<?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[Stories by Emma Sheridan on Medium]]></title>
        <description><![CDATA[Stories by Emma Sheridan on Medium]]></description>
        <link>https://medium.com/@sheridanemma?source=rss-5a1c659a8b94------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*hY8PDUqhpao7S0grsvboWw.png</url>
            <title>Stories by Emma Sheridan on Medium</title>
            <link>https://medium.com/@sheridanemma?source=rss-5a1c659a8b94------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 05:42:32 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@sheridanemma/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Facilitating effective ideation meetings]]></title>
            <link>https://uxdesign.cc/facilitating-effective-ideation-meetings-a5813713e559?source=rss-5a1c659a8b94------2</link>
            <guid isPermaLink="false">https://medium.com/p/a5813713e559</guid>
            <category><![CDATA[ux]]></category>
            <category><![CDATA[ideation]]></category>
            <category><![CDATA[design-thinking]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[design]]></category>
            <dc:creator><![CDATA[Emma Sheridan]]></dc:creator>
            <pubDate>Tue, 13 Jun 2023 00:34:46 GMT</pubDate>
            <atom:updated>2023-06-15T14:58:37.897Z</atom:updated>
            <content:encoded><![CDATA[<h4>Designing engaging and productive gatherings</h4><figure><img alt="Laptop open with blurred screen of 20 tiles of people on a video call. Plant with green leaves in focus on the right." src="https://cdn-images-1.medium.com/max/1024/0*nvyNH9X5Oqed5C4S" /><figcaption>Photo by <a href="https://unsplash.com/@sigmund?utm_source=medium&amp;utm_medium=referral">Sigmund</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>In today’s hybrid work landscape, <a href="https://www.microsoft.com/en-us/worklab/work-trend-index/great-expectations-making-hybrid-work-work">meetings have become abundant</a>, but unfortunately, many of them still suffer from inefficiency and ineffectiveness. Specifically, meetings aimed at generating ideas to address various challenges related to people, processes, or products encounter recurring issues. The lack of a clear goal in these meetings hinders active participation, and the organizer often dominates the conversation, resulting in a limited number of ideas that fail to fully solve the problem. Both the organizer and attendees are left feeling dissatisfied with the outcomes.</p><h3><strong>Intentional ideation</strong></h3><p>By approaching these pitfalls with intention and careful planning, the organizer or facilitator can overcome these issues. Here are some key aspects to aim for to host an effective ideation session:</p><ol><li><strong>Clear and well-defined goals:</strong> The meeting invitation plays a crucial role in setting expectations for participants (see <a href="https://www.designforbelonging.com/">Design for Belonging</a> to learn more about the moment of invitation). Clearly communicate the objective and purpose of the meeting, enabling attendees to prepare and engage effectively.</li><li><strong>Concise problem definition:</strong> The quality of ideas is enhanced when the group shares a comprehensive understanding of the problem. Encourage collaborative identification and documentation of pain points, allowing ideas to address the root issues more effectively. The <a href="https://www.atlassian.com/team-playbook/plays/5-whys">“5 Whys Exercise”</a> is one method to go deeper into the problem space.</li><li><strong>Cultivate psychological safety:</strong> Establish a safe environment that nurtures trust among participants. Actively design and foster psychological safety, enabling individuals to freely share their ideas and opinions. <a href="https://www.mural.co/blog/psychological-safety-for-imagination-work">This inclusive atmosphere is fundamental to unlocking creativity.</a></li><li><strong>Structured ideation:</strong> Instead of open-ended discussions starting with “What ideas do you have?”, leverage a guided ideation exercise that provides structure and constraints. Humans often struggle to ideate without boundaries, so offering a framework helps participants focus and generate more targeted ideas.</li><li><strong>Diverse communication modes:</strong> Expand beyond verbal group discussion to include other forms of communication such as writing, drawing, building, or individual and pair discussions. Embracing diverse communication modes encourages a wider range of ideas and perspectives, reducing biases and enhancing creativity.</li><li><strong>Emphasize idea quantity:</strong> Encourage the generation of a large quantity of ideas rather than settling for a limited number. <a href="https://www.forbes.com/sites/rodgerdeanduncan/2022/12/27/for-the-best-ideas-go-for-quantity-not-just-quality/?sh=28724ac54463">Quantity often leads to quality</a> as it allows for more exploration and diverse possibilities. Remember that the first idea proposed is likely not the best.</li><li><strong>Avoid idea evaluation:</strong> During ideation sessions, refrain from judging or evaluating ideas. The goal is to explore a wide range of possibilities without labeling any idea as “bad” or “wrong” until they are tested and proven.</li></ol><p>By striving for these principles, you can create a more effective and productive ideation session.</p><figure><img alt="Table with post-it notes and pens. Hands of people in the edges of the picture sharing notes." src="https://cdn-images-1.medium.com/max/1024/0*2OolRBzXLULyBwr-" /><figcaption>Photo by <a href="https://unsplash.com/de/@byfortytwo?utm_source=medium&amp;utm_medium=referral">FORTYTWO</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3><strong>Anyone can run an ideation session</strong></h3><p>To address the inefficiencies and frustrations associated with such meetings, I have developed a concise and powerful agenda leveraging design thinking methods and mindsets that can be customized for your upcoming ideation session.</p><p>The below sample agenda assumes that problem definition is clear. If that is not the case, hold a session prior to the ideation session to align on the problem. Tools such as interviewing, <a href="https://careerfoundry.com/en/blog/ux-design/affinity-map/">Affinity Mapping</a>, and developing <a href="https://www.nngroup.com/articles/user-need-statements/">User Need statements</a> and <a href="https://www.nngroup.com/articles/how-might-we-questions/">“How Might We”</a> questions can be useful in facilitating that discussion.</p><h4><strong>A sample agenda for ideation sessions</strong></h4><p><em>Estimated time needed: 45–60 minutes</em></p><ol><li><strong>Introduction &amp; ground rules (2 minutes)</strong></li></ol><ul><li>Share the agenda for the ideation session.</li><li>Review any <a href="https://www.ideou.com/blogs/inspiration/7-simple-rules-of-brainstorming">ground rules or guidelines</a> for the meeting.</li><li>Allow time for attendees to ask questions or seek clarification.</li></ul><p><strong>2. Warm-up exercise (5–10 minutes)</strong></p><ul><li>Conduct a warm-up exercise to foster creativity and build rapport among participants such as <a href="https://www.ideo.com/blog/build-your-creative-confidence-thirty-circles-exercise">30 Circles</a> or <a href="https://mbcollab.com/blog/six-creative-warmups-to-get-your-team-in-the-right-mindset">One Thing, Nine Ways</a>.</li><li>Choose an activity that aligns with the goals of the meeting and reflects the activities planned for the session. A quick Google search for “warm-up exercises design thinking” will return several potential activities.</li></ul><p><strong>3. Frame the problem (2 minutes)</strong></p><ul><li>Share a single artifact (slide, Word Doc, section of text in whiteboard tool) that serves as a summary of the problem, providing participants with a reference point to anchor their thinking and revisit as needed throughout the session.</li></ul><p><strong>4. Guided ideation &amp; dot voting (30 minutes)</strong></p><ul><li>Select 1–2 guided ideation exercises such as <a href="https://www.iamnotmypixels.com/how-to-use-crazy-8s-to-generate-design-ideas/">Crazy 8s</a>, <a href="https://www.mural.co/blog/remote-ideation-techniques">Mash-up</a>, <a href="https://www.debonogroup.com/services/core-programs/six-thinking-hats/">Six Thinking Hats</a>. A quick Google search for “ideation techniques design thinking” will return tons of potential exercises.</li><li>Provide clear instructions.</li><li>Use <a href="https://lucidspark.com/blog/dot-voting">Dot Voting</a> and/or <a href="https://www.ibm.com/design/thinking/page/toolkit/activity/prioritization">Prioritization Matrix</a> to select the most promising ideas.</li></ul><p><strong>5. Next steps &amp; closing remarks (5 minutes)</strong></p><ul><li>Assign owners or champions for the selected ideas who will be responsible for driving their implementation (if not already known).</li><li>Summarize the key decisions made and actions to be taken.</li><li>Clarify any follow-up tasks or assignments.</li><li>Express gratitude for participants’ contributions and conclude the meeting on a positive note.</li></ul><h3><strong>The facilitator’s most important work happens outside the session</strong></h3><p>To facilitate a successful ideation session, proper preparation is key for the meeting organizer. This includes planning the agenda (always set aside more time than you think), setting up the workspace (in-person or virtually using whiteboard tools such as <a href="https://www.figma.com/figjam/">FigJam</a>, <a href="https://miro.com/">Miro</a>, and <a href="https://www.mural.co/">Mural</a>), and ensuring the participants are ready (meeting invite, pre-work). Dedicate sufficient time for preparation as a facilitator so that you can fully focus on the session and engage with the participants when the time comes. For more guidance on organizing workshops, I highly recommend reading <a href="https://www.amazon.com/Workshop-Survival-Guide-educational-workshops/dp/1071344374"><em>The Workshop Survival Guide</em></a>.</p><p>After a productive ideation session, it’s important to show appreciation to the attendees for their time and ideas. Consider expressing gratitude and highlighting the impressive number of ideas generated by the group, which often comes as a pleasant surprise.</p><h3><strong>Ideation sessions do more than simply generate ideas</strong></h3><p>While the obvious goal of an ideation session is to generate ideas to solve the problem, this particular way of work also generates excitement and camaraderie amongst the team. Participants will leave the meeting feeling energized about what’s to come rather than questioning the reason for gathering in the first place.</p><p>For more on the importance of ideas and ideation, I recommend reading <a href="https://www.ideaflow.design/"><em>Ideaflow</em></a> by Jeremy Utley and Peter Klebahn from Stanford’s d.school who expound on the ideation practice moving beyond a single workshop or event. How does your team support ideation?</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a5813713e559" width="1" height="1" alt=""><hr><p><a href="https://uxdesign.cc/facilitating-effective-ideation-meetings-a5813713e559">Facilitating effective ideation meetings</a> was originally published in <a href="https://uxdesign.cc/">UX Collective</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Remedy PM loneliness by investing in PM-to-PM relationships]]></title>
            <link>https://medium.com/design-bootcamp/remedy-pm-loneliness-by-investing-in-pm-to-pm-relationships-5482ce6c5538?source=rss-5a1c659a8b94------2</link>
            <guid isPermaLink="false">https://medium.com/p/5482ce6c5538</guid>
            <category><![CDATA[product]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[teams-and-teamwork]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[collaboration]]></category>
            <dc:creator><![CDATA[Emma Sheridan]]></dc:creator>
            <pubDate>Tue, 24 Jan 2023 05:32:28 GMT</pubDate>
            <atom:updated>2023-01-24T08:36:31.903Z</atom:updated>
            <content:encoded><![CDATA[<h4>Why you should try co-PMing</h4><p>Product articles, podcasts, and newsletters are flooded with mindsets and tools to approach building a strong product triad or trio between designers, engineers, and product managers (PM).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*CIsMWRuqPghlZIlf" /><figcaption>Product trio model, diagram from <a href="https://uxcam.com/blog/modern-mobile-product-team-structure/">UXCam</a>.</figcaption></figure><p>With a quick Google search, you’ll find plenty of opinions from thought leaders in industry on how to strengthen those cross-functional relationships to deliver great products. Don’t get it twisted, I preach this working trio as well, but what about PM-to-PM relationships?</p><p><a href="https://twitter.com/shreyas">Shreyas Doshi</a>, a prevalent, respected voice in the product industry, <a href="https://twitter.com/shreyas/status/1270564761478946817">composed a tweet </a>in June 2020: “A universal aspect of Product Management that people don’t expect is just how lonely it feels to be a PM. You could be working with a team of 100 people, be in 10 meetings a day, and yet often feel super-lonely because no one else shares your struggles or truly understands them.” Plenty of people resonated and responded to this tweet (and <a href="https://productlife.to/p/-pms-need-friends">others have written on this topic of PM loneliness</a> too).</p><p>Shreyas <a href="https://twitter.com/shreyas/status/1270567509851140097">responded to commeters</a> with some suggestions to address this issue mentioning that “peer connection among PMs can do wonders” though “PMs are often so busy that they don’t find time to do this or otherwise feel guilty about spending their time on things other than project work”.</p><p>In my experience thus far, these feelings of loneliness stem from PMs lacking collaboration with other PMs. At large companies, generally, PMs are granted ownership over a particular product area and soon become the subject matter expert (SME) in their space. While this singular ownership gives them autonomy over their decisions, it does not foster collaboration and learning amongst their immediate or greater team. Additionally, insufficient relationships and poor collaboration amongst PMs fuels a competitive environment and directly influences planning, development of product sense, and the understanding of how an individual’s work relates to the greater organization’s (and in turn, company’s) vision.</p><p>I’ve been fortunate to have had managers who allowed and encouraged me and my PM teammate Dina to co-PM (jointly lead) projects, and after reflecting on my experience, I see co-PMing as an antidote to PM loneliness.</p><h3><strong>Benefits of co-PMing</strong></h3><p>I encourage this ownership approach for at least one of your projects to experience the considerable benefits for each individual PM and for the greater PM team. Below are the key benefits of co-PMing that I’ve gathered through participating in this partnership:</p><ol><li><strong>No single subject matter expert (SME)</strong>: As mentioned prior, traditionally, at large companies, PMs tend to have ownership over a particular product area and thus naturally become the SME in this space. With co-PMing, there is no single SME: there are two. If one person is out of office, there is a back-up with the same information who can move along the conversation.