojection, and some required clipping.</p><h3>Vector Data Preparation</h3><p>The Audubon Society provided a variety of landscape and administrative areas for the tool to be used as summarization areas in the widget. This powerful feature allows users to select preset boundaries and receive dynamic reports of the raster data within the area. This was the main selling point of using Weighted Raster Overlays and the Suitability Modeler.</p><p>Using Jupyter Notebooks, produced with the help of ChatGPT, we made quick work of the processing steps. Writing code as a notebook allows for short, modular code that’s reusable and easily readable. It also allowed for quick iteration if we located data issues in later processing steps. For example, if an administrative area encompassed a large area of the ocean, we could return to the processing cell and add a clip step before the export step.</p><p>Recent advancements in large language models have produced some spectacular tools that allow developers to write code quickly with AI support. For these automation scripts, I defined the input files and determined the general processing steps and the outcome. With those instructions, I asked ChatGPT to write me an ArcPy script to achieve the results. Once an answer was provided, I carefully reviewed the code to ensure it was free of bugs or typos. Then, I tested the result on my dataset to confirm it achieved the desired output. It was not perfect, but the iterations greatly sped up the scripting process.</p><h3>Weighted Raster Overlay Creation</h3><p>After reprocessing the raster datasets, we generated unique Weighted Raster Overlays for each module. This was a complex and time-consuming task, but in the end, we were able to automate all the necessary steps with ArcPy.</p><p>We created a complex set of scripts to create and configure the Weighted Raster Overlays, but the geoprocessing tools provided by Esri were not easy to use in an automated workflow.</p><p>We encountered several challenges, which were eventually overcome with automation. First, the WRO setup tool takes all the layers in a map document as input, so a script had to be written to populate each module’s map with the required layers. This script was instrumental, as we reprocessed these WROs many times during development.</p><p>Second, each layer in the WRO relied on the symbology of the layer in the map, so we had to create a layer file for each layer. However, despite this preparation, the WRO creation tool did not honor the category labels, leading to having to make complex configurations to automate the configurations of layer classifications, weights, and labels as a secondary step.</p><figure><img alt="A screenshot of Python code in a Jupyter notebook. The values shown are used to configure the weighted raster overlay." src="https://cdn-images-1.medium.com/max/1024/1*G9OSVS1CqKzKzoX69Xxcfw.png" /><figcaption>To make the process repeatable, we create complex configurations to speed up processing.</figcaption></figure><p>Along the way, we also encountered a few unexpected hurdles. There was a particular issue where the tool <a href="https://github.com/Esri/weighted-raster-overlay-service-toolbox/issues/37">failed to recognize updates</a> and another where <a href="https://github.com/Esri/weighted-raster-overlay-service-toolbox/issues/39">certain character inputs</a> led to errors. Fortunately, Esri developers swiftly addressed these issues.</p><h3>Publishing</h3><p>After completing the lengthy data processing pipeline and generating our four WRO datasets, we moved on to publishing them and preparing the web application. This phase was equally challenging, filled with complexities and some puzzling issues.</p><p>Esri ArcServer — which was supposed to publish our services — started behaving unpredictably. Sometimes publishing worked, but it often failed without providing clear error messages. Despite multiple attempts and trying various solutions, success was sporadic at best. During troubleshooting, we attempted to switch the publishing pipeline to upload data to Audubon’s SDE database. However, this proved impractical due to significant upload times over VPN.</p><p>One particularly frustrating problem involved an issue with <a href="https://learn.microsoft.com/en-us/windows/win32/fileio/naming-a-file#:~:text=Do%20not%20use%20the%20following%20reserved%20names%20for%20the%20name%20of%20a%20file%3A">reserved Windows file names</a>, where using specific names caused unexpected failures. Eventually, however, we could finally publish the services, and the final WRO services were resampled to 300m to balance fidelity and performance.</p><h3>Conclusion</h3><p>This journey through the Esri Enterprise landscape, from ArcPro to Server to Portal to ArcGIS Online, was both challenging and rewarding. Developing and iterating on the processing scripts using ChatGPT streamlined that portion of the process. This experience underscores the potential of AI as an additional tool in GIS and spatial analysis workflows, sparking intrigue and inspiration.</p><p>Our use of ArcPy in the raster data processing pipeline significantly improved the final product, enhancing its robustness and efficiency. By harnessing ArcPy’s powerful capabilities, we simplified complex tasks, ensuring higher accuracy and reliability in our data processing. This ultimately led to superior outcomes for our project.</p><p>Leveraging Esri’s suite of tools, including Enterprise, ArcGIS Server, and Experience Builder, proved to be a powerful platform for Alas Seguras. Their off-the-shelf and configurable tools allowed us to create a custom experience with minimal configuration. The robust capabilities of Esri’s ecosystem played a crucial role in delivering a final product that meets the needs of Colombia’s bird conservation efforts.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c0f8abc814d5" width="1" height="1" alt=""><hr><p><a href="https://medium.com/greeninfo-network/building-a-spatial-analysis-tool-for-bird-conservation-in-colombia-c0f8abc814d5">Building a Spatial Analysis Tool for Bird Conservation in Colombia</a> was originally published in <a href="https://medium.com/greeninfo-network">GreenInfo Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Sharing our Accessibility in Mapping Workshop]]></title>
            <link>https://medium.com/greeninfo-network/sharing-our-accessibility-in-mapping-workshop-ee7e5e35705c?source=rss----7b016ec69299---4</link>
            <guid isPermaLink="false">https://medium.com/p/ee7e5e35705c</guid>
            <category><![CDATA[maps]]></category>
            <category><![CDATA[accessibility]]></category>
            <category><![CDATA[gis]]></category>
            <category><![CDATA[accessibility-design]]></category>
            <category><![CDATA[cartography]]></category>
            <dc:creator><![CDATA[Vanessa Knoppke-Wetzel]]></dc:creator>
            <pubDate>Fri, 01 Dec 2023 16:47:17 GMT</pubDate>
            <atom:updated>2023-12-01T20:30:39.124Z</atom:updated>
            <content:encoded><![CDATA[<h3>Why give a workshop?</h3><p>For the past few years at GreenInfo Network, we have been working on making changes in our project infrastructure workflows and continue learning and growing to support accessibility needs and requirements. At the least, we want to make sure our products meet regulatory requirements for accessibility by people regardless of ability, and ideally, we would go <em>beyond</em> regulation compliance.</p><p>In 2023, we began working with <a href="https://fossheim.io/">Sarah Fossheim</a> to focus on improved accessibility in our data-rich interactive products, and we have learned a lot from working with them. There also are robust web accessibility standards that are clearly documented, but we continued to struggle finding resources and information that were more map and cartography specific. Interactive maps are so complex: they have a lot of data layers stacked on top of each other to show information, in addition to toggling capacity, possible animations, running live analysis, and more. The web standards for websites and data visualizations do not completely answer or provide solutions for best practices for web maps. We can identify what existing technologies do not have the capacity to support accessibility, but do not always know what other technologies exist that may be better. What technological solutions and alternatives are out there?</p><p>We wondered what other mapping people and organizations were doing and attended webinars and workshops and ran into people asking the same questions we were: where are the resources, what can we do to improve, and how do we consistently create accessible web maps?</p><p>We are pretty committed to learning and growth and kept trying to find “solutions”, and knew that doing some research also was required. We reached out to <a href="https://geography.wisc.edu/gis/staff/roth-robert/">Rob Roth</a> hoping Rob would want to collaborate on accessibility in mapping research. Rob’s academic work centers around interactive maps, UI/UX, and he recently helped publish the book “<a href="https://www.un-ilibrary.org/content/books/9789216040468">Mapping for a Sustainable World</a>” for the United Nations, which included a section about accessibility in mapping. Not surprisingly, Rob was as excited as us to work together to research what exists, as well as find solutions for accessibility in mapping.</p><p><strong>As an initial step towards doing research, we wanted to share information that we collectively have learned in our respective practical and academic work, as well as build community around accessibility in mapping so that we can all have these conversations and learn, share, and grow — together.</strong> Given that NACIS is where so many cartographers go to learn and share, we knew that was an ideal place to host a workshop: so, the NACIS Workshop idea was born!</p><h3>NACIS “accessibility in mapping” workshop in review</h3><p>We organized the Workshop in four parts: 1) Why accessible design matters 2) Designing <em>with/for</em> people with disabilities 3) Aesthetics and accessibility and 4) Practical examples/ accessible mapping in action. We ended the workshop with small groups breaking out to test some accessibility tools. This review will summarize parts 1–4. Feel free to take a look at the Workshop <a href="https://docs.google.com/presentation/d/1MgeKLFw-0lkMV2nBDKrSgCp8PFth17W8bq4gg7Fx7T8/edit#slide=id.p">slides</a> to dive more deeply into some of the information provided below.</p><h4>Why accessible design matters</h4><p><em>“Designers should look first to those at the </em><strong><em>margins</em></strong><em>: the people pushed to the margins in any particular design context demonstrate who and what the system is trying to exclude.” — D’Ignazio &amp; Klein in </em><a href="https://data-feminism.mitpress.mit.edu/">Data Feminism</a><em>, referencing Shaowen Bardzell</em></p><p>Rob Roth began the workshop highlighting that historically, products have been designed for the “average user”, with the goal to get products in as many hands as possible (quantity), but with disregard for the margins.</p><figure><img alt="Graphic shows a bell curve that highlights the largest section of the bell curve, labeling this section as “Generalizability to the Average Map User”. The left and right tails of the bell curve — which take up less area visually — are labeled as “Margins”" src="https://cdn-images-1.medium.com/max/1024/1*38pyIp2y3OJpI6IrFT7XJg.png" /></figure><p>As the Data Feminism quote demonstrates, building and creating products (mapping softwares to maps themselves) that are for average users means that many people are left out and ultimately, excluded.</p><p>In creating our mapping softwares and resulting products, it is important to begin with <strong>inclusive design</strong>.</p><figure><img alt="Cartoon people representing examples of people in the margins, such as Permanent (person with one arm), Temporary (arm injury), and Situational (new parent carrying a baby)" src="https://cdn-images-1.medium.com/max/1024/0*o8c-9rTeSZSYnX4W" /></figure><p>Inclusive design takes the margins into consideration and recognizes exclusion. Often when solving for one type of exclusion, the solution extends to providing a solution for multiple exclusion types. For example, solving for someone with one arm, also helps provide solutions for someone with an arm injury, or a new parent who often only has use of one arm at a time.