<?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 Katherine Olvera on Medium]]></title>
        <description><![CDATA[Stories by Katherine Olvera on Medium]]></description>
        <link>https://medium.com/@kaolvera?source=rss-527573d8209c------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*29PTR590detbp-kcZdsLgA.jpeg</url>
            <title>Stories by Katherine Olvera on Medium</title>
            <link>https://medium.com/@kaolvera?source=rss-527573d8209c------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 08:47:45 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@kaolvera/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[A Human in a Machine World]]></title>
            <link>https://medium.com/@kaolvera/a-human-in-a-machine-world-e8f2166507f?source=rss-527573d8209c------2</link>
            <guid isPermaLink="false">https://medium.com/p/e8f2166507f</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[chatgpt]]></category>
            <category><![CDATA[art]]></category>
            <category><![CDATA[ethics-in-tech]]></category>
            <category><![CDATA[midjourney]]></category>
            <dc:creator><![CDATA[Katherine Olvera]]></dc:creator>
            <pubDate>Tue, 10 Jan 2023 06:34:22 GMT</pubDate>
            <atom:updated>2023-01-10T15:20:24.405Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EtmtXytkgSN6_jZagZH2UQ.png" /><figcaption>4 variations from the prompt “Woman in a Machine World” using Midjourney</figcaption></figure><p><em>Will AI be a deus ex machina or a Pandora’s box?</em></p><p>I’ve had an idea for a few years now about creating a series of photographs documenting Texas homecoming mums in the style of Dutch masters. If you aren’t familiar with <a href="https://www.google.com/search?q=texas+homecoming+mums&amp;sxsrf=ALiCzsbefmBVRsSVwI5RVw7w_F2US4DO9w:1672888154176&amp;source=lnms&amp;tbm=isch&amp;sa=X&amp;ved=2ahUKEwjOvbaBuq_8AhUvmmoFHYrqBCUQ_AUoAXoECAEQAw&amp;cshid=1672888175828557&amp;biw=1500&amp;bih=794&amp;dpr=2">Texas homecoming mums</a>, they are these elaborate and gaudy accessories that high schoolers in Texas exchange with their dates for the homecoming game. They have teddy bears, doodads, ribbons, streamers, glitter, literally bells and whistles. I thought it would be interesting to elevate a kitschy piece of ephemera from one weird corner of Texas culture by capturing it in the style of the Dutch still-life.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uc-Wf43mc0M6aTeUhOo_4A.jpeg" /><figcaption>Drill team members wear mums for a homecoming football game in Texas from a story in <a href="https://www.dmagazine.com/publications/d-magazine/2014/october/homecoming-mums-in-bloom/">D Magazine</a>, 2014.</figcaption></figure><p>Anyway, when I heard about Dall-E and Midjourney, I was excited. These are AI tools that can generate images from written prompts. Maybe these AIs could finally visualize this idea I’d had in my head for so long! I tried several prompts:</p><ol><li><em>Ooh, this is exciting, I can just type in what I want to see — </em>“Texas Homecoming mum in the style of Dutch masters still lives” — <em>ehhh, not quite, looks like a regular Dutch still life</em></li><li><em>maybe I should describe what a mum is? — </em>“fake mum with teddy bear in the middle and with glittery streamers, ribbon, cowbells, and whistles dangling from it in the style of Dutch masters still life”<em> — closer, but it isn’t really putting all the pieces together into a whole mum</em></li><li><em>does it even understand what a mum is??? — </em>“Texas homecoming mum” — <em>so, it’s capturing the spirit of a mum, which I didn’t realize was possible, but it’s still not generating an image of the object I want it to render. It keeps on making bouquets.</em></li><li><em>give it another shot? — </em>“Texas homecoming mum in the style of the Dutch masters” — <em>interesting images, but not really what I had in mind</em></li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KMIqY9L5JAo10yPjEShItA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Vu0MgrkHLlT5wwGeBLIFFw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WzV-0tymF-sqYuhcRtFapw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*glgYnGSX1ILJ2EQJBC55hg.png" /></figure><p>Alas, their attempts were lacking. The images were fascinating, but they didn’t really seem to understand what a Texas homecoming mum was. However, it did prompt me to ask a question: what are the limits and shortcomings of AI when everyone seems so excited about its potential?