</li><li><strong>Two heads are better than one: </strong>This set-up allows you to take advantage of each other’s strengths and complement each other’s weaknesses. Additionally, having diversity of perspectives (particularly in terms of personality, values, and abilities, also known as <a href="https://hbr.org/2017/06/does-diversity-actually-increase-creativity">deep-level diversity</a>) is important for idea generation.</li><li><strong>You’re not alone</strong>: Having a partner to bear responsibilities with you for the project greatly relieves stress. There’s no longer the potential dooming feeling that it is all on you. It’s also so nice to have a confidant to vent with when things get tough or to gut check with when you’re unsure about something. This <a href="https://hbr.org/2022/10/the-power-of-work-friends">additional support system (a.k.a. a friend) contributes to improved work satisfaction</a>.</li><li><strong>Fosters learning and community building</strong>: While working on projects with Dina, we’ve naturally pushed each other to continue learning as PMs through (knowledge) sharing of conversations, experiences, and resources. Since there are two of us, each of whom operates in slightly different networks due to our prior projects and interests, we’ve both gotten exposure to work happening outside of our individual projects with this extended reach furthering community building.</li><li><strong>It’s more fun</strong>: Simply put, co-PMing is enjoyable. Together, Dina and I have shared plenty of laughs and genuine engagement with our work fueling each other throughout the duration of the projects. As Howard Schultz, former CEO and current interim CEO of Starbucks shares, “Success feels better when it’s shared.” CoPMing is a great example of that belief. Further, a powerful co-PM relationship carries energy that is contagious, potentially boosting morale for the rest of the PM team along the way.</li></ol><h3><strong>Foundations for successful co-PMing</strong></h3><p>While the benefits of co-PMing speak for themselves, there is still work to ensure the arrangement is mutually beneficial and successful. Consider these things to build and maintain a co-PM relationship:</p><ol><li><strong>Clear understanding of responsibilities</strong>: Co-PMing does not mean that you and your partner have to take 50/50 ownership over the work, and it likely wouldn’t turn out that way even if you tried. Throughout the course of the project, responsibility will eb and flow. As long as you and your co-PM are in communication and have alignment over which tasks are assigned to who, this fluctuating responsibility won’t be an issue. In fact, the shifting levels of responsibility may even be helpful if one of you is facing demanding work outside of this project, you can work together to load balance. A manager can also help establish your responsibilities from the project kick-off by assigning initial tasks to each of you.</li><li><strong>Communication is key</strong>: Keep your co-PM informed with new information you may have learned in their absence. Additionally, together, keep your manager informed on how the project is progressing including who is tackling which tasks, and how the working relationship is going. Ensure both of you are included on all conversation threads and documented as co-leaders. If you find that one of you is left out, add them in promptly and let the thread participants know you are both working on this project.</li><li><strong>Meet often</strong>: Engage with your co-PM several times a week in various online and offline channels: chat, email, (video) calls, comments in documents and presentations, or Figjam. Dina and I have scheduled daily syncs (and, we welcome unplanned calls as well). While we often both come to our syncs with a few prepared thoughts or questions, not all of our syncs come with an agenda, and honestly, some of those conversations are our best collaboration sessions. (There’s something to be said for <a href="https://www.washingtonpost.com/wellness/2022/04/04/do-nothing-mind-wander-how/">wandering minds</a>.) Additionally, we use offline forms of communication to share relevant information or any questions that come up on our own time, not waiting for our next sync unless we find the topic better to discuss that way.</li><li><strong>Trust your counterpart</strong>: Co-PMing is absolutely a collaborative relationship where competition has no place. There should be mutual investment, excitement, and interest in the work you are doing together which will help lay the foundation for a trusting partnership.</li><li><strong>Disagree and commit</strong>: Trusting your co-PM also means you are comfortable disagreeing with them. Just because you are working on the same project does not mean you have to agree with all of each other’s ideas; in fact, it is probably better if you do not always agree. Some of Dina and I’s greatest learning experiences came when we disagreed over how to move forward or about a framework we were exploring. Invite this deep collaboration: <a href="https://www.inc.com/jeff-haden/jeff-bezos-uses-disagree-commit-rule-to-overcome-an-uncomfortable-truth-about-teamwork.html">disagree and commit</a> to move forward and to avoid stalling decision making.</li></ol><p>While the concept of co-PMing or like-concepts are not new (see <a href="https://medium.com/@arielyoffie/2-styles-of-product-management-pair-pming-vs-swim-lanes-5ee7e6f4623b">pair PMing</a>, <a href="https://www.codementor.io/pair-programming">pair programming</a>, <a href="https://hbr.org/2012/04/leadership-teams-why-two-are-b">two-in-a-box leadership</a>, <a href="https://techcrunch.com/2021/05/24/2-ceos-are-better-than-1/?guccounter=1&amp;guce_referrer=aHR0cHM6Ly93d3cuZ29vZ2xlLmNvbS8&amp;guce_referrer_sig=AQAAAAxKwrHUeIxUNrCWhcA8yjCf64A--f5VRmEPNafyTpn4cdbyoGxUGiyeN8O8jMJNdhwvnRI42Wc_bRA_InvXfdFPrnDTmO2PbfqFgF98cpoYmoZec7fvFlRy55fQQhqjZTpJxcDvJDMzaabAnr-eq02Q3liL9ZCs5ZMnslgqwRFe">co-CEOs</a>), it is underutilized in product and not touted as a remedy for PM loneliness. Co-PMing is not only a great opportunity to build genuine relationships amongst your peers that will stick with you beyond a single project, it also accelerates individual learning by consistently expanding your thinking — give it a try.</p><p>How does your PM team foster collaboration amongst PMs?</p><p><em>First posted on </em><a href="https://navigatingproduct.substack.com/p/remedy-pm-loneliness-by-investing?utm_source=profile&amp;utm_medium=reader2">Navigating Product</a><em> on Substack.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5482ce6c5538" width="1" height="1" alt=""><hr><p><a href="https://medium.com/design-bootcamp/remedy-pm-loneliness-by-investing-in-pm-to-pm-relationships-5482ce6c5538">Remedy PM loneliness by investing in PM-to-PM relationships</a> was originally published in <a href="https://medium.com/design-bootcamp">Bootcamp</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Practicing design as a (non-)traditional designer]]></title>
            <link>https://medium.com/design-bootcamp/practicing-design-as-a-non-traditional-designer-f55c3fd85b6a?source=rss-5a1c659a8b94------2</link>
            <guid isPermaLink="false">https://medium.com/p/f55c3fd85b6a</guid>
            <category><![CDATA[work]]></category>
            <category><![CDATA[design]]></category>
            <category><![CDATA[design-thinking]]></category>
            <category><![CDATA[workshop]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Emma Sheridan]]></dc:creator>
            <pubDate>Sun, 23 Oct 2022 00:21:22 GMT</pubDate>
            <atom:updated>2022-10-24T08:04:43.082Z</atom:updated>
            <content:encoded><![CDATA[<h4><em>And bringing your team along for the journey</em></h4><figure><img alt="Mountain range, clear blue skies" src="https://cdn-images-1.medium.com/max/1024/1*BrCRTsRU_VAthoRX4brXGg.jpeg" /><figcaption>Mt. Rainer, Washington.</figcaption></figure><p>Everyone practices design. At some point, you’ve likely organized a birthday celebration, planned a vacation, chosen what to wear for the day, or decided what mode of transportation to take to work. In doing all of these things, you have considered the people involved, the needs of those included, and any constraints there might be such as food allergies, flight times, weather, or gas prices. You then came up with an idea and tested it out. You picked a cute but impractical pair of new shoes to wear to the office. While you looked great in the shoes, you realized it was not worth sacrificing comfort on a day where you had lots of meetings that required walking long distances. The next time you head into the office you recall your previous experience and iterate on it, this time choosing a stylish pair of sneakers. Whether you categorize yourself as one or not, you are a designer.</p><p>By title, I am a Product Manager (PM): an elusive, ever-changing, just-make-it-happen profession. Because of my position as a PM, working across disciplines, I sit in a place to push for lowercase d design as a method to get shit done. However, this is not easy. While I do believe companies have come a long way towards being design-led, many individuals still don’t understand how to leverage the power of design. To be clear, this is of no fault to them. Design is experiential. In order to believe its potential, you must experience it yourself.</p><h3><strong>Experiencing design through a workshop</strong></h3><p>I hosted an “Intro to design methods &amp; mindsets” remote workshop for my team, inspired by the <a href="https://dschool.stanford.edu/resources/dschool-starter-kit">d.school’s Starter Kit</a>. I invited engineers and PMs, and provided minimal information about what was going to happen during the 2-hour session besides it being a high-energy, interactive, and collaborative learning opportunity. During the 2 hours, my teammates worked with a partner going through the entire design thinking process (empathy, define, ideate, test, iterate) designing a solution for a low-point in each other’s yesterdays. The hours flew by as it usually does when working deeply with design. After the session, I discussed with individuals how the workshop went for them and received a flurry of positive reactions: “very engaging and fun workshop”, “learned a lot and get actionable tools to use with customers”, “so fun and learned a LOT!”, “learned a lot and look forward to applying these concepts, right away!”, “Loved the refreshing change in pace from other responsibilities to spend some time doing something new!”.</p><figure><img alt="team sits around a large table with laptops, one person is at the front of the room at a whiteboard pointing at sticky notes" src="https://cdn-images-1.medium.com/max/1024/0*YHUPN5M13I_dTZ_K" /><figcaption>Photo by <a href="https://unsplash.com/@jasongoodman_youxventures?utm_source=medium&amp;utm_medium=referral">Jason Goodman</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>The response was encouraging, though there were a few reactions that surprised me. First, many saw the workshop as an opportunity to get to know and interact with their team members. While most companies are now remote-friendly, there is still a concern for feelings of isolation amongst a team spread out over time zones and locations. Though the workshop focused on empathy as part of the design process, the empathy went beyond building a solution for your partner and to actually deepening a connection with a teammate.</p><p>Second, most participants did not know what they were signing up for when they received the invite for a design methods and mindsets workshop. Though I was vague in my invitation (intentionally), this response reminded me that while terms like human-centered design and design thinking get thrown around a lot in the tech industry and beyond, it doesn’t mean that people understand how they work in practice. When asked what comes to mind when you hear the word design, most responses were related to aesthetics, what something looks like.</p><p>And lastly, nearly all participants were uncomfortable when asked to draw. I saw many deer-in-the-headlights responses when I instructed the group to do a <a href="https://www.artsy.net/article/artsy-editorial-blind-contour-drawing-help-better-artist">blind-contour drawing</a> exercise as a warm-up activity, along with many comments of “I can’t draw” or “I’m not good at drawing” or general embarrassment over what may be produced. As outlined in <a href="https://uxdesign.cc/there-is-always-a-good-reason-to-draw-why-sketching-benefits-your-design-practice-fd0c7e43c2c8">“There is always a good reason to draw”</a> by Marjorie Broudieu and in many other articles, drawing is a highly valuable tool for your toolbox to communicate and express ideas. Yet, most adults are hesitant to use it.</p><figure><img alt="Sheet of paper with wireframe drawings on it, hands on the edge of the photo, pen in one hand" src="https://cdn-images-1.medium.com/max/1024/0*Z7lux-Ig3ZCv7Emq" /><figcaption>Photo by <a href="https://unsplash.com/@amayli?utm_source=medium&amp;utm_medium=referral">Amélie Mourichon</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h3><strong>Being a (non-)traditional designer</strong></h3><p>While I’m drafting how to continue to expose my team members to design practices, I’ve been reflecting on my allegiance to design as a (non-)traditional designer meaning I did not receive a degree in Design or in any of its flavors.<strong> </strong>More than ever, I see design as a method and mindset to be utilized when appropriate. Determining the right time to call on design and which tools would be best to use is a skill I’m continuously honing. I’m building my toolkit (leveraging existing open-source resources from great designers and organizations and learning from my fellow designers) and sharing it with others along the way.</p><p>And, I’m still learning: still learning how to push for being design-led, still learning how to lead with design, and still learning how to be a designer. It’s exciting. There’s so much opportunity to shift our thinking, and to continuously re-think: re-think what it means to design, what it means to be a designer, what it means to use design.</p><p>If you find yourself in a similar position to me as a (non-)traditional designer pushing for infusing design into the product lifecycle or any other work you may be taking on, I urge you to lean into the challenge. And, it is a challenge, and not everyone will be supportive (at first).</p><figure><img alt="Skyline, mountains in the background, Space Needle in the left foreground." src="https://cdn-images-1.medium.com/max/1024/1*cPIN-IrDR0Z1p7J1aU-WDQ.jpeg" /><figcaption>Space Needle, Seattle, Washington.</figcaption></figure><h3><strong>How to lean in</strong></h3><p>As I’ve been navigating these challenges more seriously over the past year, I’ve collected a few tangible action items that can help you along your journey to push for design:</p><ul><li>Curate a community that also thinks the way you do, that sees the value in design. They may be outside your immediate team, organization, or company.</li><li>Conduct research amongst your own team. Host mini-interviews or coffee chats with various stakeholders from different disciplines to understand how they view the value and practice of design within their own teams and where they see friction in the design process.</li><li>Use design tools in your own work (make it a daily practice, if possible) to build your success stories so you can call on them when being asked to speak to the power of design.</li><li>To scale and influence, it is important to lean on the skills of others instead of doing it all yourself. Continue to educate yourself in the different disciplines you work with to better understand how they can help you on your journey in elevating design, and even more importantly, how you can help them with theirs (whatever that journey may be).</li></ul><p>I accept and practice the belief that “the idea of design and the profession of the designer has to be transformed from the notion of a specialist function into a generally valid attitude of resourcefulness and inventiveness which allows projects to be seen not in isolation but in relationship with the need of the individual and the community” as <a href="https://moholy-nagy.org/biography/">László Moholy-Nagy, Hungarian painter, photographer, and professor at the Bauhaus school</a>, argues in <em>Vision in Motion</em>. If you can get past the (self-)established barriers of design professionalism and embrace design as an attitude, the opportunities for impact are endless.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f55c3fd85b6a" width="1" height="1" alt=""><hr><p><a href="https://medium.com/design-bootcamp/practicing-design-as-a-non-traditional-designer-f55c3fd85b6a">Practicing design as a (non-)traditional designer</a> was originally published in <a href="https://medium.com/design-bootcamp">Bootcamp</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[7 books that made me think differently about design, that weren’t about design]]></title>