</p><p>What does<strong> accessible map design</strong> look like?</p><ul><li>The ability for anyone to obtain, read, interact with, and benefit from a map</li><li>Qualities that make a map usable specifically for people with disabilities</li></ul><p>Rob shared some initial inclusive and accessible questions to ask ourselves as we create maps:</p><p>Inclusivity:</p><ul><li>Who is and is <em>not </em>served by this map? Who might benefit and who might be harmed?</li><li>Who participated in the map design and evaluation?</li><li>What abilities (visual &amp; non-visual) are needed to use the map?</li><li>What “hidden” assumptions or “unwritten” rules to use this map?</li><li>What value is the audience deriving from the map? What are the costs?</li></ul><p>Accessibility:</p><ul><li>Does the visual hierarchy follow contrast standards (3.5:1 standard)?</li><li>Does the visual hierarchy hold up in grayscale?</li><li>Does the visual hierarchy hold up in different contrast &amp; brightness settings, or environmental viewing conditions?</li><li>Are the map and map elements available in <em>vivid </em>alternative text?</li><li>Are labels legible at twice the recommended reading distance (~3x the diagonal distance of the map)</li><li>Is text rendered in dynamic HTML to allow for resizing, language translation, etc.?</li><li>Can the data be downloaded as a table (text) rather than a map?</li></ul><p>There are a lot more questions to ask and consider for accessible design, as accessibility needs are broad. We encourage you to begin with asking these questions and to alsoask more, and expand your question toolkit. It is important to begin a project with these questions, and not ask them later — when it is too late to make changes. When you begin designing with inclusivity in mind, it means that the design workflows will be improved!</p><h4>Designing with/for people with disabilities</h4><p>Travis White reiterated that <strong>accessibility is important, inclusion is essential, and in many places, accessibility is a right protected by law. Accessibility is a human right. </strong>Inclusive design is not just designing for low-vision users, but designing <strong>with/for people with </strong>disabilities in general (and NOT designing for disabilities). He shared some key takeaways from the <a href="https://www.urban.org/research/publication/do-no-harm-guide-centering-accessibility-data-visualization">Centering Accessibility in Design, Do No Harm Guide</a>:</p><ol><li>Design with accessibility in mind <em>from the beginning</em>.</li><li>Accessibility should not be a specialty.</li><li>There is no established definition for what makes a data visualization accessible.</li><li>People with disabilities should be involved in the design of data visualization products and usability testing.</li><li>There is not a single right answer for writing <a href="https://support.microsoft.com/en-us/office/everything-you-need-to-know-to-write-effective-alt-text-df98f884-ca3d-456c-807b-1a1fa82f5dc2">alt text</a>.</li></ol><figure><img alt="Screenshot excerpts from Chapter 7 of the Do No Harm Guide, visually demonstrating the amount of checklists that exist in the Do No Harm Guide that people can work through in order to assess the accessibility of their data visualizations." src="https://cdn-images-1.medium.com/max/1024/0*TJAGyT9PkPdGNbTX" /></figure><p>The Do No Harm Guide also provides a checklist to go over as you work through projects in Chapter 7.</p><p>Travis also reinforced discussions from Frank Elavsky (whose talk about Accessibility and Data Visualization can be seen <a href="https://www.youtube.com/watch?v=wqvDFYMIFYI&amp;feature=youtu.be">here</a>) and also shared Frank’s <a href="https://chartability.fizz.studio/#what-is-chartability">Chartability</a> resource for <a href="https://chartability.github.io/POUR-CAF/">reviewing accessible design work</a>. “Chartability is a set of testable questions for ensuring that data visualizations, systems, and interfaces are accessible. Chartability is organized into principles with testable criteria and focused on creating an outcome that is an inclusive data experience for people with disabilities.”</p><h4>Aesthetics and accessibility</h4><p>I (Vanessa Knoppke-Wetzel) dove into what is often a difficult topic for cartographers when creating accessible maps: where do aesthetics fit in when there are many high contrast requirements for low-vision considerations, and where do aesthetics fit in given the broader accessibility design needs?</p><p>Maps are very visual, and cartographers care deeply about aesthetics. We work hard to make maps that not only share information, highlight data analyses, <em>and</em> showcase where things are, <em>while also</em> displaying it all in a beautiful way. While there are no current all-inclusive “accessibility design in cartography” guides (that I am aware of!) such as what we learn for cartographic theory and design (e.g. Bertin’s <a href="https://www.esri.com/en-us/esri-press/browse/semiology-of-graphics-diagrams-networks-maps">Semiology of Graphics</a>; Slocum et al’s <a href="https://www.taylorfrancis.com/books/mono/10.1201/9781003150527/thematic-cartography-geovisualization-fourth-edition-terry-slocum-robert-mcmaster-fritz-kessler-hugh-howard">Thematic Cartography and Geovisualization</a>), we can together create new standards. How?</p><p>First, I ask cartographers reading this to reframe their thinking of what is “ugly” or “pretty”; what is “aesthetically pleasing” vs “what isn’t”; and to <strong>consider the context of aesthetics in the history of cartography</strong>.</p><figure><img alt="An example of a copperplate map from ~1700s. The map has a lot of visual texture — detailed marginalia, water and coastline patterns. The map and marginalia contain detailed illustrations, such as faces and clouds." src="https://cdn-images-1.medium.com/max/1024/0*72dZ881Jct82ngMZ" /></figure><figure><img alt="Screenshot examples from a variety of maps tha highlight the many visual differences that exist in maps in the history of cartography. The screenshot show a variety of line, fill, and pattern textures, symbology differences, and level of detail differences. Some screen shots include map monsters, which we do not see in maps today." src="https://cdn-images-1.medium.com/max/1024/0*D2wk9J1OPMYNKGTR" /></figure><figure><img alt="Two examples of technology/tools used to produce maps in the history of cartography: A large and complex looking lithograph press, and a small woodblock that has been carved into. The woodblock carving depics a map, and the carving tools are shown to indicate how the woodblock map was carved." src="https://cdn-images-1.medium.com/max/1024/0*Xnkf7GYxDe8U0ILu" /></figure><p><strong>Maps today do not look the same as maps made 100, 200, 300, 500, or 1000+ years ago</strong>. Maps made in the 1200s do not look the same as maps made in the 1700s, and those do not look the same as maps made in the 1900s. Today’s map aesthetics are informed by stylistic tastes, but they also are informed by technological influence.</p><p>In my opinion, mapping technologies today easily support more minimalistic/clean/streamline designs and tend to largely look the same — aesthetically — in regards to there being less visual texture compared to historical maps. For example, you expect vector lines to be perfectly smooth. Additional intricacies such as drop shadows, beautiful terrain, patterns, and more take additional time to create, but still have a cleaner, more “perfect” look compared to maps in the history of cartography. This is not to say that all maps today look <em>exactly</em> <em>the same</em>, but modern technology production methods directly affect map products and in some way support modern maps having a particular overall aesthetic as compared to the history of cartography.</p><p>To repeat: <strong>modern maps fall into their own style/aesthetic that are quite different from the 1200s, 1700s, and 1900s</strong>. Why does this matter?</p><figure><img alt="Two maps are juxtaposed to highlight the difference between historical/old maps alongside maps from today. The historical map and the current map both contain a lot of visual detail, but they present very differently because of the tools that were used to create these two different maps." src="https://cdn-images-1.medium.com/max/1024/0*ZoWROnfwigHtPhyq" /></figure><p>If we remind ourselves that cartographic design, production techniques, aesthetics, and style have changed constantly throughout history, then we also know that <strong><em>we</em></strong> have the power to participate in changing what design requirements/rules exist and redefine what our cartographic design standards are. <strong>Today, cartographers are closer to all the production technologies than ever before: cartographers exist in product design, and mapmakers using software can provide important user feedback.</strong></p><figure><img alt="A screenshot of a Google Search for “Adobe Illustrator and Cartography”. The thumbnail examples from the Google Search results show a variety of maps with different visual aesthetics." src="https://cdn-images-1.medium.com/max/906/0*umX9MoVMkbLQjOS8" /></figure><p>Additionally, we also use softwares and tools that are <em>outside</em> of the GIS world — such as the Adobe Creative Suite or Blender — to make our maps better looking. At the moment, we are limited by tools that do not support accessible design as well as they could, but there is nothing stopping us from identifying more tools or changing to existing tools that would help us create better maps and improve the tools used to support accessibility in mapping.</p><p>In cartography, there is not (yet! that I know of) a design standard on what beautiful, well-designed accessible web maps can/should look like, or a map-specific checklist such as the checklists shared by Travis. But there<em> could</em> be — and we can collectively work towards defining that. It is in our power in our daily jobs to determine what is <em>missing</em> in the static and interactive tool stacks that would improve design, and share what can be improved, as well as begin to incrementally implement what is already possible. <strong>Cartographers today have the power to be the change we want to see in the industry.</strong></p><p>We have the power to research, experiment, and define what well-designed accessible maps can look like. In the same way that <a href="https://colorbrewer2.org/#type=sequential&amp;scheme=BuGn&amp;n=3">Color Brewer</a> exists and provides some well-loved color schemes for data visualizations, or MapPublisher exists so that we can can create geospatially accurate maps in Adobe Illustrator (and Esri created ArcGIS Maps for Adobe Creative Cloud), we can find and define an improved tool stack, share information on what questions and research need to happen to improve tools and design, and push for existing softwares and tools to change, as well and updating infrastructure to ultimately change our workflows so that cartographic design is consistently accessible and inclusive from the start.</p><p>I leave you with some questions to consider:</p><ul><li>Who do we need to talk to?</li><li>What is missing from our toolkits?</li><li>What knowledge exists in other Design spaces that we can incorporate into our workflows to help support accessibility design?</li><li>What changes do we need in softwares?</li><li>What else requires research?</li><li>What questions aren’t we asking?</li></ul><h4>Practical examples</h4><p>Tim Sinnott dove into describing challenges that he has experienced while improving accessible map design in his daily work, and he also discussed opportunities and options for improving.</p><p><strong>Maps are complex</strong></p><p>Tim began by noting that because cartography is primarily a complex visual art form, improving accessibility in mapping can feel challenging.</p><figure><img alt="Interactive map example highlighting the complexities of web maps: detailed data, variety of color display, alongside additional data analysis information that is provided in a separate box overlay." src="https://cdn-images-1.medium.com/max/720/0*sdpU99_g4e_jaVd8" /></figure><p>Maps often have subtle color contrasts because of the number of data layers on the map. They tend to have many different features symbolized by points, lines, polygons, and patterns. Additionally, maps often have very small icons and text, with text that curves and aligns with map features in all directions, as well as providing a particular aesthetic to the map. Additionally, maps use color to convey meaning … and so much more!