</p><h3>The Limitations of AI</h3><h4>How do I trust the material an AI is trained on?</h4><p>The first limitation that I noticed was that these AIs seem to be trained on mainstream culture. I noticed this because the particular thing I happened to want it to render was this random trivial object I remembered from my childhood growing up in Texas. Homecoming mums aren’t mainstream. Most people outside of Texas haven’t heard of them.</p><p>And this led me to wonder what happens if your reference falls outside of the mainstream (read: White, Western) culture? An interesting feature of Midjourney right now is that it is hosted on a Discord server. This means that it works like a big chat room, and users can see both other users’ inputs and the outputs of the AI. From my limited and admittedly anecdotal experience, using prompts like “man” “woman” “boy” or “girl” without additional descriptors tends to yield white people in images.</p><p>So, what happens if you have a niche experience or an experience that hasn’t been documented since time immemorial in books, museums, magazines, photographs, movies, music, videogames or blogs? And what if your culture’s story isn’t told by practitioners, but by observers? Where will this leave diverse experiences?</p><p>The future consequence of using homogenous material to train the AI could go in two possible directions (or many directions in between):</p><ol><li>Diverse experience suddenly becomes more valuable because it is rare and because the AI cannot replace or replicate these experiences. Humans with these experiences will be needed to train the AI on their unique, embodied knowledge.</li><li>Diverse experience becomes invisible through lack of representation. The AI is seen as a source of authority, and when you aren’t represented, your experience becomes more and more obscure as the AIs continue to flood the world with mainstream content.</li></ol><p>That said, one thing that does give me hope is that, while AI’s training material might not be representative of all experience, it can be harnessed to generate art and images to fill in gaps in representation in our current media. I’ve been excited to see <a href="https://www.instagram.com/p/CmmKD5lusY7/">Heather Abbott</a> using Midjourney to create Afrofuturistic imagery and art.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aPNhzQIwSZ6VPqoljjRWfg.jpeg" /><figcaption>One of Heather Abbott’s images of Afrofuturistic children.</figcaption></figure><p>Unlike many artists who seem fearful about this new technology, Heather seems curious and excited about AI as a tool. They first started to notice AI-generated imagery popping up on Facebook, and eventually were able to track down the tool people were using — Midjourney. After signing up for Discord, they started playing around with prompts. “Black” didn’t always work as expected: AI interpreted it as a color instead of a descriptor of a person’s race. “Afrofuturism” yielded results with afros. But after some trial and error, they settled on some terms that generated interesting new images. I was struck by how much Heather’s descriptions of interacting with Midjourney felt like play, a conversation, or improvisation:</p><blockquote>I imagine something and when I see the four options it gives me, it’s not what I pictured: it creates worlds that I couldn’t have imagined.</blockquote><p>There are limitations. The AI hasn’t quite figured out hands and feet.</p><h3>Sofiia Shvets 🇺🇦 on Twitter: &quot;My AI startup will replace designers job.AI accepting the job: pic.twitter.com/5oEVF04go4 / Twitter&quot;</h3><p>My AI startup will replace designers job.AI accepting the job: pic.twitter.com/5oEVF04go4</p><p>And sometimes, it feels like you can start to pick out the specific references it’s using when there is limited visual material to pull from. But Heather notes it’s improving. What is less clear though is whether it’s learning from the artists and people that are interacting with it. If it is, Heather wants to be sure that it’s learning from people like them about cultures and material that aren’t part of the media canon.</p><blockquote>I just want to show the world what Black people would look like in the future and that they belong in the future too.