            <link>https://sheridanemma.medium.com/7-books-that-made-me-think-differently-about-design-that-werent-about-design-3afc73564f17?source=rss-5a1c659a8b94------2</link>
            <guid isPermaLink="false">https://medium.com/p/3afc73564f17</guid>
            <category><![CDATA[books]]></category>
            <category><![CDATA[reading]]></category>
            <category><![CDATA[design-thinking]]></category>
            <category><![CDATA[design]]></category>
            <category><![CDATA[book-recommendations]]></category>
            <dc:creator><![CDATA[Emma Sheridan]]></dc:creator>
            <pubDate>Mon, 20 Sep 2021 03:37:03 GMT</pubDate>
            <atom:updated>2021-09-20T03:37:03.720Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*69jHNPm17CA-MugE" /><figcaption>Photo by <a href="https://unsplash.com/@tomhermans?utm_source=medium&amp;utm_medium=referral">Tom Hermans</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Do you also love the feeling that you get when reading and you make a connection to something else you read, saw or heard? It’s kind of like an Aha! or lightbulb moment, but not quite. It appears that there’s actually science behind that — schema theory. I’m not going to go into it but you can get the gist from <a href="https://en.wikipedia.org/wiki/Schema_(psychology)">this Wikipedia page</a>.</p><p>These moments remind me of why I read. I’ve often thought about keeping an on-going list of the threads that run throughout the ranging books and articles I’ve read just to document the growing web of connections. This is also why I like to read fiction and non-fiction (ideally in equal amounts).</p><p>Since I haven’t started this list of links, I had to go back through my bookshelves to remind myself of books I’ve read that made me think differently about design, that weren’t about design. Maybe you’ll find a book that looks interesting to you or maybe you won’t, but the point (once again) is that design is everywhere. It may be easier for some to point out than others, but it’s there. Design is saying hey, hoping one day enough people will hear them. Don’t forget we are <a href="https://sheridanemma.medium.com/we-live-in-a-world-where-nearly-everything-has-been-designed-and-engineered-to-accommodate-and-even-7991c29b3964">Living in a Designed World</a>.</p><p>Before I get started, I will say a few of these books may border the Design genre, but hasn’t that delineation already been blurred? I hope after reviewing this list you’ll remember to keep your eye out for Design.</p><h3>1. The Color of Law: A Forgotten History of How Our Government Segregated America</h3><p><em>By: Richard Rothstein</em></p><figure><img alt="Book cover for The Color of Law." src="https://cdn-images-1.medium.com/max/600/1*4BbPvJzSZaIWOCgSzUHGYw.jpeg" /></figure><h3>2. How to Do Nothing: Resisting the Attention Economy</h3><p><em>By: Jenny Odell</em></p><figure><img alt="Book cover for How to Do Nothing." src="https://cdn-images-1.medium.com/max/1024/1*XE5NYJvmzibTY6pnfK3fkA.jpeg" /></figure><h3>3. The Whale and the Reactor: A Search for Limits in an Age of High Technology</h3><p><em>By: Langdon Winner</em></p><figure><img alt="Book cover for The Whale and the Reactor." src="https://cdn-images-1.medium.com/max/323/1*ApckeMyQSXAHw3U70Za1YQ.jpeg" /></figure><h3>4. The Architecture of Happiness</h3><p><em>By: Alain de Botton</em></p><figure><img alt="Book cover for The Architecture of Happiness." src="https://cdn-images-1.medium.com/max/1024/1*5IXO7W2NgwaazT6TUGnROw.jpeg" /></figure><h3>5. Flow: The Psychology of Optimal Experience</h3><p><em>By: Mihaly Csikszentmihalyi</em></p><figure><img alt="Book cover for Flow." src="https://cdn-images-1.medium.com/max/1024/1*tdMO72H4RYgr6KH6ovFJvA.jpeg" /></figure><h3>6. Braiding Sweetgrass</h3><p><em>By: Robin Wall Kimmerer</em></p><figure><img alt="Book cover for Braiding Sweetgrass." src="https://cdn-images-1.medium.com/max/647/1*QP0wMV6rYPu1vvU3FVWqCA.jpeg" /></figure><h3>7. Reclaiming Conversation: The Power of Talk in a Digital Age</h3><p><em>By: Sherry Turkle</em></p><figure><img alt="Book cover for Reclaiming Conversation." src="https://cdn-images-1.medium.com/max/1024/1*S7qaSKlPmtXMFJtMxg_lDg.jpeg" /></figure><p>I’d love to know if you’ve had a similar experience with other books! Always looking to add books to my shelf, especially design-related. Also, if you’re interested in purchasing any of these books, don’t forget to support your local independent bookstores.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3afc73564f17" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The case for hybrid insights]]></title>
            <link>https://uxdesign.cc/the-case-for-hybrid-insights-76d7c0f5637c?source=rss-5a1c659a8b94------2</link>
            <guid isPermaLink="false">https://medium.com/p/76d7c0f5637c</guid>
            <category><![CDATA[ux-research]]></category>
            <category><![CDATA[big-data]]></category>
            <category><![CDATA[design-thinking]]></category>
            <category><![CDATA[empathy]]></category>
            <category><![CDATA[design]]></category>
            <dc:creator><![CDATA[Emma Sheridan]]></dc:creator>
            <pubDate>Tue, 22 Jun 2021 21:19:19 GMT</pubDate>
            <atom:updated>2021-06-30T10:30:50.787Z</atom:updated>
            <content:encoded><![CDATA[<h4>A friendly reminder to maintain a healthy balance of narratives and numbers</h4><figure><img alt="A black and white photograph of a whiteboard with post-it notes on it." src="https://cdn-images-1.medium.com/max/1024/1*NlCX-VS3zPOFYzUXoZwenQ.jpeg" /><figcaption>Photo by Noah Trevino.</figcaption></figure><p>I first came across the term “hybrid insights” in the book <a href="https://www.creativeconfidence.com/book/"><em>Creative Confidence: Unleashing the Creative Potential Within Us All</em></a> by David M. Kelley and Tom Kelley, IDEO founder and IDEO partner, respectively. It was simply an excerpt that explained this newer (for the time, the book was published in 2013) approach to design research. The passage was titled “Hybrid Insights: Empathy in a World of Big Data” and explains hybrid insights as “an approach that integrates quantitative research into human-centered design. Hybrid insights allow us to embed stories in the data, bringing the data to life.”</p><p>Upon reading about the blurb, I felt immediately attracted to the method since I’ve always been intrigued by the use of data and by design thinking. I grabbed the phrased and added it to my LinkedIn bio (and it’s still there). Since first reading about hybrid insights a few years ago I have had the time to experience it first hand and to learn more about it. And the case for hybrid insights has taken off.</p><p>Design consultancies, such as <a href="https://new-ideo-com.s3.amazonaws.com/assets/files/pdfs/news/pp_56-61_HybridInsights_RotFall12_spreads.pdf">IDEO</a> and <a href="https://www.frogdesign.com/designmind/hybrid-research-inspires-impactful-outcomes">frog</a>, and design-conscious companies, like <a href="https://spotify.design/article/simultaneous-triangulation-mixing-user-research-and-data-science-methods">Spotify</a>, have made their stance on the importance of hybrid insights, the blend of design research and data science, which is also known as <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5602001/">mixed methods research</a> or even <a href="https://www.nngroup.com/articles/triangulation-better-research-results-using-multiple-ux-methods/">triangulation</a>. Yet, incorporating both research methods in the design process is easily forgotten or is overlooked. And it happens both ways: there can be an over-emphasis on big data without speaking to the customer or customer interviews can take center stage without using data to see the big picture or to find the extremes in the audience.</p><p>It’s easy to get caught up in one or the other, the narratives or the numbers. I’ve done so myself. I’ve started with the numbers, formulated a plan and proposed a solution, and then remembered, “I haven’t actually talked to a customer about this.” At that point, it wasn’t too late to incorporate customer feedback, but it slowed down the process, and I had to reevaluate after having chosen a course. It would’ve been better to have both the numbers and the stories from the get-go.</p><p>I’ve found that hybrid insights often gets left out of traditional design thinking methodologies and toolkits, but it’s so important, especially in the data-driven environment today. So, in order to keep hybrid insights front-of-mind throughout the design research process, here are some tips. These pointers have been touted by the companies I mentioned previously, and I also threw in my two cents.</p><h3>1. Create initial research questions that can be addressed both quantitatively and qualitatively.</h3><p>Start your design research with clear objectives so you can identify how you can use hybrid insights to your advantage. I like using a flexible template to jumpstart your planning for the design research process. There are several example templates that people have proposed such the <a href="https://www.usertesting.com/blog/ux-research-objectives">flow diagram from UserTesting.com</a> below which prompts researchers to think about how each objective will be addressed (keeping tip #2 in mind).</p><figure><img alt="A flow diagram that starts with a feature, then objective, then task." src="https://cdn-images-1.medium.com/max/315/1*EHofdtxRvnTZW4AI3kWJjQ@2x.jpeg" /><figcaption>A template to inspire your design research plan from UserTesting.com.</figcaption></figure><h3>2. Use a variety of methods to address these foundational research questions.</h3><p>In order to answer research questions in a hybrid manner, you need to use several methods to probe the situation. Spotify has a What-Why Framework that they adapted from a <a href="https://www.nngroup.com/articles/which-ux-research-methods/">Nielsen Norman Group article by Christian Rohrer</a> that breaks down potential methods depending on your needs. Be sure to pick from different quadrants.</p><figure><img alt="A coordinate plane with behavioral/attitudinal on the y-axis and quantitative/qualitative on the x-axis. Points on the grid are choices for design research methods." src="https://cdn-images-1.medium.com/max/1024/1*V_JcnykeaFDEZNZal5YUZA.jpeg" /><figcaption>What-Why Framework from Spotify for chosen mixed design research methods.</figcaption></figure><p>Additionally, you want to carry out the chosen mixed methods at the same time. (This is where I’ve previously messed up.) It’s important to wait to review the results until both methods have been carried out so that your reactions to one don’t influence your response to the other.</p><h3>3. Remember why the hybrid approach is so good.</h3><p>In the words of <a href="https://new-ideo-com.s3.amazonaws.com/assets/files/pdfs/news/pp_56-61_HybridInsights_RotFall12_spreads.pdf">Johannes Seemann at IDEO</a>, “Keeping the soul in data is the great strength of the hybrid insights approach.” And vice versa, applying the data to the narratives is another great strength of the hybrid insights approach. It probably won’t come as a shock (especially if you already try to apply mixed methods to your design research), but you can uncover more meaningful and more accurate insights this way. Often the data and the stories don’t match, so hybrid insights give you the opportunity to pry at and to understand why that is. Or in the other case, the data and the stories match, and you can be more confident you are on the right track.</p><figure><img alt="Black and white image of computers facing each other on a desk with pens and markers in the middle of the desk." src="https://cdn-images-1.medium.com/max/1024/1*ZRgfivTRlAsKYG1wLc1Jzg.jpeg" /><figcaption>Photo by Noah Trevino.</figcaption></figure><p>Depending on the responsibilities of your role and how your team is set up, hybrid insights is also a great opportunity for cross-team collaboration. Don’t be afraid to lean on those around you to fill in the pieces required to gather hybrid insights. You never know what you’ll find together!</p><p>How do you utilize hybrid insights?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/94/0*0Py8zUEMIqob1cjg" /><figcaption>The UX Collective donates US$1 for each article we publish. This story contributed to <a href="https://www.wcd.school/">World-Class Designer School</a>: a college-level, tuition-free design school focused on preparing young and talented African designers for the local and international digital product market. Build the design community you believe in.</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=76d7c0f5637c" width="1" height="1" alt=""><hr><p><a href="https://uxdesign.cc/the-case-for-hybrid-insights-76d7c0f5637c">The case for hybrid insights</a> was originally published in <a href="https://uxdesign.cc/">UX Collective</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[NFTs: The promise for authenticity]]></title>
            <link>https://sheridanemma.medium.com/nfts-the-promise-for-authenticity-6d516d81ed5a?source=rss-5a1c659a8b94------2</link>
            <guid isPermaLink="false">https://medium.com/p/6d516d81ed5a</guid>