</p><p>Interactive maps bring in even more challenges: for example, when data layers that can be turned on and off, you may be adding more data to the map, and you are also adding more complexities that require alternatives for accessible design, because toggle-able data layers and resulting map display changes are not screen readable.</p><p>When maps have such a long list of elements that need to be considered, it feels daunting to know where to begin improvements. Given this, Tim shared some good places to start:</p><ul><li>Use larger fonts, and all facing one direction (if possible)</li><li>Use text as well as color to convey meaning (and make sure color is appropriately contrasted)</li><li>Increase color contrast</li><li>Use fewer map features</li><li>Integrate patterns to complement solid colors</li><li>Use color-blind friendly palettes</li><li>Use tactile and auditory maps</li></ul><p>For interactive maps:</p><ul><li>Use standard controls and <a href="https://www.w3.org/WAI/standards-guidelines/wcag/">web accessibility standards</a> to make accessible maps more accessible.</li><li>With radio buttons, you can show many different layers (one at a time)</li><li>Paginated overlay data can help with screen readers and general accessibility</li><li>All maps on the internet can be made more accessible with titles, a long description, a table / cards / text view, and data download functionality.</li><li>We should make websites and web maps (or map equivalents), navigable by keyboard.</li></ul><p>Tim curated <a href="https://github.com/GreenInfo-Network/accessibility-in-mapping/issues/1">a list of resources</a> for folx to explore, use, learn, and grow, as well as tools to evaluate interactive products.</p><p><strong>Accessible design in action</strong></p><p>Tim dove into an example of a project where accessible design alternatives were created so that there was an equivalent data review and exploration experience.</p><figure><img alt="An interactive map example that allows a user to pan and zoom across the map to select specific areas. The map allows data layers to be toggled on and off visually. An area has been selected in this screenshot, and corresponding metrics are shown above the map." src="https://cdn-images-1.medium.com/max/1024/0*CNsbNb0j5nxhC3Zm" /></figure><p>Exploring the above interactgive map allows a user to gather information about the area they navigate to. However, interactive maps are not very explorable with screen readers and keyboard navigation (although some places are working on changing that and improving the experience, such as <a href="https://support.google.com/maps/answer/6396990?hl=en&amp;co=GENIE.Platform%3DDesktop">Google Maps</a>). Interactive maps are especially difficult for screen readers because of the numerous complexities that come with interactive features, such as turning data layers on and off orselecting a specific area toreceive the corresponding data analysis output.</p><p>Providing an equivalent experience (exploring area, getting results) is possible by providing the data in downloadable tabular form. A similar exploring/getting results experience for keyboard navigation/screen reading is building a text filtering option that still allows a user to explore the data using keyboard navigable filters to find the data.</p><figure><img alt="Showing an alternative data exploration website page that corresponds with the explorable map. Keyboard navigation can select geographies from filter menus, and then tab to the links related to selected geographies after filter selection." src="https://cdn-images-1.medium.com/max/1024/0*TlGEuXqZq1gOomDp" /></figure><p>Keyboard navigation can tab through results and a screen reader can read the data.</p><figure><img alt="After clicking on selected data link, user can tab through data to learn about their selected geographics and corresponding data metrics" src="https://cdn-images-1.medium.com/max/1024/0*HrN9jP69vRdyfLl_" /></figure><p>The above is just one practical example of accessible design in action: check out a few others in the <a href="https://docs.google.com/presentation/d/1MgeKLFw-0lkMV2nBDKrSgCp8PFth17W8bq4gg7Fx7T8/edit#slide=id.g28cc2211a80_2_11">Workshop slides</a>!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ee7e5e35705c" width="1" height="1" alt=""><hr><p><a href="https://medium.com/greeninfo-network/sharing-our-accessibility-in-mapping-workshop-ee7e5e35705c">Sharing our Accessibility in Mapping Workshop</a> was originally published in <a href="https://medium.com/greeninfo-network">GreenInfo Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Diverse Dev Careers Lead to GIN]]></title>
            <link>https://medium.com/greeninfo-network/diverse-dev-careers-lead-to-gin-d31a90a8c887?source=rss----7b016ec69299---4</link>
            <guid isPermaLink="false">https://medium.com/p/d31a90a8c887</guid>
            <category><![CDATA[design]]></category>
            <category><![CDATA[web-design]]></category>
            <category><![CDATA[developer]]></category>
            <category><![CDATA[mapping]]></category>
            <category><![CDATA[gis]]></category>
            <dc:creator><![CDATA[GreenInfo Network]]></dc:creator>
            <pubDate>Fri, 23 Jun 2023 17:14:24 GMT</pubDate>
            <atom:updated>2023-06-23T15:35:32.845Z</atom:updated>
            <content:encoded><![CDATA[<p><em>By Kimberly Becerril</em></p><p>The rise of GIS and computer science in the 1960s invoked new frontiers in the realm of data analysis and visualization. However, during their first few decades, the complexity and computational power these technologies required limited their convenience to academic institutions. About a decade after the commercialization of GIS, in 1996 GreenInfo Network (GIN) aspired to push the edge of GIS by making its utility accessible to nonprofits. As a geospatial analysis consultant for nonprofits, the organization balanced the role of technical acuity along with a keen discernment for compelling visualizations. These attributes — in tandem with GreenInfo’s commitment to building long-term client partnerships and enduring socially impactful projects — are possible through the staff’s use of web and software development in their projects.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*M3M2Wq2FOhM0I6e6eo4tTw.jpeg" /></figure><p><strong>As a former Geospatial Analyst at GIN</strong>, I applied Python programming in projects as a means to create more efficient and replicable data processing. My coding journey started with seeking small and accessible assignments to improve GIN’s workflow and my own workflow in client projects. I also used tools like ModelBuilder in ESRI’s ArcGIS Pro, to create visual diagrams of geoprocessing tools with iterative capabilities. Collaborating and connecting with colleagues notably shaped my perception of the different approaches and applications of web and software development. Currently, as a Master of City Planning candidate at the Massachusetts Institute of Technology, I am deepening my appreciation, application, and knowledge of software development.</p><p>To glimpse into the diverse career paths that are possible to become a web and/or software developer in the geospatial sector, this blog explores the experiences and trajectories of three GIN staff. In addition, they share career insights and considerations for those interested in entering this sector.</p><h3>Interaction Designer and Senior Cartographer</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ct53Ktu7S4N7d8PUErbUnQ.jpeg" /></figure><p><a href="https://www.greeninfo.org/about/staff/tim-sinnott"><strong>Tim Sinnott</strong></a> joined GreenInfo Network in 2006. Tim’s educational background is in Geography and Environmental Science in addition to Studio Art with a focus on sculpture. In many ways, these foundational areas of study are reflected in the work designed by Tim today. As an Interaction Designer and Cartographer, he found a niche in the geospatial sector wherein his work serves as the bookends of a project. Tim utilizes Figma and Adobe XD for wireframing and prototyping in addition to CSS and HTML to clean up visualizations in the backend. During a project’s early stages, Tim’s creative wireframe designs simplify the complex and walk users through a site’s capabilities. He then returns to a project during its final stages to tailor the backend details essential to GreenInfo Network’s award-winning visualizations.</p><p>Tim’s eye for standards, balance, and functionality steered him to his new focus, website accessibility. He contemplates, “How can you make websites universally accessible and consumable?” Website accessibility opens resources to more people and improves the site experience for all.</p><p>For prospective people in this sector, Tim suggests joining geospatial and data development communities whether locally or virtually. Some communities include <a href="https://twitter.com/MaptimeOAK">Maptime</a>, based in Oakland, CA, and <a href="https://slack.hotosm.org/">Open Street Map</a> Slack. There are a variety of groups globally offering a sense of projects, workplaces, and opportunities in this sector.</p><h3>Senior Web GIS Developer</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*87Gnwp9mZyn3vx_j9XH_Hw.jpeg" /></figure><p><a href="https://www.greeninfo.org/about/staff/gregor-allensworth"><strong>Gregor Allensworth</strong></a> started at GreenInfo Network in 2011. Gregor’s undergraduate education is in Psychobiology and Neuroscience but shifted toward Computer Science concurrently. Developments in computer operating systems, processors, and the invention of the HTTP protocol made the mid-1990s an optimal time to work with computers. These technological developments made computers functionally accessible in ways they previously fell short of. In some manner, Gregor’s experience with web development grew in unison with the rise of computers which elucidates the deep and thorough knowledge of web development he offers to projects. In the early 2000s, his work shifted from application and file-sharing technologies to mapping, prompted by his employer. Gregor welcomed this change because of his childhood association with maps in science fiction. Through this, he helped develop one of the first map builder applications. Gregor’s comprehensive understanding of web development in addition to his over decade-long place in the GreenInfo community anchors his projects with ingenuity and resiliency. These projects tether the forefront of interactive geospatial technology on top of a foundational basis ensuring their longevity.</p><p>Gregor’s project websites’ relatively long lifespans are partly a result of adequate cyber security and long-standing scripts. Hence, for new web and software developers, he emphasizes the importance of web security and sustainability though these topics are often missing in coding boot camps. To build a resilient website, they are essential to understand and apply.</p><h3>Geospatial Analyst and Developer</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8zAupLc5sPeebq9ptK1vFQ.jpeg" /></figure><p><a href="https://www.greeninfo.org/about/staff/jayita-bhojwani"><strong>Jayita Bhojwani</strong></a> joined GreenInfo Network in 2019. Jayita’s university studies focused on Electrical Engineering and Computer Architecture, after which work experiences sparked her interest in software testing and automation. Jayita continued her education through community college and online classes to develop a stronger software development foundation and grow her technological skills. After fourteen years in the software QA field, Jayita sought to change careers, to focus on using technology for social good. GIS opened this door, and Jayita built her knowledge and experience through certificate courses, attending conferences, and volunteering to do GIS for organizations with environmentally focused values.</p><p>Today Jayita builds some of GreenInfo Network’s most complex data pipelines using Python with ArcPy and PostgreSQL. Her background in software systems, coupled with her methodical strategies and documentation, allow Jayita’s data pipelines to scale efficiently and offer swift interpretability to those working with them, despite the level of complexity her projects pose.