</blockquote><p>When your past isn’t well-represented in traditional media, maybe you can use an AI to create a new future.</p><h4>Do AIs create circular logic?</h4><p>While we can’t know all the material that a particular AI is trained on, we know something about the material: it was made by humans. And one thing we know about humans is that they make errors, sometimes beautiful ones, but errors none the less.</p><p>Errors in our information systems have a way of becoming sticky. I read about an example of the stickiness of errors in James Gleik’s <a href="https://www.amazon.com/Information-History-Theory-Flood-ebook/dp/B004DEPHUC/ref=tmm_kin_swatch_0?_encoding=UTF8&amp;qid=1672879422&amp;sr=8-1">The Information: A History, A Theory, A Flood</a>. Prior to calculators and computers, merchants, bankers, and navigators referred to logarithm tables to make difficult math calculations. However, they had to trust that the table they were referring to was accurate. And they sometimes contained errors, like typos from the printer. These miscalculations could lead to major real-world problems, including ship-wrecks. And since many of these tables were copied from one edition to the next, sometimes without correction, the errors stuck.</p><blockquote>Unfortunately, one of the nineteen corrections proved erroneous, so the next year’s Nautical Almanac printed an “erratum of the errata.” This in turn introduced yet another error. “Confusion is worse confounded,” declared The Edinburgh Review . The next almanac would have to put forth an “Erratum of the Erratum of the Errata in Taylor’s Logarithms.<br>-The Information: A History, A Theory, A Flood</blockquote><p>When it becomes faster and easier to generate content because AIs reduce friction to writing with an authoritative tone and creating realistic images, will that mean that our errors solidify, becoming harder to fix because they proliferate so freely and quickly throughout our information system?</p><p>Say AIs were invented in 1500, before the time of Copernicus. Would we forever be condemned to tools that told us that the sun revolved around the earth? Or would we be able to introduce novel discoveries, paradigm-shifts, in some way to the system? Would singular and small human voices still be able to share important ideas that change our collective thought, or would our information become frozen in an arbitrary point in time?</p><h4>What effect will AIs have on our information landscape?</h4><p>AIs are extremely good at finding and recognizing patterns, patterns that humans may not even notice. Sometimes the patterns highlight an important but underappreciated nuance, but they may also be insignificant, irrelevant, or just <a href="https://www.tylervigen.com/spurious-correlations">coincidental</a>. The AI can’t tell the difference. But pattern, or correlation, doesn’t equate to accuracy or truth.</p><p>I appreciate that ChatGPT comes with a disclaimer:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*x34O5Ap4rzDvhm85ofYlJA.png" /><figcaption>Prominent disclaimer shown while using ChatGPT</figcaption></figure><p>However, that disclaimer doesn’t follow the content it creates everywhere it goes on the internet. <a href="https://www.nytimes.com/2018/03/08/opinion/sunday/truth-lies-spread-online.html?action=click&amp;module=RelatedCoverage&amp;pgtype=Article&amp;region=Footer">We’re already seeing the damage wrought by bad actors using social media on our information landscape.</a> With a tool that can generate such truthy text at such a quick pace, how will this impact our ability to find sources and facts amongst the torrent of content? What new tools will we need to be able to guard against disinformation?</p><p>Finally, there’s a philosophical question to wrestle with.</p><h4>Do AIs create?</h4><p>Creative work requires an idea, an input. In the scenario about the Dutch homecoming mum that I mentioned at the beginning, I wrote the prompt. I was the person manipulating the material of the AI. The AI didn’t come up with that idea on its own.</p><p>Sol Lewitt explored the nature of creativity, authorship, and production in his work.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NdZxbmyyKsjEugc0zWNQIQ.jpeg" /><figcaption>Sol LeWitt, <em>Wall Drawing #368</em>, 1982/2016, color inks and india ink, dimensions variable. Installation view. Photo: Steven Probert. © Estate of Sol LeWitt/Artists Rights Society (ARS), New York.