            <category><![CDATA[authenticity]]></category>
            <category><![CDATA[cryptocurrency]]></category>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[nft]]></category>
            <category><![CDATA[art]]></category>
            <dc:creator><![CDATA[Emma Sheridan]]></dc:creator>
            <pubDate>Fri, 14 May 2021 23:15:09 GMT</pubDate>
            <atom:updated>2021-06-24T04:42:03.976Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="Gold picture frame with gold dollars signs falling from the top to bottom of the frame." src="https://cdn-images-1.medium.com/max/1024/1*LVD0V6MxIX8oytbiS7iXcA.gif" /><figcaption>The first NFT “Quantum” by Kevin McCoy and Anil Dash.</figcaption></figure><p>If you have been keeping up with the technology world recently, it is likely that you have heard the acronym NFT floating around. NFTs or non-fungible tokens are the latest hot commodity in the blockchain world. Many of the headlines are related to digital art and NFTs. Recently, artist Beeple’s digital collage “Everydays: The First 5000 Days” sold for a <a href="http://decrypt.co/62898/the-10-most-expensive-nfts-ever-sold">staggering $69 million</a>. Jack Dorsey, co-founder and CEO of Twitter, sold his first tweet as an NFT for $2.9 million. American DJ 3LAU collaborated with digital artist Slimesunday to produce an NFT that includes a music video and dance track that the buyer can choose, selling for $1.33 million. Other artists have been making significant compensation for their work as well and new uses for NFTs are being explored constantly. It is a relatively new technology that has not been widely investigated philosophically by scholars yet. I am diving into NFT technology to raise interesting and enduring philosophical issues surrounding authenticity.</p><h3><strong>The brief history of NFTs</strong></h3><p><a href="http://www.theverge.com/2021/4/2/22364240/nft-blockchain-artist-hackathon-kevin-mccoy-ani%20l-dash">The first NFT</a>, “Quantum,” was a piece of digital art created in 2014 by Kevin McCoy, a prominent media artist, in partnership with Anil Dash, now CEO of Glitch, a small technology company. A few years after “Quantum”, in 2017, the rise of <a href="https://ethereum.org/en/">Ethereum</a>, a blockchain platform that powers the second-largest cryptocurrency ether, led to Cryptopunks, the first NFT on Ethereum used for trading unique cartoon characters. During this time, CryptoKitties also surfaced as a way to adopt and trade virtual cats using NFTs. Soon NFTs caught on for more mainstream businesses such as Nike. In 2019, Nike patented CryptoKicks using NFTs to verify collectors’ shoes. In 2020, the NBA established TopShot in partnership with Dapper Labs, a blockchain company, to host an online marketplace for virtual player trading cards and other media such as highlight clips using NFTs to provide authenticity. Today, NBA top-shot has 800,000 accounts and is leading the <a href="http://www.theverge.com/22348858/nba-nft-top-shot-dapper-labs">NFT market with $500 million in sales</a>. To understand the recent craze over NFTs, we can look at sales data from NonFungible.com that shows how NFT sales peaked during the first three months of 2021. Transactions have slowed since the end of March 2021 in which the month saw almost $150 million in sales. While purchase levels <a href="http://www.reuters.com/technology/after-first-quarter-frenzy-nft-market-shows-signs-stabilisin%20g-2021-05-04/">seem to be on the decline</a>, sales have remained higher than pre-pandemic levels. The technology does not seem to be fading away anytime soon.</p><figure><img alt="Digital art collage of 5000 daily artworks." src="https://cdn-images-1.medium.com/max/1024/1*deHLCtgUc-8zvdLSfuK4gA.jpeg" /><figcaption>Beeple’s “Everydays” digital collage NFT that sold for $69 million.</figcaption></figure><h3><strong>What is an NFT?</strong></h3><p>It is important to understand the technology behind an NFT as its features lend itself to the philosophical questions regarding authenticity I am addressing shortly. An NFT or non-fungible token is a unique digital asset with certifiable ownership that lives on a blockchain; hence, the descriptor <em>non-fungible</em> meaning the asset cannot be replaced with something else–it is one-of-a-kind. <a href="http://www.theverge.com/22310188/nft-explainer-what-is-blockchain-crypto-art-faq">For those unfamiliar with blockchain technology</a>, in brief, blockchain is a data structure for storing transactional records otherwise known as “blocks” on a “chain” of public databases. These blocks are connected in a network of adjacent nodes which is called the “digital ledger”. The intrigue of blockchain technology, and consequently NFTs, is its ability to be transparent in recording changes to the ledger thus creating greater trust in transactions. NFTs exist as assets operating on blockchain technology. Most NFTs are part of Ethereum blockchain, but exist on other ledgers as well. Ether, the cryptocurrency used on Ethereum, gives individuals the opportunity to bid on and buy NFTs. If an NFT is bought on Ethereum, Ethereum will hold a public record of that individual’s ownership of that NFT. Additionally, anyone can “mint” an NFT, which is the process for making a digital asset part of the Ethereum blockchain (or another blockchain of choice), through online services. After minting an NFT, the new digital asset is stored in a digital wallet. The wallet can then be linked to an individual’s account for digital marketplaces such as <a href="https://opensea.io/">OpenSea</a>, the first and largest marketplace for crypto collectibles and NFTs, where individuals can then buy and sell their NFTs.</p><p>When an NFT is purchased, the individual only purchases a link to the digital asset. Further, it is the link that lives on the blockchain. The data from the digital asset does not reside on the blockchain. For instance, when purchasing digital art as an NFT, the owner receives a hash code–for example, the hash code for<em> </em>“Everydays” is 6314b55cc6ff34f67a18e1ccc977234b803f7a5497b94f1f994ac9d1b896a017–or lone key of numbers and letters that directs the owner to the transaction on the blockchain that signifies their unique ownership over this NFT. The digital art itself does not live on the blockchain. It is important to point out that owning an NFT does not grant the owner any intellectual property rights over the digital asset associated with the NFT. Thus, in the instance of digital art, the NFT owner could not necessarily print out the piece of art and reproduce the artifact in any way that they would like to. The NFT also does not prevent copies of the digital art from circulating. Other instances of that piece can exist simultaneously, however, the replicas would not be <em>the</em> NFT. Despite this lack of property rights or control over reproduction, NFTs’ ability to be mathematically unique and transactionally transparent via blockchain is the key intrigue of NFTs. These particular attributes of NFTs promise authenticity among digital assets, and this quest for authenticity is not new.</p><h3><strong>The desire for authenticity</strong></h3><p>For a while now, there has been an obsession over obtaining the “authentic” or “original” work, particularly, but not exclusively, in the art world, with the end goal of establishing and certifying the value of a piece so that it can be monetized. The first written instance of an original work of art having “authenticity” was by German philosopher Walter Benjamin in 1935 in his essay <a href="https://web.mit.edu/allanmc/www/benjamin.pdf">“The Work of Art in the Age of Mechanical Reproduction”</a> where he explores art reproduction of various mediums and the philosophical issues that arise because of art’s ability to be copied. Since Benjamin’s writing, the desire for authenticity has not waned. People have been pining over and paying millions for true Warhols or Van Goghs rather than for a reproduction. Challenges surrounding proving authenticity and sniffing out frauds continue to persist with many cases getting to the point where the law is needed to settle disputes. However, according to lawyer Peter R. Stern, interviewed in an article by <em>The New York Times</em>, <a href="http://www.nytimes.com/2012/08/06/arts/design/when-judging-arts-authenticity-the-law-vs-the%20-market.html">“Ruling on Artistic Authenticity: The Market vs. the Law,”</a> “A decision by a court in the United States that a work is authentic may or may not have any value. It’s totally up to the market.” The law can not always resolve issues of art authenticity since the value of art is assigned by the people. With digital art becoming more prevalent, these issues of authenticity that arise due to the reproducibility of art has not gone away. Before arriving at the role of NFTs in this issue, I am first investigating the desire for the authentic asset, despite being digital or physical. I am focusing on authenticity in relation to art as there exists foundational undertakings on the subject. Regardless of the attention given to art, the discussion of authenticity can be broadened to other mediums that NFTs cover such as music and trading cards.</p><p>Nigel Warburton, a British philosopher, wrote about the “fetishistic element” of art associated with the original or authentic. In his 1997 paper “Authentic Photographs,” Warburton examines the medium of photographs with their ability to be copied. In reference to photography and what constitutes an authentic print, Warburton explains that</p><blockquote>the very fact that the photographer was involved in this direct way in the causal process of print-making, even if the photographer did not print the work, is significant. We should not underestimate the importance of this aspect of an original print, one which is frequently a psychological spur to imaginate involvement with the photographer’s embodied intentions.</blockquote><p>In order to certify, even a print, as authentic, Warburton claims the photographer must be involved: “only the photographer can determine that a print is of sufficient quality for it to count as an authentic print.” This is because “[a]nyone else can only hazard a guess, and may in the process seriously damage our understanding of a photographer’s style, and thus his or her possibility of creative expression.” The certification of the authentic ensures that the artist’s intentions and craft is not tainted. Benjamin, in his paper on art reproducibility, also comments on the aspiration to be connected to the artist’s process and history: “The authenticity of a thing is the essence of all that is transmissible from its beginning, ranging from its substantive duration to its testimony to the history which it has experienced.” An original piece of art carries with it a story that is valuable in itself. Additionally, Benjamin believes “the uniqueness of a work of art is inseparable from its being imbedded in the fabric of tradition,” or “in other words, the unique value of the “authentic” work of art has its basis in ritual, the location of its original use value.” There is a particular, “aura” as Benjamin explains, that “withers in the age of mechanical reproduction” thus revealing the desire for the authentic. It is this quality that accompanies the original that creates value for many individuals.</p><p>Related to the fetishistic element Warburton brings up, there is also a desire for closeness to the artist through owning the original work. Benjamin mentions</p><blockquote>the desire of contemporary masses to bring things ‘closer’ spatially and humanly, which is just as ardent as their bent toward overcoming the uniqueness of every reality by accepting its reproduction. Every day the urge grows stronger to get hold of an object at very close range by way of its likeness, its reproduction.</blockquote><p>In Benjamin’s instance, he claims people are willing to accept copies in order to get closer to the actual object since they can not have the original. This comment still reflects the desire for proximity to the artist.</p><p>Warburton proposes two kinds of authenticity–token authenticity and type authenticity–that make a useful distinction when examining NFTs. For Warburton’s classification, the key difference between the two types of authenticity is the presence of this fetish:</p><blockquote>In such cases of <em>token</em> authenticity, the fetish element of authenticity is preserved: we take token authenticity as the only reliable guide that compliance conditions have been met, which has the added benefit that the photographer will have ‘passed’ the print in question. But there is a further variety of authenticity: <em>type</em> authenticity. Here the fetish or relic element is lost. The photographer certifies a type, and implicitly sets compliance conditions.</blockquote><p>Further, “if two prints were virtually (or even <em>actually</em>) indistinguishable, only the one certified by the photographer would count as genuine or authentic,” according to Warburton. While Warburton was contemplating these questions surrounding authentic photographs in 1997, digital photography was just coming into the scene. Warburton hypothesized that the easily reproducible nature of digital photographs would create a shift away from token authenticity toward type authenticity. Warburton’s speculation on authenticity of digital versus physical photographs relates closely to the discussion today with digital work and NFTs. Just as digital photographs are inherently reproducible, so is digital art and nearly any other digital asset that could exist as an NFT.</p><p>As discussed by Warburton and separately by Martin Zeilinger–a researcher and philosopher who is interested in digital art and emerging technologies–in his paper “Digital Art as ‘Monetised Graphics’: Enforcing Intellectual Property on the Blockchain”, being reproducible means that little to no value is assigned to copies. However, all the value is given to the original. In Zeilinger’s discussion of the nature of digital work, he points out the challenge digital art faces in being authentic:</p><blockquote>the digital is almost always easy to reproduce and share, suggesting, again that immaterial digital-born artworks cannot easily attain the status of the unique, desirable collector’s item that continue to inform the standards by which the commercial art world measures value in both its aesthetic and commercial connotations.</blockquote><p>Zeilinger uses this point to support his claim that platforms like Monegraph, a blockchain media rights company which can be likened to today to NFT online marketplaces, will not help artists in their attempt for control over their work: “it may be the nature of the digital itself that stands in the way of turning digital art into a stable financial instrument.” However, Monegraph and other similar platforms work to create authenticity where it does not currently exist. Zeilinger explains this phenomenon as “an ontological transformation from infinitely reproducible intangible artefacts to ones that are artificially scarce and therefore, potentially, valuable.” Since we desire authentic art for its connection and closeness to the artist and for its promise of particular qualities, and now face problems with authenticity for digital work, NFTs have arisen as a potential solution in order to fabricate authenticity to assign value to digital assets which can then be bought and sold.