</p><p>For those interested in software development and GIS, Jayita recommends pursuing developmental opportunities through courses, learning from the code and workflow of other developers, and seeking constructive feedback on scripts. These strategies are imperative to building good programming practices, and rapidly improving one’s problem-solving approaches. Ultimately, cultivating a community is invaluable for acquiring learning opportunities whether through independent or collaborative projects.</p><h3>All Pathways Lead to GreenInfo Network</h3><p>For people interested in pursuing a field at the intersection of GIS, User Interface, and Software and Web Development, the experiences of Tim, Gregor, and Jayita capture a small sample size of the breadth of educational backgrounds and experiences an individual may have to become a web and software developer in the geospatial sector. Despite their different pathways, they share a dedication to organizations and projects working in the public interest. It is organizations like GreenInfo Network and individuals like Tim, Gregor, and Jayita that make a meaningful impact in society.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9wbnq88VC2-RtRRBba0zcQ.jpeg" /><figcaption>GreenInfo Network’s 2023 staff retreat</figcaption></figure><p>GreenInfo Network was founded to ameliorate nonprofits’ access to geospatial technologies and it succeeded. The community of partners and employees built around and within this organization for the past 27 years echo this organization’s founding principles and mission statement of<em> </em>creating, analyzing, visualizing, and communicating information in the public interest. When starting my next year of graduate school, continuing to build a foundation in software development will be informed by the experiences of my former colleagues and the community I started building in my graduate program. Though my career path is shifting towards City Planning, the relevancy of GIS and web and software development has only increased.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d31a90a8c887" width="1" height="1" alt=""><hr><p><a href="https://medium.com/greeninfo-network/diverse-dev-careers-lead-to-gin-d31a90a8c887">Diverse Dev Careers Lead to GIN</a> was originally published in <a href="https://medium.com/greeninfo-network">GreenInfo Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Creating the Zoning Atlas Editor]]></title>
            <link>https://medium.com/greeninfo-network/creating-the-zoning-atlas-editor-14f641329c10?source=rss----7b016ec69299---4</link>
            <guid isPermaLink="false">https://medium.com/p/14f641329c10</guid>
            <category><![CDATA[maps]]></category>
            <category><![CDATA[data]]></category>
            <category><![CDATA[regulation]]></category>
            <category><![CDATA[zoning]]></category>
            <dc:creator><![CDATA[GreenInfo Network]]></dc:creator>
            <pubDate>Thu, 15 Jun 2023 19:02:30 GMT</pubDate>
            <atom:updated>2023-06-15T19:02:30.773Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="A mixed urban scene from a high level looking off into the distance. In the near ground is a tight suburban neighborhood. Further off are medium rise buildings. In the far distance, dense high rises fill the horizon. This shows the variety of land uses and how they are impacted by zoning." src="https://cdn-images-1.medium.com/max/1024/1*KQ70txgOSiCzaqlDBu1_-w.jpeg" /><figcaption>Photo by<a href="https://unsplash.com/@chuttersnap?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> CHUTTERSNAP</a> on<a href="https://unsplash.com/photos/0bRBlE9699Y?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> Unsplash</a></figcaption></figure><h4><strong>From Spreadsheets to a Full-Stack Web Application</strong></h4><p><em>By Stephen Smith, Project Manager, GreenInfo Network</em></p><p><strong>Introduction</strong></p><p>Have you ever pondered the true significance of zoning maps? These intricate documents hold the key to understanding the laws that govern much of the United States’ land use, influencing our economy, environment, and society. The National Zoning Atlas (NZA) — a groundbreaking project by the Legal Constructs Lab at Cornell University’s College of Architecture, Art, and Planning — aims to unravel the complexities of zoning regulations by creating a transparent and accessible central database and public, online map. Through the Atlas, users can compare jurisdictions, identify trends, and make informed decisions in planning and policy-making at various levels. GreenInfo Network partnered with the NZA team, led by Cornell Professor Sara Bronin, to develop a modern web application that improves the collection and harmonization of information about the text of zoning codes and the districts created by zoning maps. This series will examine the project background and our high-level technical</p><p><strong>Project Goals</strong></p><p>Our primary goal was to replace the decentralized and error-prone zoning collection process, which relied on Google Sheets and file-sharing services like Dropbox, with a more efficient and user-friendly system. The previous data collection system required data to be saved independently by each analyst or team, dispersing information across many different systems and servers. The previous process was also prone to errors. For example, prior NZA analysts sometimes input information into the wrong row or column in the Google Sheet, or included entries using the wrong units (e.g., square feet, instead of acres). This new system creates a central database, with internal data-verification controls, that makes it easier for contributors to the National Zoning Atlas to collect and submit information consistently and accurately.</p><blockquote>The methodology for researchers to enter data into the new system can be found <a href="https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4476927">here</a>.</blockquote><p>Our secondary goal was to display the data from multiple project teams analyzing different states and regions into a unified public mapping application. We’ll talk more about the map in the next post since it’s still under development, but for now, we’ll say that it will be a technically intricate modern mapping application displaying a massive dataset in a user-friendly interface. By making the zoning map available to the public and providing a unified data source, the NZA hopes to encourage informed decision-making and promote zoning reform on a broader scale.</p><figure><img alt="A top down view of a multi-story building under construction. A crane splits the middle of the image. The ground is covered with construction materials and eqiupment." src="https://cdn-images-1.medium.com/max/1024/1*2Cze2i4q33rmtTa9JWJd7Q.jpeg" /><figcaption>Photo by<a href="https://unsplash.com/@unstable_affliction?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> Ivan Bandura</a> on<a href="https://unsplash.com/photos/0-no6ywKMPY?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> Unsplash</a></figcaption></figure><p><strong>Technical Approach</strong></p><p>To achieve the first goal, we built a user-friendly, password-protected web system enabling teams to collect and enter data about zoning codes seamlessly, and to collaborate with each other in the process. We utilized an open-source stack built in Cornell’s Azure tenant, which allowed us to deploy a web server and install Django as a backend. We also deployed an Azure Blob storage for storing uploaded files and data and an Azure Database to host the PostgreSQL instance.</p><p>For the front-end development, we used accessibility best practices and the Bootstrap 5 framework to create pages that enabled users to view, upload, import, and edit jurisdictions and zoning districts. To facilitate data pre-loading, we incorporated <a href="https://handsontable.com/">Handsontable</a>, which provided users with an intuitive way to paste zoning and jurisdiction data they discovered during their research.</p><p><strong>Results</strong></p><p>Our hard work and dedication to the National Zoning Atlas project have led to several significant accomplishments:</p><ol><li><strong>Active Participation</strong>: We successfully created a user-friendly system that has integrated the work of prior teams and is being used by all new teams, who are now actively contributing information to the National Zoning Atlas. This collaboration has facilitated the ongoing growth and development of the zoning database.</li><li><strong>Expanding Coverage</strong>: These dedicated teams are currently working on around 6,600 jurisdictions (e.g., cities and towns), comprising approximately 25,000 zoning districts. They add to data previously collected for hundreds of jurisdictions through the prior process and being imported into the atlas by the central NZA team. This extensive coverage demonstrates the immense scale and impact of the project, which continues to expand as more data is collected and submitted.</li><li><strong>Upcoming Public Map Release</strong>: Soon, the NZA will release the public map, allowing researchers, the general public, and advocates to explore and access the wealth of zoning data. This milestone will enhance transparency and provide valuable insights to inform decision-making and promote zoning reform efforts nationwide.</li></ol><p>The results we’ve achieved thus far are a testament to the potential of the National Zoning Atlas to revolutionize how we understand and engage with zoning laws. As the project progresses, we expect to see even more positive outcomes and benefits for researchers, policy advocates, and the public.</p><figure><img alt="A hand holding a pen hovers over an illustrated map of a city with points of interests and areas labeled. The left side of the photograph shows a notebook and colored pencils." src="https://cdn-images-1.medium.com/max/1024/1*3WuRMLai820Nz7mHcGWugA.jpeg" /><figcaption>Photo by<a href="https://unsplash.com/es/@arttravelling?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> oxana v</a> on<a href="https://unsplash.com/photos/qoAIlAmLJBU?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText"> Unsplash</a></figcaption></figure><p><strong>Conclusion</strong></p><p>By transforming a decentralized and error-prone data collection system into a sleek, full-stack web application, our team played a pivotal role in revolutionizing how zoning code data is understood nationwide. Once the public map is released, we will provide more information about our approach to ensuring users have the best experience. In upcoming posts, we will delve deeper into the technical aspects of the NZA Editor and share insights on the challenges and successes we encountered during the project.</p><p>Stay tuned for more updates on the National Zoning Atlas and the incredible journey from spreadsheets to a state-of-the-art web application!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=14f641329c10" width="1" height="1" alt=""><hr><p><a href="https://medium.com/greeninfo-network/creating-the-zoning-atlas-editor-14f641329c10">Creating the Zoning Atlas Editor</a> was originally published in <a href="https://medium.com/greeninfo-network">GreenInfo Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[OpenHistoricalMap Cartography — Past, Present, and Future]]></title>
            <link>https://medium.com/greeninfo-network/openhistoricalmap-cartography-past-present-and-future-97e1b7bb1cdb?source=rss----7b016ec69299---4</link>
            <guid isPermaLink="false">https://medium.com/p/97e1b7bb1cdb</guid>
            <category><![CDATA[interactive-maps]]></category>
            <category><![CDATA[cartography]]></category>
            <category><![CDATA[vector-tiles]]></category>
            <category><![CDATA[openstreetmap]]></category>
            <dc:creator><![CDATA[Tim Sinnott]]></dc:creator>
            <pubDate>Fri, 21 Apr 2023 15:26:20 GMT</pubDate>
            <atom:updated>2023-04-21T15:26:20.501Z</atom:updated>