</figcaption></figure><blockquote>LeWitt is perhaps best known for his “wall drawings,” which he conceived as large-scale line drawings that could be created by anyone following a set of written instructions. These instructions, which LeWitt referred to as “concepts,” were often quite simple and straightforward, yet they could be used to create highly complex and varied works of art.</blockquote><blockquote>LeWitt’s philosophy of art was based on the idea that the concept behind a work of art was more important than the finished product. He believed that the process of creating a work of art was as important as the finished product, and that the viewer’s experience and interpretation of the work was just as important as the artist’s intention.</blockquote><blockquote><em>-</em>ChatGPT, Can you explain Sol LeWitt’s philosophy of art?</blockquote><p>But we also have to consider the authorship and possession of the material of the AI. The training material that is.</p><p><a href="https://www.nytimes.com/2022/12/31/opinion/sarah-andersen-how-algorithim-took-my-work.html">We have never before been able to so perfectly and quickly generate images in another artist’s style</a>. Sure there has been collage, remixing, appropriating, and copying. But this feels somehow new. Now, with enough training material from any single artist, living or dead, you could use an AI to generate content to compete directly with that artist for work without having invested the time and labor in developing that style yourself. Who is entitled to the collective creative labor of all of humanity to train an algorithm? And who then is entitled to its output?</p><p>This is where a lot of the questions, concerns, and anxieties about the future and AI seem to stem from. What does AI mean for the future of creativity and knowledge work? People are mentioning that an AI could write <a href="https://www.reuters.com/legal/transactional/will-chatgpt-make-lawyers-obsolete-hint-be-afraid-2022-12-09/">legal contracts,</a> technical documentation, Excel formulas, software code, <a href="https://www.theatlantic.com/technology/archive/2022/12/openai-chatgpt-writing-high-school-english-essay/672412/">high school english papers</a>, dating app replies, customer service problem tweets, either now, or in the not so distant future.</p><p>And yes, that’s all possible as an output, but, as of right now at least, it still needs an operator. It still needs someone to evaluate the outputs for accuracy and tweak it until it’s correct. And for many “knowledge work” and creative jobs, what people are actually paying for isn’t a deliverable, it’s an outcome and a person to hold responsible when things don’t go right.</p><p>In my estimation, AIs today don’t create. They respond. They don’t evaluate, they iterate. It’s a tool. An incredibly powerful, novel, surprising, and inspiring one, but a tool nonetheless.</p><p>AI could do for today what photography did for painting. When photography came along and was able to so quickly and accurately capture an image, painting was freed to become even more expressive as a medium. Painting was used to be a direct way to capturing the result of the artists movements, their emotions, their feelings, and their inner psychology, everything besides the image or representation of the thing that a photograph would do just as well for. Maybe AI will do the same for some of our current information and content mediums.</p><p>In that way, it could free us of some of the minutiae of creative production work. It could allow us to focus more on the interesting creative work, the evaluation, critique, and ideas. Or to focus again on physical (real world) craft, which the AI can’t do. And it can help us more quickly communicate visual ideas or approximations of visual ideas to each other quickly.</p><p>But I hope we don’t do this completely at the expense of other artists, the ones who have labored in creating the material to train the AI. If we do, we will lack new material to move our creative culture forward. We must find a way to properly value <em>both</em> the input and output, even if the input becomes more obscured.</p><p><strong>Best case scenario:</strong> AIs take over the jobs of both customer service and arguing with customer service, leaving us humans free to create and enjoy our lives in the real world.</p><h3>What does the future hold?