</p><p>NFTs intend to solve the problem of verifying art by using blockchain technology to provide unique hash codes that signify an individual’s ownership of the NFT. While this is the promise of NFTs, that does not necessarily mean the technology is capable of doing so in the real world. Zeilinger believes that the question of authenticity has shifted when we talk about NFTs:</p><blockquote>Technically, our understanding of the image’s ‘authenticity’ would no longer be bound to any particular copy of it but rather to a distributed blockchain hash that accurately and irreversibly documents the GIF’s [or other digital asset] origin, ownership history, etc. Conceptually, the authenticity of the image would no longer be inscribed in a scarce ‘original’ and linked to the creator of this artefact but rather in its distinctiveness as verified by the blockchain hash.</blockquote><p>In his view, NFTs are ineffective at providing this authenticity that can lead to potential value thus not giving artists a stable income they desire.</p><h3><strong>What’s next for NFTs?</strong></h3><p>Since the technology is still in its infancy, there remains a host of possibilities for future uses and an unpredictable future for the technology. There is discussion among content creators on YouTube to use NFTs to establish shares in their videos by minting short clips of a larger video in an effort to lessen reliance on YouTube’s monetary policies. Other possibilities exist for using NFTs to decentralize big business, but many people including Zeilinger and Dash, the first NFT creator, suspect that NFTs will continue to benefit the big companies supplying the technology rather than the artists. It is useful to consider the original intention of NFTs in relation to the state of NFTs today.</p><figure><img alt="Pixelated figure with a hat and sunglasses on smoking a pipe." src="https://cdn-images-1.medium.com/max/336/1*SuKB4fIwT4uPfQW2VhWGYg.png" /><figcaption>CryptoPunk #7804 sold for $7.5 million.</figcaption></figure><p>In a recent <em>Atlantic </em>article, <a href="http://www.theatlantic.com/ideas/archive/2021/04/nfts-werent-supposed-end-like/618488/">“NFTs Weren’t Supposed to End Like This,”</a> Dash explains how McCoy, the artist Dash partnered with for the first NFT, wanted a way for artists to have ownership over their work. This conversation between Dash and McCoy happened during the time when people were reposting artists’ works that were not their own on Tumblr, generating engagement (and likely monetary compensation). The original intention for the NFT technology Dash and McCoy developed was for providing artists with ownership rights and protection over their work. Today, “Quantum” is valued at $7 million and Dash has no stake in the high value of that asset. Dash points out how artists are not the ones profiting from NFTs as originally intended, rather it is the tech industry seizing opportunity in a new multi-billion dollar market. Additionally, since platforms like Ethereum are responsible for maintaining the public record of ownership, individuals are dependent on the platform continuing to operate into the far future if they hope to keep their NFT asset for a while. This establishes a new dependency for artists hoping to make money by getting around other monetary reliances. However, many artists like digital artist Alfredo Salazar-Caro, the one interviewed on <em>Wall Street Journal</em>’s “The Future of Everything” podcast episode <a href="http://www.wsj.com/podcasts/wsj-the-future-of-everything/how-nfts-could-disrupt-the-art-mark%20et/fa20d12b-81e2-4897-8bea-ca53e14b6671">“How NFTs Could Disrupt the Art Market,”</a> are hopeful that NFTs is the way forward for selling their art. Yet, sales for Salazar-Caro so far have been unsuccessful.</p><figure><img alt="Rotating cube GIFT of LeBron James." src="https://cdn-images-1.medium.com/max/1024/1*IZq3SeNPj5VwskT3FHP7Aw.gif" /><figcaption>NBA TopShot NFT for sale on their collectors platform.</figcaption></figure><p>There is speculation on whether NFTs will remain a collector’s economy or whether it will disrupt economies. As mentioned prior, NBA’s TopShot is leading the NFT market in sales with its trading cards and highlight clips–a great example of the use and enthusiasm for NFTs for collectors. It is possible that the NFT market stays niche, but there is also continued expansion of uses particularly in virtual worlds like Decentraland. Public voices such as Gary Vaynerchuk, CEO of VaynerMedia, among others are making big bets on the NFT market. <a href="http://www.cnbc.com/2021/05/05/gary-vee-launching-art-nfts-linked-to-ethereum-cryptocurren%20cy.html">Vaynerchuk has even suggested that small businesses should utilize NFTs for memberships, clubs, and discounts</a>. Again, Vaynerchuk is highlighting the desire for verifiable authenticity that NFTs could bring. It will be interesting to see how the NFT market transforms along with blockchain and cryptocurrency in the coming years. Despite whether or not NFTs are here to stay, the issue of authenticity for digital assets will persist because of the reproducible nature of the medium and the desire to hold the original asset.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6d516d81ed5a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The power of storytelling in the remote workspace]]></title>
            <link>https://sheridanemma.medium.com/the-power-of-storytelling-in-the-remote-workspace-648ed74e8fe?source=rss-5a1c659a8b94------2</link>
            <guid isPermaLink="false">https://medium.com/p/648ed74e8fe</guid>
            <category><![CDATA[storytelling]]></category>
            <category><![CDATA[communication]]></category>
            <category><![CDATA[remote-working]]></category>
            <category><![CDATA[presentations]]></category>
            <category><![CDATA[presentation-design]]></category>
            <dc:creator><![CDATA[Emma Sheridan]]></dc:creator>
            <pubDate>Thu, 08 Apr 2021 23:42:17 GMT</pubDate>
            <atom:updated>2021-04-08T23:42:17.697Z</atom:updated>
            <content:encoded><![CDATA[<h4>Why you should care about your slides</h4><figure><img alt="Cartoon-like drawing of a woman standing next to a presentation." src="https://cdn-images-1.medium.com/max/1024/1*9o-i4TX397YNczCGVXtqqg.jpeg" /><figcaption>Care about your slides! (Photo from Salesforce.)</figcaption></figure><p>For awhile now, I’ve been a stickler for aesthetically pleasing presentations. Fonts and font sizes must be consistent, headings and margins should be aligned across pages, and colors should stick to a theme. I will notice that one picture that’s shifted slightly off the center line and cringe internally. Maybe you think this is extreme, that this level of polished is not worth the time, or that what really matters is the content on the slides. I urge you to think twice about that. In the remainder of this piece, I’ll share some things I’ve learned recently in hopes of convincing you to spend a little extra time putting together your next presentation. In particular, I’ll tie this skill to the new normal of the remote work environment and how I’ve come to find storytelling even more important when you can’t be with your co-workers in-person.</p><h3>A presentation gone awry</h3><p>Half way through my virtual internship, I was asked to prepare a few slides to update the greater team on the work I’d been doing and some interesting things I discovered. The meeting where I planned to share these slides was scheduled for an hour and a half. The meeting starts, and we begin on the first slide. The slide featured three bulleted lists of size 12 font that completely filled the screen. There was enough content on the one slide for the entire meeting–and it nearly was. The team spent 40 minutes talking about the first slide. Maybe it was just me, but I stopped paying attention to the presentation after minute 5 on the same slide. This pained me. During this time, my teammate messaged me privately to say that there would not be enough time for my slides since we were behind schedule. To be honest, I was disappointed. I had a really good story and was excited to share it. And it wasn’t just me that was affected by spending 40 minutes on one slide. Additionally, it took too long to realize we needed to keep the presentation moving. The remainder of the slides were rushed and things were overlooked.</p><p>I will acknowledge that this meeting was the first of its kind, so there was a major learning curve to the time and formatting for the presentation. But nevertheless, I think it’s a great opportunity to show how to improve storytelling. And, I know I’m not the only one who has witnessed a presentation that has gone awry.</p><h3><strong>Make it digestible</strong></h3><p>The less content on a slide, the better. The average human attention span is 8 seconds. Ask yourself: What is the most important piece of information this slide needs to convey? And there should only be one important piece of information per slide. I’ve been told you should stay on a slide at most one minute before moving on. This has proven to be good advice. If you need to tell your audience to save their questions and comments until you get through a section, do that. Do what you need to keep the presentation moving and everyone on the same page.</p><h3><strong>Make it a story</strong></h3><p>How do you not lose the audience by slide 3? Tell a story–one with a beginning, a middle, and an end. You don’t have to be J.K. Rowling to be able to create a storyline for your presentation. One of the simplest ways to work on improving your story for a presentation is to use a tool called a Story Spine. The Story Spine was developed to give structure to creating a story and has been used by playwrights. It goes like this:</p><blockquote>Once upon a time… (introduction to customer/persona)</blockquote><blockquote>Every day… (routine)</blockquote><blockquote>But one day… (event that breaks the routine)</blockquote><blockquote>Because of that… (pain point)</blockquote><blockquote>Because of that… (pain point)</blockquote><blockquote>Because of that… (pain point)</blockquote><blockquote>Until finally… (a solution)</blockquote><blockquote>And ever since then… (new routine)</blockquote><p>By following this structure, you can create a story that puts the customer at the center of your presentation. However, don’t feel like you have to be rigid in maintaining this outline for your story. Adapt this methodology to your needs or use something else entirely. Having a story laid out like this will directly transition into making slides for your presentation and save time when it comes to putting the slides together.</p><h3><strong>Make it polished</strong></h3><p>You don’t have to be a visual designer to make nice looking slides. To me, having a polished presentation comes down to this: If you put tons of time and effort into gathering the content that you’re going to share, you should put the same amount of time and effort in making sure it’s presentable. Else, you’re doing a disservice to all the work you did to get to the point of presenting.</p><p>At a minimum, proofread, use consistent colors and fonts, use the gridlines to help align content, and limit the amount of text on a single slide. If you feel like you need a lot of text, try breaking one slide into three slides and replacing words with images, icons, graphs or other visuals. And, there’s no need to use the same layout for every slide. Change it up! Keep the people engaged. If your presentation is quite long, try breaking it up with section title slides. An agenda at the start of your presentation will also help everyone with pacing if you have lots of slides to get through.</p><h3><strong>F the script</strong></h3><p>Don’t read from a script. A script can be useful for planning what you want to mention on each slide, but reading from a script sounds robotic and is a surefire way to lose your audience. You want to know your slides so well that you don’t need notes. One thing I like to do to make sure I don’t forget an important point is to have enough words on the slide to trigger my comments about a point. Also, if you’re into animations, they can help you move through a slide as well.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/400/1*MYtuyXgsRSvXhujhSBNdIw.gif" /></figure><h3><strong>Why storytelling matters even more right now</strong></h3><p>While many companies are still working remote, and plan to continue to allow their employees to work remotely for the foreseeable future, storytelling has an increased power in rallying the troops around the problems and solutions. I found that it was much harder to share insights regarding a customer problem when stakeholders could not simply walk by my workspace. Instead, most interactions are scheduled. I needed to put time on my manager’s calendar in order to walk through the customer pain points I identified. And further, my time with my manager needed to be productive and all digital. Storytelling (and a good presentation deck) was essential in being able to communicate efficiently.</p><p>It wasn’t until my final presentation of the summer that I realized just how effective my methods of storytelling were in rallying the team around my project. Don’t get it twisted though, I’m still learning. I’d love to hear what others tips and tricks are for improving storytelling and for creating an engaging and memorable presentation.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=648ed74e8fe" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The curb cut effect: How universal design makes things better for everyone]]></title>
            <link>https://uxdesign.cc/the-curb-cut-effect-universal-design-b4e3d7da73f5?source=rss-5a1c659a8b94------2</link>
            <guid isPermaLink="false">https://medium.com/p/b4e3d7da73f5</guid>
            <category><![CDATA[universal-design]]></category>
            <category><![CDATA[inclusion]]></category>
            <category><![CDATA[tech]]></category>
            <category><![CDATA[ux]]></category>
            <category><![CDATA[accessibility]]></category>
            <dc:creator><![CDATA[Emma Sheridan]]></dc:creator>
            <pubDate>Mon, 01 Feb 2021 02:37:53 GMT</pubDate>