            <content:encoded><![CDATA[<h3>OpenHistoricalMap Cartography — Past, Present, and Future</h3><p>GreenInfo Network’s cartographic journey for producing a cartographic style for <a href="https://www.openhistoricalmap.org/">OpenHistoricalMap</a> (OHM), a modern interactive map with the added dimension of time, began several years ago with a cataloging of historical maps to be used as inspiration. There were many considerations when reviewing old maps and how they could and would be applied to a brand new style for an OHM base map.</p><p>First and foremost, we needed to consider that our new map would be applied throughout all times and geographies. This would not simply be a map styled for 1930’s America or 1700’s Japan, for example. Second, this map would be a modern interactive map that had to globally style across all zoom levels. From city streets to states to continents, a basic style would need to work for all of these scenarios. Finally, similar to OpenStreetMap, the amount and variety of types of data are massive. Our map would need to be able to coherently style many different types of data. Most historical maps have a single purpose and were designed to highlight that purpose. For that reason, historical maps also greatly limit the number of types of information displayed, where OHM is an ever-expanding landscape of data.</p><p>With the above in mind, we compiled a long list of historical maps several years ago for our team to review, which is available in this <a href="https://docs.google.com/document/d/1I7f2bIqtOkCLghgKcmLK9K_cqaGTMDWxBI1zf-99mck/edit">Google Doc</a>. After some time, we settled on an aesthetic inspiration coming from the “<a href="https://www.davidrumsey.com/luna/servlet/s/5u4ok2">Geographia highway map of the United States</a>” by Alexander Gross (1950). Click the link to view this map in high resolution in David Rumsey’s Map Collection.</p><figure><img alt="Historical road map focused on the midwest and northeast United States" src="https://cdn-images-1.medium.com/max/1024/1*UIYz5CztcJiMz5SZfTU5-g.png" /><figcaption>Midwest and Northeast crop of “<a href="https://www.davidrumsey.com/luna/servlet/s/5u4ok2">Geographia highway map of the United States</a>” by Alexander Gross (1950)</figcaption></figure><figure><img alt="Historical road map focused on Central California and Nevada" src="https://cdn-images-1.medium.com/max/1024/1*nXPQPEoIB-lFXY2r8y08og.png" /><figcaption>Central California, Nevada, and Utah crop of “<a href="https://www.davidrumsey.com/luna/servlet/s/5u4ok2">Geographia highway map of the United States</a>” by Alexander Gross (1950)</figcaption></figure><p>This map seemed like a great reference for our initial cartographic design in the first draft OHM style, given its handful of bright colors and minimal amount of basic layers. We would be able to quickly prototype a map that is immediately recognizable as a nod to vintage road maps.</p><p>Especially inspirational in the Alexander Gross map are the dusty dusky blue gray water, the warm off white land color, and the bright colors and stark contrast of the highways against the light background. These distinct colors would be the foundation of our vintage road map style, from which we would build out more subtle, complementary symbologies for land cover types, buildings, icons, and more.</p><figure><img alt="Zoomed in view of Southern Louisiana of “Geographia highway map of the United States” by Alexander Gross (1950), highlighting dusky blue waters and black and red highways" src="https://cdn-images-1.medium.com/max/791/1*gp4FOK5L-JPBtaA4WoF95g.png" /><figcaption>Southern Louisiana crop of “<a href="https://www.davidrumsey.com/luna/servlet/s/5u4ok2">Geographia highway map of the United States</a>” by Alexander Gross (1950)</figcaption></figure><h3>Design</h3><figure><img alt="Prototype design created in Mapbox Studio, focused on Vermont and New England, influenced by vintage maps" src="https://cdn-images-1.medium.com/max/1024/1*Cf5CYQGRGgAKXAA8Rnqaxg.png" /><figcaption>Prototype design created in Mapbox Studio, focused on Vermont and New England, influenced by vintage maps</figcaption></figure><figure><img alt="Prototype design created in Mapbox Studio, focused on Chicagoland, influenced by vintage maps" src="https://cdn-images-1.medium.com/max/1024/1*_62Lcoe1-SgZ7_nbD91zgQ.png" /><figcaption>Prototype design created in Mapbox Studio, focused on Chicagoland, influenced by vintage maps</figcaption></figure><figure><img alt="Prototype design created in Mapbox Studio, focused on downtown Chicago, influenced by vintage maps" src="https://cdn-images-1.medium.com/max/1024/1*u4NU-tWPbcrqMxIeoIXvog.png" /><figcaption>Prototype design created in Mapbox Studio, focused on downtown Chicago, influenced by vintage maps</figcaption></figure><p>This first approach to an OHM style attempted to take global data at all zoom levels and symbolize it in a way that referenced 1950s road maps. Highways were given a rust red color to make them stand out, as was very typical with vintage maps of this era. We pulled the water color almost directly from the 1950 Alexander Gross map posted above. This was a great starting point to get the OHM cartography ball rolling.</p><p>While working on initial iterations, we used Mapbox Studio and their data as to test the style, since Mapbox pulls data from OpenStreetMap (OSM) which has a large amount and variety of data to work with, and an interface to quickly prototype styles.</p><p>After we knew what we wanted the style to look like, we left Mapbox and used Maputnik. Using Maputnik, we could apply the styling we created in Mapbox to OHM data (rather than Mapbox/OSM data), and we began to test the cartography across different locations, zoom levels, and times throughout history. As the style was reviewed, it became clear that the strong focus on highways and car-centric roads was not the best for all locations at all times. We toned down the saturation of water and shifted all roads to white. Roads were still highlighted by width, cases, and shield symbols and labels. This gave us more of a basic cartographic palette to work with, and allowed us to emphasize a variety of types of data, not simply transportation layers.</p><p>After some time with this new iteration of the style, some OHM users and contributors expressed how they were missing the initial draft’s variance in color that symbolized different road types, and thought colors had done a nice job of differentiating the types of roadways in the application. We listened to feedback and returned to applying colors to roads and created a few new options, but desaturated the new style compared to our original designs. We settled on this symbology, as seen in Mapbox Studio:</p><figure><img alt="Prototype map design created in Mapbox Studio, focused on Downtown Chicago, updated with less saturated colors." src="https://cdn-images-1.medium.com/max/1024/1*zf2qF1IbmZv6RTt8Ie8lIA.png" /><figcaption>Prototype map design created in Mapbox Studio, focused on Downtown Chicago, updated with less saturated colors.</figcaption></figure><p>Here is that basic style applied to OHM data in Maputnik in the London area:</p><figure><img alt="Cartography with more mellow colors applied to OpenHistoricalMap data, map focused on London, UK." src="https://cdn-images-1.medium.com/max/1024/1*v6XsBXipaOZTrLYmdEr2LQ.png" /><figcaption>Cartography with more mellow colors applied to OpenHistoricalMap data, map focused on London, UK.</figcaption></figure><p>Interestingly, and maybe not surprisingly, the evolution of the OHM style started to push away from an historical map style and towards a modern interactive map style. As we bring in additional layer types to OHM, we need to be able to balance all of our layers’ symbologies more equally, and this is the focus in our current approach.</p><p>In the future, there should be opportunities to bring new influences from vintage maps, like the <a href="https://www.openhistoricalmap.org/#map=4/19.89/-19.86&amp;layers=W&amp;date=2010-05-10&amp;daterange=1923-01-01,2023-12-31">Woodblock</a> and <a href="https://www.openhistoricalmap.org/#map=8/35.183/136.445&amp;layers=J&amp;date=2020-03-30&amp;daterange=1923-01-01,2023-12-31">Japanese</a> Scroll inspired styles, and we should consider more of these options moving forward!</p><h3>How Data Affects Design</h3><p>One of the latest challenges for the OHM base map cartography has been a lack of data, especially at a regional and global level. Below are a couple of screenshots of the most recent regional look in the production version of the OHM map style:</p><figure><img alt="Previous OpenHistoricalMap country-level zoom without landcover and hillshade, focused on the United States" src="https://cdn-images-1.medium.com/max/956/1*lH3g2V3UzQqd7FspVRxi3w.png" /><figcaption>Previous OHM country-level zoom without landcover and hillshade, focused on the United States</figcaption></figure><figure><img alt="Previous OpenHistoricalMap multi-country-level zoom without landcover and hillshade, focused on Europe" src="https://cdn-images-1.medium.com/max/811/1*k8DRzF-NkHFEBV09ZmpJ8A.png" /><figcaption>Previous OHM multi-country-level zoom without landcover and hillshade, focused on Europe</figcaption></figure><p>This map had been a nice base, but it was very limited in what it was displaying. The limited data only included land and water; and country and state boundaries and labels. To supplement our sparse data, we had some recent discussions about adding in a few global zoom levels of both hillshading and land cover / land use (LCLU). These additional layers would add an engaging and informative reference, while helping to visually resolve that issue.</p><p>To that end, we built out raster tiles that combine both hillshading and LCLU that work well at a global level, and fade to transparent at a regional level. At a global level, these relief and landcover tiles will be static (they will not change over time). At the regional level, current and future OHM land cover data will replace this static layer. The global LCLU data have been styled in a manner very similar to the current land cover data in OHM to maintain a consistency in our cartography.</p><p>To determine the best color saturation for these raster tiles, we proposed three versions: one at full color, one with a 25% white overlay on top of LCLU data, and one with a 50% white overlay on top of LCLU data.</p><figure><img alt="Trio of maps focused on the United States showing three different saturations of land cover" src="https://cdn-images-1.medium.com/max/1024/1*kBzKOgwZEvqxlQzLhdtWGg.png" /><figcaption>Trio of maps focused on the United States showing three different saturations of land cover</figcaption></figure><p>We ultimately decided to split the difference and move forward by integrating the 25% white overlay, the middle option above. This seemed to offer just the right level of detail without being too distracting, and this layer fades nicely into the rest of the OHM data as you zoom in. You can see our latest basemap live at <a href="https://www.openhistoricalmap.org/">https://www.openhistoricalmap.org/</a>.</p><p>If you would like to contribute to OpenHistoricalMap, please consider <a href="https://www.openhistoricalmap.org/user/new">registering for an account</a>!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=97e1b7bb1cdb" width="1" height="1" alt=""><hr><p><a href="https://medium.com/greeninfo-network/openhistoricalmap-cartography-past-present-and-future-97e1b7bb1cdb">OpenHistoricalMap Cartography — Past, Present, and Future</a> was originally published in <a href="https://medium.com/greeninfo-network">GreenInfo Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[A Conversation about the History of GIS with Madelyn Glickfeld]]></title>
            <link>https://medium.com/greeninfo-network/a-conversation-about-the-history-of-gis-with-madelyn-glickfeld-ce4cd3274df1?source=rss----7b016ec69299---4</link>
            <guid isPermaLink="false">https://medium.com/p/ce4cd3274df1</guid>
            <dc:creator><![CDATA[Vanessa Knoppke-Wetzel]]></dc:creator>
            <pubDate>Wed, 12 Apr 2023 14:58:42 GMT</pubDate>