</h3><p>I approach AIs, machine learning, and algorithms with a mixture of caution and curiosity. An AI is a black box. It’s trained on more data than we could comprehend. It also means we can’t know or audit the training material for accuracy or representation.</p><p>The patterns it finds have the power to be used for incredible good (<a href="https://www.cancer.gov/news-events/cancer-currents-blog/2022/artificial-intelligence-cancer-imaging#:~:text=Scientists%20have%20developed%20AI%20tools,this%20area%20is%20quickly%20evolving">helping doctors screen for breast cancer</a>) or horrifying evil (<a href="https://radiolab.org/episodes/40000-recipes-murder">discovering new and terrifying chemicals that could be used in biological warfare</a>). And they can reinforce biases that already exist in <a href="https://theintercept.com/2021/01/30/lapd-palantir-data-driven-policing/">policing</a>, <a href="https://twitter.com/dhh/status/1192540900393705474?lang=en">banking</a>, and a plethora of other areas, including who is represented in media and art, magnifying errors made long ago in an obscure, powerful, and authoritative tool that doesn’t answer to anyone. It obscures responsibility.</p><p>My fear is that, like with many other AI tools that have recently been unleashed upon us, an unwitting public, by smart but unthinking tech bros, the power of ChatGPT, Midjourney, Dall-E, and others to do extraordinary harm hasn’t been considered. They don’t think about consequences. What happens when your face is used to create revenge porn with Midjourney? Or is used to endorse something you’d never endorse? What happens when ChatGPT creates a recipe for a home cleaning solution that could kill you? Or writes a news story with completely artificial sources? Where and how does one seek redress?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*DfBAqt-QMWzxFTNeEDcJng.jpeg" /><figcaption>The movie Desk Set (1957) was prescient in considering the tensions and problems humans would face as technology encroached further into our information space: <em>“Oh, yes, we’ve looked that up for you, and there are certain poisons which leave no trace, but it’s network policy not to mention them on our programs.”</em></figcaption></figure><p>The current technologists claim that they are <a href="https://openai.com/about/">trying to do good for mankind</a>, but based on the track record of big tech, we shouldn’t take them at their word. We need to see action. We need to see that they’ve actually invested in and thought about how to protect vulnerable populations, and that they’ve given just as much thought and care to those gnarly problems as they have to these cool new gadgets. If you can come up with AI, don’t tell me you can’t come up with a better way to keep people from being violated in new ways each day on the internet.</p><p>I hope that as these tools become more prevalent that we will employ people of all backgrounds, ethicists, integrity teams, social scientists, anthropologists, and smart regulations (if they can keep up) to ensure that the future we create is better than the one we exist in today. And to ensure that the machines serve humanity instead of the other way around.</p><p><strong>More Reading and Listening about AI &amp; Our Future</strong></p><ul><li><a href="https://www.thisamericanlife.org/757/the-ghost-in-the-machine">The Ghost in the Machine</a> <em>This American Life</em> — a writer recounts the experience of writing a difficult personal narrative story with an AI</li><li><a href="https://podcasts.apple.com/us/podcast/hard-fork/id1528594034">Hard Fork (Podcast</a>) <em>NYTimes</em></li><li><a href="https://podcasts.apple.com/us/podcast/on-with-kara-swisher/id1643307527">On (Podcast</a>) <em>Kara Swisher</em></li><li><a href="https://podcasts.apple.com/us/podcast/the-ezra-klein-show/id1548604447">The Ezra Klein Show (Podcast)</a> <em>NYTimes</em></li><li><a href="https://www.nytimes.com/spotlight/artificial-intelligence">Artificial Intelligence: A Special Report</a> <em>NYTimes</em></li><li><a href="https://www.nytimes.com/2023/01/06/science/robots-artificial-intelligence-consciousness.html">‘Consciousness’ in Robots Was Once Taboo. Now It’s the Last Word.</a> <em>NYTimes</em></li><li><a href="https://time.com/6132399/timnit-gebru-ai-google/">Why Timnit Gebru Isn’t Waiting for Big Tech to Fix AI’s Problems</a> <em>Time Magazine</em></li><li><a href="https://thebaffler.com/salvos/sentience-and-sensibility-ogieblyn">Sentience and Sensibility</a> <em>The Baffler</em></li></ul><p>Collections on Artificial Intelligence:</p><ul><li><a href="https://www.theatlantic.com/category/ai-artificial-intelligence/"><em>The Atlantic</em></a></li><li><a href="https://www.nytimes.com/topic/subject/artificial-intelligence"><em>NYTimes</em></a></li><li><a href="https://www.wired.com/tag/artificial-intelligence/"><em>Wired</em></a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e8f2166507f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[A Bone to Pick with Skeleton Screens]]></title>