            <atom:updated>2021-02-02T23:59:29.043Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="An image of a curb cut on the corner of a street. The sidewalk grades down to the street to allow for smooth crossing." src="https://cdn-images-1.medium.com/max/1024/1*py_-ZxUCxSjkXo5r4Eo3sg.jpeg" /></figure><p><strong>It was a beautiful day in Santa Barbara, California,</strong> and I was walking down to work listening to an episode of my favorite podcast <em>99% Invisible</em> (I highly recommend). This <a href="https://99percentinvisible.org/episode/curb-cuts/">episode talked about curb cuts</a>–the ramps graded down on sidewalks to meet with the street. Here is yet another example of design that goes unnoticed by many. Little did I know that there was an extensive history behind this seemingly simple design that makes walking through the streets more accessible for those with disabilities and even those without.</p><p>In fact, the impact of this kind of design is so significant that the term <em>curb cut effect </em>was coined. The curb cut effect states that when you design for disabilities, you make things better for everyone in the process. For example, curb cuts on sidewalks made it easier for people pushing strollers, the elderly with walkers, or those rolling a bag behind them. Implementing this design across the country was no easy task.</p><h3>A quick history lesson: curb cuts</h3><p>It all started with the curb cut movement in the revolutionary 60s in Berkeley, California. Ed Roberts, an activist and graduate student at University of California, Berkeley, was the leader of this revolution. When Roberts was young, he became sick with polio which left him paralyzed from the neck down. The only part of his body he could move were two fingers on his left hand. Though it was difficult, he was determined to go to school and make a difference. He went to class with the help of an attendant pushing his wheelchair and had a classmate make carbon copies of their notes. Roberts’s story was an anomaly for the time, and he soon began making headlines. Shortly after Roberts started at Berkeley, quite a few other disabled students arrived as well. Together, they began a dialogue regarding the civil rights of disabled peoples, such as the right to be included in public life with something as simple as ease of travel through the streets.</p><p>There are rumors that Roberts and friends went out at night and tore up the corners of sidewalks to install their own ramps demanding action from the city. Though these stories may have been exaggerated, they reflect the revolutionary spirit of the times. Soon, the world’s first curb cut program was implemented in the city of Berkeley. On September 28, 1971, it became city policy to make streets and sidewalks accessible for the disabled in major commercial areas.</p><p>In the 70s, other changes to the built environment, such as wheelchair lifts on buses, accessible bathrooms, and low countertops, came into play. Demonstrations and protests were held across the country demanding these changes be implemented in all facilities. These years of protest culminated with the 1990 <a href="https://www.ada.gov/">Americans with Disabilities Act</a> which “prohibits discrimination and guarantees that people with disabilities have the same opportunities as everyone else to participate in the mainstream of American life — to enjoy employment opportunities, to purchase goods and services, and to participate in State and local government programs and services.”</p><h3>What still needs work: the built environment</h3><p>It is important to recognize how far we’ve come from the 60s, but it’s equally important to see what still needs work in terms of making design in the built environment universally accessible. We have not reached total inclusion yet. Let’s look at Uber, for example.</p><figure><img alt="An image of a driver loading a person in a wheelchair into the back of a wheelchair accessible car using a ramp for a ride." src="https://cdn-images-1.medium.com/max/960/1*C4SXXFoJksdozPxMT9k0vg.jpeg" /></figure><p>For many of us, Uber’s ride-share service is quick and easy to use. But what if you’re in a wheelchair? In 2016, Uber launched UberWAV, a service that provides wheelchair-accessible vehicles (WAV). Since its implementation there have been <a href="https://www.law.com/therecorder/2020/03/13/judge-doesnt-buy-ubers-claims-that-it-is-not-a-transportation-company-in-ada-suit/?slreturn=20200224175817">tons of lawsuits</a> regarding sub-par service, inconsistent access to wheelchair-friendly cars, and longer wait times for UberWAV. In New York, there’s approximately 200 wheelchair-friendly vehicles out of the more than 58,000 cars in the city. The UberWAV service clearly still needs work to ensure equal and universal access. Additionally, UberWAV is not offered in all cities where Uber operates. <a href="https://www.uber.com/us/en/drive/services/uberwav/">According to Uber’s website</a>, UberWAV is currently only available in Chicago, New York City, Philadelphia, and Washington, DC.</p><p>The New York City Subway poses additional problems for wheelchair users. Elevator accessibility across stations is poor with 80% of the stations being inaccessible to disabled people. Further, preventative maintenance is rarely–definitely not in a timely fashion–performed on the elevators of New York’s aged subway system. As a part of MTA’s 2020–2024 capital plan, <a href="https://new.mta.info/sites/default/files/2019-09/20-24%20Capital%20Plan%20Overview.pdf">70 new stations will become ADA-accessible</a>. This is a step in the right direction.</p><figure><img alt="An elevator shaft on the ground level of a subway station. The elevator shaft is located in the middle of the platform." src="https://cdn-images-1.medium.com/max/1024/1*bnSkDt5rvOWJmGO5WfOBMQ.jpeg" /></figure><h3>What still needs work: the virtual space</h3><p>Typically, universal design is discussed in relation to physical space, but now it has been applied to virtual space as well. The information technology industry uses the term <em>electronic curb cuts. </em>Much of what we interact with today is our devices: phones, tablets, computers, etc. However, Americans with disabilities are <a href="https://www.pewresearch.org/fact-tank/2017/04/07/disabled-americans-are-less-likely-to-use-technology/">3 times as likely</a> than those without a disability to not go online. Additionally, across ages, those with disabilities often don’t even own a laptop, smartphone, or tablet.</p><p>Many of these technologies provide some accessibility settings such as enlarged font for those with poor vision, text-to-speech and voice-to-text features, subtitles and captioning, and LED flash for alerts. Further, standards such as <a href="https://www.w3.org/WAI/standards-guidelines/wcag/">Web Content Accessibility Guidelines (WCAG)</a> exist for web pages and applications. Generally, these standards refer to how text, images, and sounds are presented and structured for web content. For example, “All non-text content that is presented to the user has a text alternative that serves the equivalent purpose.” Often, these accessibility features are not prioritized in the development of web content and other applications. Accessibility should not be an afterthought, rather considered from the creation of the technology by calling on universal and inclusive design processes to guide the development from the start.</p><h3>Help one, help all</h3><p>While there are many areas that need improvement, it doesn’t hurt to point out good examples of universal design and how they have helped everyone. A classic example is Oxo Good Grips.</p><p>The creation of <a href="http://universaldesign.ie/What-is-Universal-Design/Case-studies-and-examples/3-case-studies-on-UD/#oxo">Oxo Good Grips</a> started with the goal of creating a vegetable peeler that was easy and comfortable to use for someone with arthritis. In developing the product, the grip was tested with people of various ages, hand sizes, hand strength and dexterity. The consideration of a functional and attractive product for all peoples led to OXO Good Grips being a widely-recognized design and brand.</p><figure><img alt="From left to right, a peeler, can opener, handheld grater, tongs, ice cream scooper, and a whisk." src="https://cdn-images-1.medium.com/max/500/1*fuvPWHSHa7X1Ca7gyDvLug.jpeg" /></figure><p>A perhaps lesser-known universal design success is the <a href="https://www.archdaily.com/495736/house-of-disable-people-s-organization-cubo-force4">House of Disable People’s Organization</a> in Taastrup, Denmark. Designed by architect firms Cubo Arkitekter and FORCE4 Architects, this office building promotes equal access, and further encourages users to be self-reliant, through several design features. Many of these considerations are small low-tech features such as floor numbering along the railings or moving the elevator closer to the stairs in order to avoid disrupting conversation if two individuals need different modes of transport. (I recommend reading the linked article to learn more about the design.)</p><figure><img alt="Rendering from inside of House of Disable People’s Organization. 5 story building with blue, red, green columns. Open layout." src="https://cdn-images-1.medium.com/max/780/1*G0dj70DThpvzyDoJIZ89cQ.jpeg" /><figcaption>A rendered image of the inside of the House of Disable People’s Organization.</figcaption></figure><p>Lastly, in tech, assistive technologies (AT) are a great example of electronic curb cuts that often exhibit the curb cut effect. Let’s take Closed Captioning for a brief example. Closed Captioning was originally created so that the deaf and hard-of-hearing could watch television. But today, people watching football games in loud bars or watching the news while at the gym benefit from captioning; and, we have the deaf community to thank for that. Though there are plenty of examples like Closed Captioning where everyone benefits from AT, there is still a lack of consideration for accessibility in the technology world today.</p><figure><img alt="An icon image of the Closed Captioning logo that has two black letter c’s in a white box with rounded edges." src="https://cdn-images-1.medium.com/max/259/1*pde5IM99NwkkLO4Ka6-Ejw.png" /></figure><p>Fortunately, some companies such as Apple have been committed to accessibility for all for a while now making their built-in features widely advertised, and the company has received several awards for their inclusive designs. The company quite clearly prides themselves on these accomplishments and <a href="https://www.apple.com/newsroom/2020/07/apple-creatives-and-disability-rights-activists-reflect-on-30-years-of-the-americans-with-disabilities-act/">promotes their achievement</a> that “every product and service in the Apple ecosystem is designed with accessibility built in.” Apple’s accessibility features are also great examples of the curb cut effect. If you are able-bodied, you have probably benefitted from features such as Siri, Dictation, FaceTime, Safari Reader, and more, which are considered accessibility features.</p><h3>Accessibility is not a barrier</h3><p>To conclude, accessibility should not be considered a barrier. Designing with all people in mind doesn’t hold you back but rather forces you to make the experience better for everyone. These universal design successes should encourage companies to make their products and services universally accessible, and should encourage cities to make the built environment universally accessible.</p><p>Though this is not a focus of this particular article, there must also be a push for greater education and awareness of incorporating accessibility throughout the design process in both the physical and virtual realms. Let’s get the conversation concerning equitable access going sooner. Practicing empathy (<a href="https://medium.com/design-ibm/practicing-empathy-through-design-thinking-6f45d2cbe9ff">shocker!</a>) will be key to championing accessible experiences for all to enjoy.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/94/0*S6p7LRiXe_TqEDHT" /><figcaption>The UX Collective donates US$1 for each article published on our platform. This story contributed to <a href="https://www.bayareablackdesigners.com/">Bay Area Black Designers</a>: a professional development community for Black people who are digital designers and researchers in the San Francisco Bay Area. By joining together in community, members share inspiration, connection, peer mentorship, professional development, resources, feedback, support, and resilience. Silence against systemic racism is not an option. Build the design community you believe in.</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b4e3d7da73f5" width="1" height="1" alt=""><hr><p><a href="https://uxdesign.cc/the-curb-cut-effect-universal-design-b4e3d7da73f5">The curb cut effect: How universal design makes things better for everyone</a> was originally published in <a href="https://uxdesign.cc/">UX Collective</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Un-bottling the data]]></title>
            <link>https://medium.com/data-science/un-bottling-the-data-2da3187fb186?source=rss-5a1c659a8b94------2</link>
            <guid isPermaLink="false">https://medium.com/p/2da3187fb186</guid>
            <category><![CDATA[data]]></category>
            <category><![CDATA[data-exploration]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[towards-data-science]]></category>
            <dc:creator><![CDATA[Emma Sheridan]]></dc:creator>
            <pubDate>Mon, 02 Dec 2019 17:42:42 GMT</pubDate>
            <atom:updated>2019-12-03T17:44:58.626Z</atom:updated>
            <content:encoded><![CDATA[<h4>Exploring the big data world of wine.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PnZ6g2BrXYvQC_lCpfDAsQ.png" /></figure><h3>The Motivation</h3><p>In this blog post, we will go through several different data exploration methods using data on wine! We will show step by step code and explain our processes along the way. Before we started diving into the code, we posed a few questions for ourselves that we were hoping to answer about the wine dataset:</p><ol><li>Which countries have the most reviewed wine?</li><li>Where are the world’s best wines from?</li><li>Where does the priciest wine come from?</li><li>Is there a correlation between wine price and points?</li><li>What words describe the top 10 wine types?</li><li>Can the description predict a feature of the wine?</li></ol><p>These questions and other findings will be addressed in the remainder of the blog post using Natural Language Processing (NLP), word clouds, map visualizations, and other calculations in Python.</p><h3>The Data</h3><p>We found our dataset, “Wine Reviews,” on <a href="https://www.kaggle.com/zynicide/wine-reviews">Kaggle</a>. The dataset consists of over 130,000 wine review entries. Each entry contains the country the wine came from, a description of the wine, the designation (vineyard within the winery where the grapes are from), the number of points WineEnthusiast rated the wine on a scale of 1–100, the price for a bottle of wine, the province or state the wine is from, the region or wine growing area, the title of the wine review, the variety or type of grape used to make the wine, and lastly, the winery. Part of the raw data can be seen below.</p><pre>wine_reviews = pd.read_csv(“winemag-data_first150k.csv”)<br>wine_reviews.dropna() # drop any empty entries<br>del wine_reviews[&#39;Unnamed: 0&#39;] # delete first column <br>wine_reviews.head()</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7pdqdVyPhGQZxa-Dvl7ghg.png" /><figcaption>The first five entries from the Wine Reviews dataset.</figcaption></figure><h3>Which countries have the most reviewed wine?</h3><p>The top 10 countries make up 143,344 reviews which accounts for 95% of all reviews. There are 46 total countries in the dataset.</p><p>The US has the most wine reviews in this dataset with 62397 reviews (41.34%). This is not surprising since the dataset is from WineEnthusiast, a New York-based company. Following the US are Italy (15.56%), France (13.98%), Spain (5.48%), and Chile (3.85%).