            <atom:updated>2023-04-13T14:13:36.660Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="Madelyn Glickfeld" src="https://cdn-images-1.medium.com/max/1024/1*QRCwkkJ6pDyfwS2Y9umW2A.jpeg" /></figure><p><em>Madelyn Glickfeld has served on the GreenInfo Network Board for 20 years, and is retiring soon. Before she leaves us (only in Board capacity 😊), I wanted to interview Madelyn to hear about her experiences with GIS throughout her career.</em></p><p><strong>Vanessa: I read your </strong><a href="https://www.ioes.ucla.edu/person/madelyn-glickfeld/"><strong>UCLA biography</strong></a><strong> and know you have a strong history of public environmental leadership and service, but I would love to hear you talk about what you have been doing the last few years in your own words, as well as give a summary of your career — anything you would love to highlight.</strong></p><p>Madelyn: My career since the 80s has been a combination of public service and full-time work. I have served time with several other nonprofits, and also owned a consulting firm. Since the 90s, I have worked for UCLA.</p><p>At the start of my career, I did planning consultant work at the time. I ended up working on the first national environmental impact statement (EIS) that was ever submitted to the federal government — an expansion plan of the State of Alaska Airport System! At the time, I had no idea what an EIS was supposed to have in it. My firm did not include a travel budget for me, so I had to communicate with the project manager who would bring back information that I needed. No internet yet — no world wide web. I had to figure it out as I went.</p><p>I worked on many environmental projects at the very cusp of the 70s, when other state and federal environmental legislation was coming in, focused on land and preserving open space. I thought about how to protect land: How can we use the law to avoid development in places that were important? I learned how to use<a href="https://en.wikipedia.org/wiki/Ian_McHarg"> Ian McHarg</a>’s methodologies for site planning to do overlays of key information with aerial photography, base maps and overlaps with a graphic artist inserting color adhesives for different categories. Ian McHarg historically was very influential in the environmental movement, as he brought ecological planning methods into landscape architecture, city planning, and public policy spaces. He was also influential in the use of maps.</p><figure><img alt="An illustration depicting developed areas that did not destroy trees and natural elements in order to have buildings exist in the area" src="https://cdn-images-1.medium.com/max/1024/1*c9p_0yWn3E4kn606S0_eDg.png" /><figcaption>Ian McHarg emphasized designing with nature in his work (<a href="https://www.wiley.com/en-us/Design+with+Nature%2C+25th+Anniversary+Edition-p-9780471114604">Design with Nature</a>)</figcaption></figure><p><strong>Vanessa: Within all this amazing environmental work, of course a lot of it likely had to be mapped. I would love to hear about how you started working with GIS and maps at the start of your career and how that relationship progressed over time.</strong></p><p>Madelyn: Using McHarg’s methodologies, I developed a means for site planning that included mapping. My work has always been very spatial: where things were, how they relate to other things… it has always been an important factor. For the site planning, I used McHarg’s overlay method, which is the core of what GIS is today: building data layers on top of each other in order to visualize and analyze the layers in order to make informed decisions. I layered a bunch of transparent overlays of the topography, vegetation, aspect, waterways, and other important data layers that you needed to use to understand where it was okay to develop, and where to avoid development plans.</p><figure><img alt="A graphic depicting the layering structure of data layers — stacked on top of each other, with different labels for each layer. Layer examples: hydrology/groundwater; geology; soils; etc" src="https://cdn-images-1.medium.com/max/850/1*ME26IUKWT0KiYQG5_7Lkiw.png" /><figcaption>McHarg’s layer representation</figcaption></figure><p>In my consulting firm we often used mapping, as it was incredibly important. It just was more arduous than today and it took longer, as everything was completely manual. We assembled from available data that we had: from aerial photographs, field surveys of land, and many (a lot) of USGS printed maps. We would calculate how to scale the maps up and down and interpolate scaling and topography. We also usually had a graphic artist who would help us in some of our work. We had to adhere colors to the maps to show what represented what, where. What you can do now in GIS is easy and fast, but it just took a really long time back then, and required a few people to get it done.</p><p>Recently, I have worked on a project in Salinas Valley (California) that has not yet had any mapping work, but can — and should! This project works with agricultural farm workers that live in small communities that are, per se, not allowed to be in the cities that have not historically been allowed to be built in cities outside. Farm worker housing communities have developed outside of cities, and some of the issues are that all of them have shallow groundwater, and many have polluted water due to a variety of reasons. In the project, I worked with engineers to design and build advanced water treatment facilities designed and sited in the communities for clean drinking water, and make the area safe for them to live and work in.</p><p>The required intake groundwater monitoring shows that over the last five years, nitrate pollution has steadily increased. It raises several issues where GIS information is needed. Where are the increases in pollution and where is it coming from? How has the long drought and this year’s soaking affected pollution? There likely is an increase of accumulation of fertilizers from nearby soils in the groundwater as well. How do we help to prevent the continuing deterioration of the water, when the water already was not good to begin with? There are so many other questions to answer here as well, but because I am such a spatial person, I know these questions can be answered by mapping all of the water intake data and groundwater areas, and doing analyses to help determine what some of the issues are, as well as well as provide guidance to what some potential solutions could be.</p><p>I am retiring, but I hope future studies of groundwater pollution trends include maps and spatial analysis.</p><p><strong>Vanessa: I hope so too! The project work is important, and also sounds incredibly interesting.</strong></p><p><strong>You sent an interesting </strong><a href="https://www.esri.com/en-us/what-is-gis/history-of-gis#:~:text=GIS%20Goes%20Commercial&amp;text=In%20need%20of%20analyzing%20an,Esri%20into%20a%20software%20company"><strong>timeline/article</strong></a><strong> about the history of GIS. The timeline specifically flagged when “GIS went commercial” in the 1980s. I am really curious to hear about (1) your mapmaking experiences <em>prior </em>to GIS “going commercial”; and (2) what changes you noticed between 1990 and now, as GIS became more and more accessible to more people — more software and technologies became available for practitioners to use, more open source options, more education around GIS, and more variety of products (print, online).</strong></p><p>Madelyn: I remember in the 60s, 70s, and 80s, there was almost no GIS at all, at least not in the way that we know it today: it was in a nascent state.</p><p>One of the most interesting experiences I had was when I was really interested in making maps and went to a storefront in Redlands. The storefront office housed a one person — Jack Dangermond (later founder and CEO of Esri) — mapping and analysis/planning shop. I wanted to see how he was making his maps. He was making them the same way we made them — inspired by McHarg, using layer overlays, etc! He talked a lot about what things could look like in the future. He taught himself not only the technology of doing better manual mapping, but also how to program and translate early academic GIS software into his first desktop computer GIS. It was exciting to have been able to see all of that!</p><p>Today, mapping and GIS systems are so complex. I used GIS to identify the best way to pair strong water systems in an area with many small water systems in Southern Los Angeles County. Lots of factors need to be considered and I had a hard time balancing the complexity of maps with the need to communicate with maps in a way people could readily understand.</p><p>I had a hard time translating from the really complex academic research and data work that I was doing, into something more simpler so people could use the research, data, and maps in order to use the research results. That “translation” is so important: making the map digestible, easy-to-read, well designed, so that it isn’t just for internal use, but for external use so a variety of people can use the maps. Without GreenInfo Network’s Maegan Leslie Torres, I wouldn’t have been able to make the research accessible.</p><figure><img alt="A beautiful map of the Northern great plains, which has a lot of data on it. The map example is meant to highlight the visualization of complex topics" src="https://cdn-images-1.medium.com/max/1024/1*49iJAsZI7ii65V6odDKIMw.png" /><figcaption>An <a href="https://www.greeninfo.org/work/project/tnc-resilient-northern-great-plains">award-winning example</a> of a map series that Maegan created to help share a complex message.</figcaption></figure><p><strong>Vanessa: This is such an interesting juxtaposition to me, hearing you talk about how complex your work and research — so many maps and data pieces involved. This is interesting compared to when you were starting out in your career, as you noted that GIS wasn’t so complex at the time.</strong></p><p><strong>The complexity of data and data analyses these days really does often require expertise and guidance in order to make the information digestible outside of academia, which, for geospatial data, often requires understanding cartography/data visualization requirements. There are so many reasons why I love Cartography, but taking a complex problem and figuring out how to “translate” it through the analysis and design is one of the many reasons.</strong></p><p><strong>This conversation is also interesting to me because despite the increase in complexity, there also was an increase in accessibility — more people can 1) make maps and 2) access maps (read maps, hold maps, use maps) — compared to when your career started. You noted earlier that it was a lot of labor in order to produce a couple maps, so maps were not produced at the scale or speed that they are today.</strong></p><p><strong>Technology and accessibility conversations always make me think of my thesis work, which was in maps made before the 1900s. Maps were still important and useful back then, but they tended to require a lot more labor and were far less accessible. The average person did not see the historical maps that many love to look at and collect today.</strong></p><p><strong>And yet, despite this increase in being able to produce more maps and read and use maps, there also is just so much data available as well, as you said. With so much data accessible, it feels easy to try and add it all and use it all. But, adding ALL the data to a map doesn’t necessarily mean the map is readable or digestible right away. Maps today really often need cartographers and other data visualization experts to help make sure the message of the map is understandable, so it is awesome that you worked with Maegan and other mappers to make sure your messages came across.</strong></p><p>Madelyn: This complexity is another reason I love mapping: it has breadth and depth. It can be used in a complex way and you don’t need to change anything if only you are using it or your team is using it and it is incredibly useful to make a difference, but the maps and GIS work still have a lot of power if you take advantage of cartography and create a map in such a way that it is shareable with a variety of audiences.</p><p>An example of this kind of project is something else that GreenInfo Network helped with. I worked with a brilliant climate scientist at UCLA who was downscaling climate change projections for use in regions of California. The first downscaling projection was done for Los Angeles County, and we had to learn how to communicate the local impacts of climate change on maps. GreenInfo Network’s founder, Larry Orman, helped with that a bit. That work then was published and it went far. The data from the climate scientist was actually connected to the physical characteristics of a place through the GIS and mapping work: topology, demographics, heat effects, and other data that may have gone into it. The way GIS technologies have advanced and jumped so far ahead to be so immediately useful in a variety of ways: GIS has had such an incredible jump ahead compared to when I started my career! You can do anything and everything now — it is so complex.