            <link>https://medium.com/viget-collection/a-bone-to-pick-with-skeleton-screens-eb006536ca1c?source=rss-527573d8209c------2</link>
            <guid isPermaLink="false">https://medium.com/p/eb006536ca1c</guid>
            <category><![CDATA[ux]]></category>
            <category><![CDATA[user-research]]></category>
            <category><![CDATA[user-experience]]></category>
            <category><![CDATA[website-design]]></category>
            <dc:creator><![CDATA[Katherine Olvera]]></dc:creator>
            <pubDate>Thu, 19 Oct 2017 20:51:14 GMT</pubDate>
            <atom:updated>2017-10-19T22:06:26.184Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-OKdG49z-OT6nCaYNQcpAg.png" /></figure><p>by Kathryn Faulkner &amp; Katherine Olvera</p><p><em>Facebook and Google use skeleton screens to make their apps feel faster. Should you be using them too?</em></p><p>In the fight for the short attention span of our users, every performance gain, whether real or perceived, matters. This is especially true on mobile, where despite our best efforts at performance, a spotty signal can leave users waiting an interminable few seconds (or more) for content to load.</p><p>Design’s conventional answer to unpredictable wait times has long been the loading spinner; a looping animation that tells the user to “Hold on. Something’s coming,” whether that something is one or ten seconds away.</p><p>More recently, a design pattern known as progressive loading has gained popularity. With progressive loading, individual elements become visible on the page as soon as they’ve loaded, rather than displaying all at once. See the following example from Facebook:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/388/0*93UHgWimut08Ig3N." /><figcaption>Progressive loading on the Facebook app</figcaption></figure><p>In the Facebook example above, a skeleton of the page loads first. It’s essentially a wireframe of the page with placeholder boxes for text and images.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/377/1*o2u-5_Uon_6_QNLy8_kgQw.png" /><figcaption>Facebook’s skeleton screen</figcaption></figure><p>Progressive loading with skeleton screens is thought to <a href="https://www.lukew.com/ff/entry.asp?1797">benefit the user</a> by indicating that progress is being made, thereby shortening the perceived wait time. Google, Medium, and Slack all use skeleton screens to make their apps feel more performant.</p><p>So, should we all be using skeleton screens to make our apps feel faster? To answer this question, we decided to do some lean research into the effects of different loading techniques on perceived wait time.</p><h3>Test Design</h3><p>We created <a href="https://viget.optimalworkshop.com/chalkmark/meal-delivery">a short test</a> for mobile devices that measured users’ perceived wait time for three different loading animations of identical length: a loading spinner, a skeleton screen, and a blank screen.</p><p>The quickest way to build the test was to use animated GIFs to simulate each loading animation and put them inside an existing testing framework. (We chose Chalkmark by Optimal Workshop.) Users who opted into the test from a mobile device were asked to complete a simple task and then randomly shown one of the GIFs. Following the task, which was a red herring, they were asked a series of follow-up questions about how long they waited for the page to load.