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1008/1*N5oQcnBDkTb3VGf3VN2yjg.png" /></figure><p><strong><em>The map below is interactive, please visit </em></strong><a href="https://wsjung.github.io/caobd_project/maps/d3-freq.html"><strong><em>this link</em></strong></a><strong><em> to see it in action.</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XJCKSEtekNUipQol_vSsQw.png" /></figure><h3>Where are the world’s best wines from?</h3><p>Surprisingly, the UK leads the list when it comes to wine review points with an average of 92.89 points. It’s interesting to see that the US did not make the top 10. Looking back at the frequency data, the UK only has 9 reviews which is far too little compared to 62397 of the US. Our data is skewed, and this isn’t quite the fairest comparison.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AMOi_hdhSaTCtN_7xgyFxQ.png" /></figure><p><strong><em>The map below is interactive, please visit </em></strong><a href="https://wsjung.github.io/caobd_project/maps/d3-points.html"><strong><em>this link</em></strong></a><strong><em> to see it in action.</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*y2WkyNU5Uo0FRHeAE3Zqaw.png" /></figure><h3>Where does the priciest wine come from?</h3><p>The UK hits number one for priciest wine with an average price of $47.50. France is in a close second with an average price of $45.61, and Hungary follows with $44.20. The United States ranks 8th with an average price of $33.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gW2oiMyQoKpzirBOZzzM2A.png" /></figure><p><strong><em>The map below is interactive, please visit </em></strong><a href="https://wsjung.github.io/caobd_project/maps/d3-prices.html"><strong><em>this link</em></strong></a><strong><em> to see it in action.</em></strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qtV8UT6nsKVkVOZowm8-GQ.png" /></figure><h3>Is there a correlation between wine price and points?</h3><p>First, we can look at some summary statistics for price and points.</p><pre>wine_reviews.describe()</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jZPPKPIpnTjY1WvNPSRnFw.jpeg" /></figure><p>Let’s visualize the distribution of wine price and points.</p><pre>plt.figure(figsize=(15,5))<br>plt.subplot(1, 2, 1)<br>plt.hist(wine_reviews[‘price’], color=&#39;#ba281e&#39;)<br>plt.xlabel(“Price”)<br>plt.ylabel(“Frequency”)<br>plt.title(“Distribution of Wine Prices”)</pre><pre>plt.subplot(1, 2, 2)<br>plt.hist(wine_reviews[‘points’], color=&#39;#ba281e&#39;)<br>plt.xlabel(“Points”)<br>plt.ylabel(“Frequency”)<br>plt.title(“Distribution of Wine Review Points”)</pre><pre>plt.tight_layout()<br>plt.show()</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-K26Hkl6ivvnoqXGuzp-Qg.png" /></figure><p>We can now look at the relationship between the variables price and points. Using the seaborn package we can visualize the variables’ relationship.</p><pre>import seaborn as sns</pre><pre>plt.figure(figsize=(10,14))<br>ax = sns.jointplot(x=&#39;price&#39;, y=&#39;points&#39;, data=x[x[&#39;price&#39;] &lt; 200], kind=&#39;hex&#39;, gridsize=20, cmap=&#39;Reds&#39;)<br>plt.tight_layout()<br>plt.show()</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/437/1*W6j4Ul07dRh-s4Oj65s7Mg.png" /></figure><p>There is a positive relationship between points and price–as the price of the wine increases, the points given to the wine increase as well. The average price is around $20 and average points is around 88. We can also look at the distribution for wines that received 100 points. We see that even cheaper wines can receive full marks!</p><pre># prices of wines that received perfect points <br>perfect_score = wine_reviews[‘points’]==100<br>perfect_wines = wine_reviews[perfect_score]<br>plt.hist(perfect_wines[&#39;price&#39;], color =&#39;#ba281e&#39;)<br>plt.xlabel(&quot;Price&quot;)<br>plt.ylabel(&quot;Frequency&quot;)<br>plt.title(&quot;Wines with 100 Points&quot;)<br>plt.show()</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/382/1*WyQ08fnsrF2aIjlxO59WeA.png" /></figure><h3>What words describe the top 10 wine types?</h3><p>Using word clouds, we analyzed the descriptions for the top 10 wine types. We took out words like <em>wine, bottle, glass, flavors, grapes, pour, sip, </em>and <em>taste </em>so that we were left with solely descriptors. The code below shows how we created the two of the individual word clouds. The rest of the word clouds were created in the same way.</p><pre># creates the word clouds<br>bordeaux_wc = wordcloud.WordCloud(background_color=”white”, max_words=1000, mask=transformed_wine_mask, contour_width=3, contour_color=’firebrick’).generate(bordeaux_style_white_blend)</pre><pre>syrah_wc = wordcloud.WordCloud(background_color=”white”, max_words=1000, mask=transformed_wine_mask, contour_width=3, contour_color=’firebrick’).generate(syrah)</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MnWBw-eN475Lg9ijp4GdUg.png" /></figure><p>As you can see from the word clouds above, the top words for the top 10 wines include <em>fruit, spicy, dry, sweet, aroma, palate, herb, and tannin</em>. Some of the wines have more specific descriptors:</p><ul><li>Bordeaux Style White Blend: <em>apple, citrus, honey, grapefruit, acidity</em></li><li>Syrah: <em>pepper, complex, licorice, young, elegant</em></li><li>Muscat: <em>dessert, acidity, balanced, apricot, honeysuckle</em></li><li>Prugnolo: <em>leather, smoke, tobacco, crisp, thick</em></li></ul><h3>Can the description predict a feature of the wine?</h3><p>In order to approach this question, we explored the data looking for features that could be used for prediction. We found that the variables <em>province</em> and <em>variety</em> have a largely skewed distribution, but <em>points</em> had a normal distribution. We decided to use the description to predict the points a wine would receive.</p><p><strong><em>Creating and training the model:</em></strong> First, we preprocessed the description text using word embeddings from a pre-trained global vectors text file that rates words on the relationship they have between each other. We then used 3 different models to evaluate how to predict points based on the description. These models included a 1D convolutional neural network (CNN), 1D CNN with a long-short term memory (LSTM) layer, and a 1D CNN with gated recurrent units (GRU). CNN is a basic model for reading word embeddings. LSTM is good at finding patterns in sequences of words. GRU is another recurrent neural network architecture, and similar to LSTM, GRU is good at understanding word context. We did not use purely an LSTM model because they do not perform well for sentiment predictions. The code below is for the 1D CNN.</p><pre><em># set parameters:</em><br>max_features = vocab_size<br>maxlen = 132<br>BATCH_SIZE = 128<br>embedding_dims = 50<br>filters = 250<br>kernel_size = 3</pre><pre>cnn = Sequential()<br><br><em># start with an efficient embedding layer which maps</em><br><em># vocab indices into embedding_dims dimensions</em><br>embedding_layer = Embedding(vocab_size, 100, weights=[embedding_matrix], input_length=maxlen , trainable=<strong>False</strong>)<br><br>cnn.add(embedding_layer)<br><br><br><em># we add a Convolution1D, which will learn filters</em><br><em># word group filters of size filter_length:</em><br>cnn.add(Conv1D(BATCH_SIZE,<br>                 kernel_size,<br>                 padding=&#39;valid&#39;,<br>                 activation=&#39;relu&#39;,<br>                 strides=1))</pre><pre><em># we use max pooling:</em><br>cnn.add(GlobalMaxPooling1D())<br><br><em># We add a vanilla hidden layer:</em><br>cnn.add(Dense(BATCH_SIZE))<br>cnn.add(Activation(&#39;relu&#39;))<br><br><em># We project onto a single layer</em><br>cnn.add(Dense(1))<br><br>cnn.compile(loss=&#39;mse&#39;,<br>              optimizer=&#39;adam&#39;,<br>              metrics=[&#39;accuracy&#39;])</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*deaG0hqeawVbYkALLGrYMg.png" /><figcaption>Model summary output for 1D CNN.</figcaption></figure><p><strong><em>Evaluating the models: </em></strong>The best performing model was the 1D CNN that got an accuracy of 30% on the validation set and an average error of 3 points (error based on mean squared error). We also tried breaking the points into classes in a 5 point range. Using this method, the model had an accuracy of 62% with an average error of 0.3 points. Our results tell us that it is hard to predict an exact point value for a given description because there is so much variation. However, it seems that we can predict descriptions within a range of points pretty well.</p><p><strong><em>Problems:</em></strong> The descriptions given were some what problematic for predictions. There are only 20 unique reviewers that contributed to this dataset. Each reviewer has their own style of writing making the prediction problem a bit more challenging.</p><h3>Reflections</h3><p>This project was a great opportunity for us to use most of the skills that we learned throughout the semester in our Computational Analysis of Big Data class. Unfortunately, we didn’t have access to a dataset that didn’t focus on American wines. If time permitted, we would’ve liked to do a US-specific analysis as well. We would’ve liked to use a better dataset to perform other predictions such as looking at how the origin of a wine affects its rating.</p><h3>References</h3><ul><li><a href="https://amueller.github.io/word_cloud/auto_examples/masked.html#sphx-glr-auto-examples-masked-py">Masked wordcloud - wordcloud 1.6.0.post1+g8217e20 documentation</a></li><li><a href="https://www.d3-graph-gallery.com/bubblemap.html">Bubble map | the D3 Graph Gallery</a></li><li><a href="https://seaborn.pydata.org/api.html">API reference - seaborn 0.13.2 documentation</a></li></ul><p><em>Project done by Derek Albosta, Woo Jung, Emma Sheridan, and Erik Tacam for Computational Analysis of Big Data course. This projects source code can be found on </em><a href="https://github.com/wsjung/caobd_project"><em>GitHub</em></a><em>. The interactive map visualizations can be found </em><a href="https://wsjung.github.io/caobd_project/"><em>here</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2da3187fb186" width="1" height="1" alt=""><hr><p><a href="https://medium.com/data-science/un-bottling-the-data-2da3187fb186">Un-bottling the data</a> was originally published in <a href="https://medium.com/data-science">TDS Archive</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Looking for the perfect words? Generate them.]]></title>
            <link>https://medium.com/data-science/looking-for-the-perfect-words-generate-them-b28b818c2f24?source=rss-5a1c659a8b94------2</link>
            <guid isPermaLink="false">https://medium.com/p/b28b818c2f24</guid>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[lstm]]></category>
            <category><![CDATA[poetry]]></category>
            <category><![CDATA[neural-networks]]></category>
            <dc:creator><![CDATA[Emma Sheridan]]></dc:creator>
            <pubDate>Mon, 02 Dec 2019 13:05:52 GMT</pubDate>
            <atom:updated>2021-07-01T05:43:20.790Z</atom:updated>
            <content:encoded><![CDATA[<h3>Looking for the Perfect Words? Generate Them</h3><h4>Using an LSTM model to generate poems.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*kvy_G-Y6YHmEubCe" /><figcaption>Photo by <a href="https://unsplash.com/@thoughtcatalog?utm_source=medium&amp;utm_medium=referral">Thought Catalog</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a>.</figcaption></figure><h3>The Motivation</h3><p>Ever try to find the right words to describe how you are feeling? Want to profess your love to your significant other in a poem? Why don’t you just generate one based on thousands of well-known poems? This is what we attempted to do using a Long Short-Term Memory (LSTM) model.</p><h3>What is a LSTM model?</h3><p>An LSTM model has an artificial recurrent neural network (RNN) architecture. Since it is a recurrent network, it means that the model has feedback connections. Additionally, it can process entire sequences of data such as sentences or video. LSTM is often used for handwriting recognition and speech recognition. It is good at working on these types of problems because of its selective remembering patterns. A typical LSTM unit consists of a cell, an input gate, an output gate, and a forget gate. When information enters the unit, it can either be forgotten, inputted to the rest of the cell, or summarized and outputted.</p><p>The <strong><em>forget gate</em></strong> removes information from the cell via the multiplication of a filter (see the sigmoid function above on the far left with the arrow to the multiplication symbol).</p><p>The <strong><em>input gate</em></strong> is how new information can be added to the cell. This happens in three steps. First, the sigmoid function acts as a filter to decide what values need to be added. Next, the tanh function creates a vector of all possible values that could be added. The output vector will be in the range of -1 to 1. Finally, the values from the sigmoid filter and the tanh function are multiplied together and then added to the cell.</p><p>The <strong><em>output gate</em></strong> selects useful information from the cell state and shows it as output. This happens in three steps as well. First, the tanh function is applied to have values between -1 and 1. Again, we will use the sigmoid function as a filter. To complete the step, the vector created by the tanh function is multiplied with the filter and sends it as output to the next hidden cell.</p><p>In conclusion, all these gates make LSTM models much better than your typical convolutional feedforward network or RNN at sequence prediction. This is why we chose to use an LSTM model to generate poems.</p><h3>Our Data</h3><p>The data for training our LSTM model for poem generation came from the complete collection of poems from PoetryFoundation.com. The data can be found <a href="https://www.kaggle.com/tgdivy/poetry-foundation-poems">here on Kaggle</a>. Each poem entry contains several tags that describe the poem. We decided to focus on three categories of poems: Love, Nature, and Sadness. To get the right data, we needed to extract poems from the dataset that have certain tags. We decided to cap the characters at 1000 for each poem (we played around with more characters) because that way we included enough of the poem, and didn’t put too much strain on our computers. In the code below, you can see how we got the data.</p><pre><strong># include 1000 characters from each poem</strong><br>pd.options.display.max_colwidth = 1000 </pre><pre><strong># load dataset from poetryfoundation.com dataset</strong><br>df = pd.read_csv(&#39;PoetryFoundationData.csv&#39;) </pre><pre><strong># dropping any entries with no tags</strong><br>df = df.dropna()</pre><p>Here is an example of what the raw data looks like.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Q1hOx1nL4skPJ4ZNPtX0fA.png" /><figcaption>Raw poem data from Kaggle dataset. Photo by author.</figcaption></figure><p>As you can see in the raw data above, there are several tags. For the purpose of this blog post, we are going to show you only how we extracted poems with the tag “love” (see code below). We also generated poems using other tags which you can see towards the end of this post.