</p><p>With that said, everything is online, and that can be a little harder for making things more accessible to everyone. You think about people who have seen all of GIS’ iterations over time and tried to keep up and stay in-the-know, it’s difficult to stay caught up. We see all the changes and see where things were to where they grew to, to where they are now: there have been many different GIS revolutions at different stages. At GreenInfo Network as well, when Dan Rademacher, GreenInfo Network’s current Executive Director came, he brought in some of that revolution too. You (Vanessa) came in with a diverse background in historical cartography, web cartography, education, and more. When María del Carmen Lamadrid came in as a design researcher and UX designer, GreenInfo Network really started doing work that asked and worked towards how to use mapping to engage people — the way in which I wanted to when I was a planner.</p><p><strong>Vanessa:</strong> <strong>How has GIS had an impact on policy and social-good work throughout your career?</strong></p><p>Madelyn: I think that without GIS mapping, we would never have been able to acquire and protect as many resources as we have in California. We just wouldn’t have had any idea of where the most important resources were. Visualizing where things are is really important. Being able to take that. It has been used to buy “this parcel,” “that parcel.” It has allowed us to be able to step back and see all that data, analyze it, and do more with it, for environmental issues, but also for a huge area that is incredibly important: equity work.</p><p>Front-line communities — often called disadvantaged communities, but they do not like being called that, so I call them front-line communities — are systematically disadvantaged in <em>literally</em> everything. Environmentally speaking, they often are forced to live near polluted sites; are exposed to polluted water, air, and stormwater. GIS and mapping has helped them to highlight the inequity in order to work towards remedying through advocacy, policy work, protest, and more.</p><figure><img alt="The visual is a screen shot from a map application that users can use to review the water rights in their area in California, in order to use the information for advocating for themselves" src="https://cdn-images-1.medium.com/max/1024/1*zUGikJfap7xhGY7c1wMqRg.png" /><figcaption>GreenInfo Network worked on a <a href="https://www.greeninfo.org/work/project/searching-california-water-rights">California Water Rights project</a> that is a great example of how GIS helps to support communities, and gets relevant and important information into their hands.</figcaption></figure><p><strong>Vanessa: Yes! I agree. Maps have so much power to unearth things people do not know, or do not discuss, or try to ignore. The maps tell the stories that complicated data may not immediately be able to convey. This is yet another reason why I love making maps.</strong></p><p>Madelyn: Yes, we are similar: we care so much and are drawn to this work because it is meaningful and because it has the power to change things. One of the most important things in my career was discovering (and being shocked) that so many communities didn’t have safe drinking water. In California, supposedly a state that “has everything” and the richest state in the union, a million people live without clean water<em> or</em> without any water. I stopped believing that water is just “a given”. A lot of the details about issues such as these would never have been understood without GIS, and solutions would not be possible without mapping it all.</p><p><strong>Vanessa: You have been a Board member at GreenInfo Network for 20 years and you have previously mentioned that GreenInfo Network feels like coming home. I would love to hear more about this, and what it has been like working with GreenInfo Network over the years.</strong></p><p>Madelyn: It feels like coming home because I really am such a spatial person. In fact, I think my career has been hindered without being able to do GIS from the start. I was doing so many other things, and so I never had the time to allocate to frequently use GIS to become adept at it. I have always been a frustrated GIS person. What I would have done — if I had a chance, would have been jumping into a GIS position at the get-go.</p><p><strong>Vanessa: I know how you feel. I know for a lot of my career, there have been so many software, technologies, programming languages that are “good to know” and they are constantly changing and growing. It is impossible to keep up with it all, and I also never had the time to keep up in the way that I often wanted to. At some points in my career, I did get stressed thinking I didn’t know “enough things” because there is just SO MUCH out there. But I now know that it is absolutely OK to “not know everything,” because the reality is in the GIS world today, it is impossible to know it ALL.</strong></p><p>Madelyn: Exactly. When I was starting out, I think I was intimidated by it in some ways, but I have always found great people to work with who could do it for me. One of the best people I have worked with was Larry Orman, who used a lot of GIS in his work, both before and after he founded GreenInfo Network. I hired GreenInfo Network when I was at the California Natural Resource Agency, where I helped direct work on conservation mapping for the entire state. He and his staff did a lot of amazing work for us, including creating one of the first internet mapping tools. Larry and the team worked with Esri at the time to build a California data inventory and programming. Later, Larry said he would take the database he helped build with me and make it into something bigger by adding local parks and all kinds of other data. So I was happy to see this database grow. The work that Larry did with me at the Natural Resources Agency was the foundation for the<a href="https://www.calands.org/"> California Protected Areas Database</a>, which was built subsequent to our work in 2001–2002. It continues to be updated by GreenInfo Network with the Natural Resources Agency today, and it is such an essential database.</p><figure><img alt="a screenshot of the California Protected Areas Database website" src="https://cdn-images-1.medium.com/max/1024/1*0omyv4EWncgzCmRg7AhNBg.png" /><figcaption>Today, the California Protected Areas Database that GreenInfo Network developed now holds 15,000 parks and preserves owned by over 1,000 agencies/organizations, protected as permanent open space. A dozen key agencies and foundations have helped support this work. You can learn more about it <a href="https://www.greeninfo.org/work/project/cpad-the-california-protected-areas-database">here</a>.</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ce4cd3274df1" width="1" height="1" alt=""><hr><p><a href="https://medium.com/greeninfo-network/a-conversation-about-the-history-of-gis-with-madelyn-glickfeld-ce4cd3274df1">A Conversation about the History of GIS with Madelyn Glickfeld</a> was originally published in <a href="https://medium.com/greeninfo-network">GreenInfo Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[A Tale of Two Senates]]></title>
            <link>https://medium.com/greeninfo-network/a-tale-of-two-senates-4bb18fa80832?source=rss----7b016ec69299---4</link>
            <guid isPermaLink="false">https://medium.com/p/4bb18fa80832</guid>
            <category><![CDATA[mapping]]></category>
            <category><![CDATA[democracy]]></category>
            <category><![CDATA[redistricting]]></category>
            <dc:creator><![CDATA[Dan Rademacher]]></dc:creator>
            <pubDate>Thu, 06 Apr 2023 23:08:14 GMT</pubDate>
            <atom:updated>2023-04-06T23:08:56.754Z</atom:updated>
            <content:encoded><![CDATA[<p>How 3+ million Californians ended up with an assigned state senator.</p><h3>The Short Version</h3><p>Every 10 years, California redraws the boundaries for Congress, State Assembly, and State Senate.</p><p>Congress and Assembly are up for election in full every 2 years. So when the new districts get approved, they all roll out at once.</p><p>The State Senate is different. They serve four years and half the body is re-elected every two years. Senators running in even-numbered districts run in one cycle. Senators in odd-numbered districts run in the next.</p><p>That means you can never replace all the districts at once. Half the districts come on line in one election, and then the other half come two years later. In the meantime, for the first two years of a cycle (2023 and 2024 in this case), things are just confusing. Sometimes old and new districts overlap, so some folks get two senators. Other times, old and new don’t even come close to each other, so large areas have no senator at all.</p><p>The state’s redistricting commission publishes <a href="https://www.wedrawthelinesca.org/final_maps">final maps</a> for all levels in a simple, clear page. That’s good, but it’s also very easy to grab those shapes, put them on a map, and think everything is clear and settled.</p><p>It’s not, at least for folks who live in Chico or the Eastern Sierra. The new senate districts in those areas haven’t come up for election yet. So where does that leave the 100,000 people who live in Chico? Or advocates who might be looking to support conservation in the Eastern Sierra?</p><p>It turns out that the Senate Rules Committee assigned a senator to cover those areas, and that’s what you see if you search on the official <a href="https://findyourrep.legislature.ca.gov/">Find Your California Representative</a> web application. But that application also says, in bold red letters:</p><p>As a result of redistricting, your area may be represented by two Senators, or may not be represented by any Senator, during the 2023–24 legislative session. If your area is unrepresented as a result of redistricting, the Senate Rules Committee has assigned a Senator to provide constituent services to your area during the 2023–24 legislative session.</p><p>OK, so we know the Senate appointed someone to cover those areas. But where’s the shapefile that tells us who those senators are? How much extra area are they covering? It took us a while and had some twists and turns, but we eventually found a source, which we’ll share later in this post.</p><p>We also figured out that more than 3 million people are in areas with a senator assigned to them. <strong>If those folks all lived in one place, they’d be the second-largest city in California after Los Angeles.</strong></p><h3>The Long Version</h3><p>In January 2023, a long-time client asked us to update the legislative districts within a project map and gallery we built a few years ago. The application in question has nothing to do with redistricting, but it does include legislative districts for California, to help elected leaders and advocates see how public projects are distributed across important political and social geographies.</p><p>When the request came in, I thought: Easy! We have worked on multiple redistricting projects so we know all about this. We helped create <a href="https://planscore.org/">PlanScore</a>, we redesigned and rebuilt <a href="https://redistricting.lls.edu/">All About Redistricting</a>, we support the tech side of <a href="https://redistrictingdatahub.org/">Redistricting Data Hub</a>, we helped create <a href="https://drawcongress2020.law.stanford.edu/">DrawCongress</a>, and we’re currently engaged with leading folks at the University of Florida and the College of William and Mary on yet more projects related to redistricting, voting laws, and more.</p><p>OK, so this should take like 2 hours, right?</p><p>Wrong.</p><p>At first, it was all going swimmingly: We knew we could easily get boundary files from this page: <a href="https://www.wedrawthelinesca.org/final_maps">Final Maps — We Draw the Lines CA</a>. We also needed to list the elected officials by name tied to their district. We didn’t find a ready source for district boundaries with names that we were confident would be current, but it was quick work to make clean tabular lists of district numbers and members from lists on the <a href="https://www.congress.gov/members?pageSize=100&amp;q=%7B%22member-state%22:%22California%22,%22congress%22:118%7D">Congress</a>, <a href="https://www.assembly.ca.gov/assemblymembers">Assembly</a>, and <a href="https://www.senate.ca.gov/senators">Senate</a> websites.</p><p>In no time, we had updated files for Congress, Senate, and Assembly that we had joined on district numbers, so we had shapes on the map and names attached to shapes. We checked by number to make sure that the join was correct, and that no one was missing.</p><p>Ship it!</p><p>We did. And everything seemed fine. Until.</p><p>We heard that Senate District 19 was “showing wrong map locations.” And, wow, was it ever!</p><p>Senator Monique Limón represents SD 19 centered on Santa Barbara. You can’t mistake that if you look at her <a href="https://sd19.senate.ca.gov/district">online district map</a>.</p><figure><img alt="Map showing coastal area around Santa Barbara and the northern Channel Islands shaded in, Senate District 19." src="https://cdn-images-1.medium.com/max/653/0*D4fz8c9cuwlwGQCI" /><figcaption>Senate District 19, until January 2025.</figcaption></figure><p>OK, so how did we have SD 19 way over in the Eastern Sierra? The green area on the right is the SD 19 in California’s new redistricted Senate Boundaries, and the pink on the left is the area Limón currently represents:</p><figure><img alt="Map of two separate shaded areas: One coastal near Santa Barbara and the other very large covering much of eastern Riverside and San Bernardino counties." src="https://cdn-images-1.medium.com/max/849/0*RWd40XOF41L9UyIC" /><figcaption>Left, in pink, in SD 19 until Jan 2025. Right, green, SD 19 after Jan 2025.</figcaption></figure><p>Not even close!</p><p>We thought we’d simply done something wrong, but we couldn’t figure out what it was until we read through <a href="https://ballotpedia.org/California_State_Senate_District_19">the Ballotopedia page for Senate District 19</a>, which states what is going on here more clearly than anything else we found:</p><blockquote>As of February 2023, two sets of overlapping California State Senate maps are in effect because the state’s adopted redistricting plan takes effect in two phases. <strong>Even-numbered districts use boundaries from the post-2020 census map, while odd-numbered districts use boundaries adopted after the 2010 census.</strong> As a result, an area may be represented by two Senators, or may not be represented by any Senator, during the 2023–24 legislative session. The state Senate map enacted after the 2020 census will fully take effect after the 2024 elections.</blockquote><p>Aha! At last, a clear statement that we could also turn into easy database queries:</p><ol><li>Load the old districts</li><li>Load the new districts</li><li>Query both and if the result is ODD, use the old district and if EVEN, use the new district.</li></ol><p>We did that but we weren’t yet looking too closely at this part of the explanation:</p><blockquote>As a result, an area may be represented by two Senators, or may not be represented by any Senator, during the 2023–24 legislative session.</blockquote><p>Surely that’s just a few tiny spots, right? Like some edge cases most folks wouldn’t notice.</p><p>Well, not exactly. Here’s where we see big holes in the Sacramento Valley, the Central Coast Gabilan Range, and the Mojave:</p><figure><img alt="Map of combination of old odd and new even districts, with large holes around Chico, south of Gilroy, and in eastern San Bernardino and Riverside counties." src="https://cdn-images-1.medium.com/max/1024/1*bs6lWXtFbM1TNyRhS1aKew.png" /><figcaption>The combination of old even and new odd districts</figcaption></figure><p>At this point, we were entertaining the idea of just eliminating Senate District as a search option for two years, since it seemed too hard to explain what the heck is going on.</p><p>But I kept coming back to the <a href="https://findyourrep.legislature.ca.gov/">Find Your California Representative</a> tool. If I search an address in Chico, I see that Brian Dahl (R-SD01) has been assigned to that area. And <a href="https://dahle.cssrc.us/1st-senate-district">the map on his website</a> reflects that.</p><p>If the Find Your Rep tool could answer correctly, there must be a spatial file under the hood somewhere. I fired up Developer Tools, switched to the network panel, and searched for the Chico Old City Hall address, 441 Main St, Chico, CA 95928.</p><p>That led me to this ArcGIS REST service: <a href="https://services5.arcgis.com/0CGHmi5SBMvfD65u/arcgis/rest/services/AD_SD_2011_2021_SDRulesAssign/FeatureServer/5/">https://services5.arcgis.com/0CGHmi5SBMvfD65u/arcgis/rest/services/AD_SD_2011_2021_SDRulesAssign/FeatureServer/5/</a></p><p>There’s probably a better way to do it, but to get all the data out of an endpoint like that, I learned a trick of going to <a href="https://services5.arcgis.com/0CGHmi5SBMvfD65u/arcgis/rest/services/AD_SD_2011_2021_SDRulesAssign/FeatureServer/5//query">the Query form</a>, entering “1=1” into the WHERE field, “*” into the “Out Fields:” form, and GeoJSON into the format.</p><p>That gave us a 50+MB GeoJSON with over 2000 polygons in it and fields for the following:</p><ul><li>SD_2011: old districts, still in play for odd-numbered districts now</li><li>SD_2021: new districts, in play for even-numbered districts</li><li>SD_assign: the missing piece!</li></ul><p>The file is a bit squirrelly to use. Basically, for any location, use the following rules:</p><ul><li>Use the assignment in SD_assign regardless of the other two values.</li><li>If the result in SD_2011 is an odd number, use that. The 2011 odd districts are in place until the end of 2024.</li><li>If the value in SD_2021 is even, use that. The SD_2021 even districts came in with the November 2022 election.</li></ul><p>This can result in two senators assigned to a single location. That’s OK and it’s accurate to the situation until the November 2024 elections.</p><p>Here are the areas where a senator has been assigned — filling in the gaps in the previous map:</p><figure><img alt="Map of assigned senator areas, around Chico, south of Gilroy, and eastern San Bernardino and Riverside Counties" src="https://cdn-images-1.medium.com/max/1024/1*7csyu93VlqyBELvhSrUT1A.png" /><figcaption>Areas where senators were assigned.</figcaption></figure><p>I will say that this spatial data is not my favorite. At GreenInfo, we like to avoid slivers and gaps as much as possible — meaningless little edge effects in data that end up clogging analysis and causing confusion.</p><p>Here’s an area in San Leandro that looks like textbook slivers:</p><figure><img alt="Zoomed-in map of San Leandro with tiny, serpentine shapes colored in where districts were slightly misaligned." src="https://cdn-images-1.medium.com/max/970/1*2TiH5ltoUQGhUea_y-93HQ.png" /></figure><p>Still, we can set that aside and confidently assign at least one senator to any point in California.</p><h3>Mapping Semi-representation</h3><p>In any case, with the solution above, we were done with what was needed for the project at hand. But I couldn’t quite let go of this issue. It’s just so weird that for 20% of the time, some number of Californians are represented by a senator they never voted on.</p><p>How many Californians? I didn’t know, but we happened to have some high-resolution population data we developed for another project that would make it pretty easy to find out. For a park equity and access project we’ve been working on in collaboration with UCLA, The Wilderness Society, and Resources Legacy Fund, we used <a href="https://research.facebook.com/downloads/high-resolution-settlement-layer-hrsl/">Meta’s High Resolution Settlement Layer</a> combined with more recent data from the American Community Survey to estimate the number of people in each cell of a 30 meter grid covering California. (Read more about that over at <a href="https://parkaccess.org/about/">parkaccess.org</a>, in the section “Understanding where people live” .)</p><p>Using that data, I loaded the 11 million points with population into Google’s BigQuery, uploaded the shapes from our Find Your Rep data where SD_assign had a value, and then intersected just the population points within those areas. So that gives us an estimate of how many people have Senators assigned to them in 2023–2024.</p><p>The answer is about 3.3 million people. They are scattered around the state, but if they were all in one city, they would be the second-largest city in California after Los Angeles.</p><p>Here it is summarized by which district voters were assigned, to, sorted by most to least assigned voters:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/7cf95263ebc3b8c7c53a8889f0acb096/href">https://medium.com/media/7cf95263ebc3b8c7c53a8889f0acb096/href</a></iframe><p>And <a href="https://clausa.app.carto.com/map/d5a1d730-0c97-4328-9567-b54e39e40bc7">here’s an interactive map</a> that shows where all those people are, with a widget that calculated the viewport (note that the widget rounds down to 3 million on the full state display)</p><figure><img alt="Image of an interactive map showing detailed info on where assigned-senator  voters live" src="https://cdn-images-1.medium.com/max/1024/0*ffZxtU6PAHyBCICo" /><figcaption>Access this map on <a href="https://clausa.app.carto.com/map/d5a1d730-0c97-4328-9567-b54e39e40bc7">CARTO.com</a></figcaption></figure><p>When we first fell down the rabbithole of “some people have no senator,” I asked for some advice about this from Justin Levitt and Doug Spencer, who run <a href="https://redistricting.lls.edu/">All About Redistricting</a> (which we were honored to help redesign and rebuild a few years ago).</p><p>It turns out that a few states do deal with this issue better than we do in California — they truncate terms once a decade so everyone starts over with new districts. But most states do what California does: Assign legislators to provide constituent service in gap areas.</p><p>I wonder about that phrase “constituent service.” It’s important to have that, but you also want your elected officials actively looking out for your interests and advocating for the people and places in your area that matter to your communities. It’s hard to imagine that happening when a region is tacked onto an existing district for a couple of years.</p><p>However, apparently the courts have generally not seen this as a problem. People get to vote for their senator eventually. The paper <a href="https://digitalcommons.nyls.edu/cgi/viewcontent.cgi?article=1034&amp;context=redistricting_resources">Staggered Elections and Redistricting</a> provides more legal overview and also a list of states with staggered elections and how they handle redistricting.</p><p>One improvement I’d like to see is simply making the data of who is being assigned easier to find and understand. To be fair, the California Senate seems to have done a decent job of highlighting assigned areas on the district map pages of each senators’ websites where that’s relevant (at least on the 8 or 10 that I checked).</p><p>At the end of this process, we discovered <a href="https://sdmg.senate.ca.gov/Current">this page</a> from the Senate Office of Demographics (I didn’t know there was such an office!). That page explains things reasonably well, but without any maps or data. And even then we stumbled on that URL as a link from an individual senator’s website. I’m still not sure how to get there from <a href="https://www.senate.ca.gov/">senate.ca.gov</a>. You won’t find it on the site map.</p><p>Even better than a page on the Senate website would be making this part of our nonpartisan redistricting commission’s work, hosted at <a href="https://wedrawthelines.ca.gov/">WeDrawtheLines.ca.gov</a>. They could publish a single clear and well-documented layer of the actual districts in effect for 2 years each decade. maybe there could even be some public oversight and input into who gets assigned to whom. Maybe that’s a recipe for delay, but it also sounds more like democracy than getting someone assigned to you by a rules committee.</p><p>If every state at least made the assignment data easy to find, maybe we could do population analyses for more states and really understand the scope of this issue. If there are over 3 million people in California with assigned Senators until 2025, how many are there across the whole country?</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4bb18fa80832" width="1" height="1" alt=""><hr><p><a href="https://medium.com/greeninfo-network/a-tale-of-two-senates-4bb18fa80832">A Tale of Two Senates</a> was originally published in <a href="https://medium.com/greeninfo-network">GreenInfo Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>