</p><pre>poems = df[‘Poem’]</pre><pre><strong># you can add additional tags to searchfor if you want more poems</strong><br>searchfor = [‘Love’]<br>lovePoems = poems.loc[df[‘Tags’].str.contains(‘|’.join(searchfor))] <br>lovePoems = lovePoems.to_string()</pre><h3>Training the Model</h3><p>Once we gathered the poems we needed, we were able to create training and testing data. In the code below, we created our training data. First, we had to map the characters to indices and vice versa.</p><pre>chars = sorted(list(set(lovePoems)))<br>char_indices = dict((c, i) for i, c in enumerate(chars))<br>indices_char = dict((i, c) for i, c in enumerate(chars))</pre><p>Next, we had to create the X and Y data sets for training. This process chops up the giant string of poems into sequences with a length of 50 characters.</p><pre><strong># process the dataset</strong><br>seqlen = 50<br>step = seqlen</pre><pre>data_X = []<br>data_y = []</pre><pre>poemLines = []</pre><pre><strong># creates poem lines </strong><br>for i in range(0, len(lovePoems) — seqlen — 1, step):<br> poemLines.append(lovePoems[i: i + seqlen + 1])<br> <br><strong># creating x and y data</strong><br>data_X = np.zeros((len(poemLines), seqlen, len(chars)), dtype=np.bool)<br>data_Y = np.zeros((len(poemLines), seqlen, len(chars)), dtype=np.bool)</pre><pre>for i, poemLines in enumerate(poemLines):<br> for t, (char_in, char_out) in enumerate(zip(poemLines[:-1], poemLines[1:])):<br> data_X[i, t, char_indices[char_in]] = 1<br> data_Y[i, t, char_indices[char_out]] = 1</pre><p>After generating our data, we built our model. There is a LSTM layer and a Dense layer in its architecture. We will discuss below, but we played around with the number of hidden units for the LSTM layer. In addition, we tried a model with 2 LSTM layers and 3 LSTM layers. Keeping all the variables the same, we did not see a difference in performance of our model when changing the number of layers.</p><pre><strong># create the model</strong><br>model = Sequential()<br>model.add(LSTM(80, input_shape=(seqlen, len(chars)), return_sequences=True))<br>model.add(Dense(len(chars), activation=&#39;softmax&#39;))</pre><pre>model.compile(loss=&#39;categorical_crossentropy&#39;, optimizer=RMSprop(learning_rate=0.01), metrics=[&#39;categorical_crossentropy&#39;, &#39;accuracy&#39;])<br>model.summary()</pre><h3>Evaluating the Model</h3><p>Then, we were ready to fit our model. After training with a batch size of 128 and 10 epochs, we achieved an accuracy of 0.68.</p><pre><strong># evaluate the model on validation set and visualize <br></strong>history = model.fit(data_X, data_Y, validation_split = .33, batch_size=128, epochs=10)</pre><p>We also evaluated the model on a validation dataset and visualized the results. The plot below shows the training and validation loss when we had 128 hidden units for the LSTM layer.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/398/1*x3xnOStUyWhHltNrorY5zw.png" /><figcaption>Visualization of model training and validation loss when LSTM layer had 128 hidden units. Photo by author.</figcaption></figure><p>However, when we fit our model with 80 hidden units for LSTM layer, we saw the plot below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/386/1*5p8ycKL7wGuIO0r5uUyDQg.png" /><figcaption>Visualization of model training and validation loss when LSTM layer had 80 hidden units. Photo by author.</figcaption></figure><p>When comparing the two plots, we can see that the model with 80 hidden units in LSTM layer has a better fit than the model with 128 hidden units. The model with 128 hidden units shows a pattern of overfitting that we read about in our research. Over time, the training and validation data diverge which is not what we are looking for. However, with the 80 hidden units, the training and validation loss both decrease together indicating a good model fit.</p><p>We also realize that our accuracy could have been higher. We tried training our model with 3000 characters from each poem in the dataset. This increases the amount of data we had, and we were able to increase the accuracy of the model to 78%–a 10% increase. If we had more time and more computing power, we could use more of the data. We used a smaller amount of data than what we had access to because it took too long to train.</p><h3>Generating Poems</h3><p>After settling on the architecture of the model, we generated some poems using the code below.</p><pre><strong># helper function that samples an index from probability array<br></strong>def sample(preds, temperature=1.0):<br>    preds = np.asarray(preds).astype(&#39;float64&#39;)<br>    preds = np.exp(np.log(preds) / temperature)  # softmax<br>    preds = preds / np.sum(preds)                <br>    probas = np.random.multinomial(1, preds, 1)  # sample index<br>    return np.argmax(probas)</pre><pre><strong># generate poem<br></strong>def generating_poem(poem_length, start_index, diversity, _):<br>    start_index = start_index<br>    diversity = diversity</pre><pre>generated = &#39;&#39;<br>    poemLines = lovePoems[start_index: start_index + seqlen]<br>    generated += poemLines<br>    print(&#39;----- Generating with seed: &quot;&#39; + poemLines + &#39;&quot;&#39;)<br>    sys.stdout.write(generated)</pre><pre>for i in range(poem_length):<br>        x_pred = np.zeros((1, seqlen, len(chars)))<br>        for t, char in enumerate(poemLines):<br>            x_pred[0, t, char_indices[char]] = 1.<br>            <br>        preds = model.predict(x_pred, verbose=0)<br>        next_index = sample(preds[0, -1], diversity)<br>        next_char = indices_char[next_index]</pre><pre>poemLines = poemLines[1:] + next_char</pre><pre>sys.stdout.write(next_char)<br>        sys.stdout.flush()<br>    print()</pre><p>Below, we generated a poem with a length of 500 characters, a start index (which is a string of the sequence length from the original text), and diversity of 0.5. The diversity input changes the sampling probabilities. After a few trials, we found that a diversity of 0.5 gave us the best poems. A diversity of 0.1 made the poems too repetitive and a diversity of 1 made poems with little sense.</p><pre><strong># input desired poem length, start index, diversity</strong><br>generating_poem(500, 60, .5, _) </pre><p>Here is the output from a run of the code above:</p><pre>----- Generating with seed: &quot;n\r\r\n\r\r\n\r\r\nAnd it won’t be multiple choice&quot;<br>n\r\r\n\r\r\n\r\r\nAnd it won’t be multiple choiced in the sky love and made the back between the forest and the from the black and sky feet promised which her desire. I could come before the point to the collation.\r\r\nI am a dear the shadows and which stare of the father, the cheek and boxasted seeman that the confecting the can sweating while I moved me can don’t know you want to still watch the stater and we can see a poem who deep of bene to the didn’t were been on the breaks dark in the tongue and blood from the color in the morning on t</pre><p>As you can see the output includes new line indicators (\r\r\n). We played around with removing these from our training string, but we decided that keeping these characters allows our model to structure our generated poems like real poems. Below, we show a few more poems that we generated, formatted nicely.</p><p><strong>Love Poem #1:</strong></p><pre>Roy and Glen</pre><pre>lost themselves</pre><pre>is the world the prides or story in the starting the discapive to the stranged start us the wind. The down the stucky strange and the last the sky</pre><pre>each an all the spon and carry me heart and set, and the bright have so sand for the stranged of the stranger. I see we between my leaves stranged the skies in the annt, sweet of at the same shade to heart, and the can don’t grape in not so self</pre><p><strong>Love Poem #2:</strong></p><pre>The woman carried the bucket</pre><pre>and the man with sing are fingers the soft in this do, and of the conficial from the strangering to my slight which the sky stead between the body of a sation under there is the spring became and prices air strange of me to mouth have at his blossoms the sands in the trangs of the stared a paration in the man spring of carvant</pre><p><strong>Sad Poem #1:</strong></p><pre>Among these in this place my father lay</pre><pre>At the end of everything</pre><pre>In the curked the pather she have years with the single to the mander,</pre><pre>as mance of the should the wordder the stear the procked the goor in hone the searther the marmer.</pre><pre>I was a sinder the pourth of the say and word the same the see we life, geser and he was down the all dading morned with the stead of the sleet on the surion, not liver, a mone the mardens colling to me see she seeple of the sead of the becausion the spean.</pre><pre>Lone the bood the swear the looked wor a man her beants of the rom</pre><p><strong>Sad Poem #2:</strong></p><pre>tooed with red…</pre><pre>Imagine half your face</pre><pre>rubbed out yet</pre><pre>you are suited up</pre><pre>where were souned a reared semerit</pre><pre>a manding words and the did read to mister the flown the stone are where dristed were the san land a come was the son like word the light of the wise to belend</pre><pre>and she teather, you been a die and lacken the wanted from the cores and the mand of Comples, never a come your mone the mone of the she she park of hall that like the sard of the she was the pleared in the say back the meed the say with to be the ray.</pre><pre>A dried be sun to carr take in th</pre><p><strong>Nature Poem #1:</strong></p><pre>Soon the fish will learn to walk. Then humans will come ashore and paint dreams on the wall distant when the rain she who water the world of mean the flower </pre><pre>the flowers the flacks the ancent of my walls and shadows the river and stars them like the sung and flue</pre><pre>the sea when the bodge his not behind the streets of the remement and she to the sings in the screen, the name the stars of the days on the the sounds the dark skin the mother the same and this pity, the clouds have a state is sings and someone land, and at my shell stars</pre><pre>as the little still make a way t</pre><p><strong>Nature Poem #2:</strong></p><pre>e; it is obvious) and everything about him is beautiful. His hand swells from the bite [spread?] of the sky in the more the will condine her the distant green the forms, the wind my bear the side the roots where a merch her cold of high a constiled in the charged the body the stars in the wind sauter be. The cold with the world from the remember, and he stream and sting for the season the men rain</pre><pre>a wood with the was in the silence and men the world it roses where the cattent the capt and seas the dark and still as the grass</pre><pre>where he the stars and past of the the color the beats head</pre><h3>Comments on Generated Poems</h3><p>Overall, our poem output seems to be generating poems with a format similar to our dataset. They have similar amounts of “\r\r\n” like sections to represent the spacing, and they are pretty good at generating sentences. The general mood of each category seems to reflect the data that it came from as well. However, they still struggle to make logical sense and are highly based on the starting index. The starting index plays a large role in determining how the poem turns out. The words in the starting seed are often repeated throughout the poem.</p><p>In the future to improve the output of these poems, we could have the poems indexed so the starting seed starts at the beginning of a poem each time rather than the middle or end of the poem. We could do work to have it train on one full poem so it doesn’t just start and end randomly–creating a more concise poem.</p><p>The logical sense of each sentence is another thing to improve. However, this is a great challenge in poetry because poems are artistic and don’t necessarily follow traditional grammar rules. We could do more research into how to make poems more coherent as well.</p><h3>Reflections</h3><p>When we started the project we originally intended to use a generative adversarial network (GAN) where the generator was a recurrent neural network (RNN). After doing some research, we found <a href="https://becominghuman.ai/generative-adversarial-networks-for-text-generation-part-1-2b886c8cab10">several articles</a> that mentioned difficulty and less accuracy when using the GAN on text formats. We also saw that an LSTM model was recommended so we switched to that instead. It would have been nice to use a GAN because we would have been able to combine the latent space of data instead of making a new model for each type of poem.</p><p>Another thing we could have tried was separating the poem by time period. The poems written by Shakespeare have a very different language than modern poems and that could make for a confusing data set for the model to generate from. If we are to add onto this project we could separate by time period so the user could choose the style they want their poem to come out as.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*nGmyPVyB_zeHwEy6" /><figcaption>Photo by <a href="https://unsplash.com/@thoughtcatalog?utm_source=medium&amp;utm_medium=referral">Thought Catalog</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Another feature that could be added to future work would be poem titles. If we had more time, we could have formatted the data to also take in the poem title along with the poem content. We would then generate a title based on the tags that would match each generated poem. However, this may lead to the titles being unrelated to the actual poem content.</p><p>Overall, we realize there are many things we could try improve the model’s poem generation. However, we made a good start at poem generation and have learned a lot about how LSTM models work and how to tweak them.</p><h3>References</h3><ul><li><a href="https://www.analyticsvidhya.com/blog/2017/12/fundamentals-of-deep-learning-introduction-to-lstm/">Long Short Term Memory | Architecture Of LSTM</a></li><li><a href="https://towardsdatascience.com/ai-generates-taylor-swifts-song-lyrics-6fd92a03ef7e">AI Generates Taylor Swift’s Song Lyrics</a></li><li><a href="https://machinelearningmastery.com/diagnose-overfitting-underfitting-lstm-models/">How to Diagnose Overfitting and Underfitting of LSTM Models</a></li><li><a href="https://becominghuman.ai/generative-adversarial-networks-for-text-generation-part-1-2b886c8cab10">Generative Adversarial Networks for Text Generation — Part 1</a></li><li><a href="https://medium.com/cindicator/music-generation-with-neural-networks-gan-of-the-week-b66d01e28200">Music generation with Neural Networks — GAN of the week</a></li></ul><p><em>Project completed by Emma Sheridan and Jessica Petersen for Artificial Neural Networks course. The source code for this project can be found on </em><a href="https://github.com/emmasheridan/ANNfinalproject/"><em>GitHub</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b28b818c2f24" width="1" height="1" alt=""><hr><p><a href="https://medium.com/data-science/looking-for-the-perfect-words-generate-them-b28b818c2f24">Looking for the perfect words? Generate them.</a> was originally published in <a href="https://medium.com/data-science">TDS Archive</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>