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/361/0*Ud7hkrgNxpTni5J4." /><figcaption>The skeleton screen variant in the test we deployed</figcaption></figure><p><em>Follow-up questions:</em></p><ol><li>Based just on what you can recall, please respond to the following statement: “The recipes loaded quickly for me.” [Strongly agree, Moderately agree, Neutral, Moderately disagree, Strongly disagree, I didn’t notice]</li><li>From what you can remember, estimate the amount of time it took for the meals to load. [1 second, 2 seconds, 3 seconds, 4 seconds, 5 seconds]</li></ol><p>We also measured the time it took users in each group to complete the red herring task. Based on some of the literature we’d read, it seemed plausible that a skeleton loader might actually speed up task completion by orienting users more quickly to the structure of the page.</p><p>Roughly half (70) of the participants were sourced through Amazon Mechanical Turk and paid for their participation. The rest were organically sourced through Viget’s social channels. The results were replicated across both groups of participants.</p><h3>Hypotheses</h3><ol><li>Users in the skeleton screen group will perceive the shortest wait time.</li><li>Users in the skeleton screen group will complete the task most quickly.</li></ol><h3>Results</h3><p>We gave the test to 136 people, and the skeleton screen performed the worst by all metrics. Users in the skeleton screen group took longer to complete the task, were more likely to evaluate their wait time negatively (by answering the first question with “Strongly disagree” or “Moderately disagree”), and guessed that the wait time had been longer than users who saw the loading spinner or a blank screen.</p><h4>Table 1. Test Results</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/973/1*0Wurzk5y7h8kL7nRIrUwqw.png" /><figcaption>This table shows how participants responded, on average, to each of the three simulated loading animations.</figcaption></figure><p>Participants in the loading spinner group were most likely to evaluate their wait time positively (by answering the first question with “Strongly agree” or “Moderately agree”) and had the shortest average perceived wait time.</p><h3>Analysis</h3><p>The unexpectedly weak performance of the skeleton screen may be due to one or more of the following reasons:</p><ol><li>Skeleton screens are somewhat novel and attract more attention than the ubiquitous loading spinner.</li><li>Skeleton screens work better in familiar interfaces and can be off-putting in new settings when users don’t know what to expect.</li><li>Skeleton screens work best when wait times are very short.</li></ol><p>Our hunch is that each of these reasons has some merit, but more testing is needed to know for certain. Either way, skeleton screens aren’t a silver bullet for increasing perceived performance and should be used thoughtfully.</p><p>Have you implemented or experienced skeleton screens in the wild? We’d love to hear your thoughts. Please leave us a comment.</p><p><em>Originally published at </em><a href="https://www.viget.com/articles/a-bone-to-pick-with-skeleton-screens"><em>www.viget.com</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=eb006536ca1c" width="1" height="1" alt=""><hr><p><a href="https://medium.com/viget-collection/a-bone-to-pick-with-skeleton-screens-eb006536ca1c">A Bone to Pick with Skeleton Screens</a> was originally published in <a href="https://medium.com/viget-collection">Inspire + Advance</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What UX Designers Can Learn From IKEA]]></title>
            <link>https://medium.com/viget-collection/what-ux-designers-can-learn-from-ikea-68fb068796b2?source=rss-527573d8209c------2</link>
            <guid isPermaLink="false">https://medium.com/p/68fb068796b2</guid>
            <category><![CDATA[interaction-design]]></category>
            <category><![CDATA[onboarding]]></category>
            <category><![CDATA[ux-design]]></category>
            <category><![CDATA[ux]]></category>
            <dc:creator><![CDATA[Katherine Olvera]]></dc:creator>
            <pubDate>Tue, 13 Jun 2017 19:09:26 GMT</pubDate>
            <atom:updated>2017-06-13T20:17:00.778Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RW9suLSZpGqZYCRPRIJ_ZQ.png" /></figure><p>I recently moved and ended up buying a lot of IKEA furniture. While assembling the different pieces, I began to notice how IKEA devises their instructions to gently lead builders through complex tasks.</p><p>One of the customer’s first interactions with an IKEA product will be to build it, and this experience will likely shape the customer’s lasting impression of both that piece of furniture and IKEA as a brand. This is a <a href="https://www.viget.com/articles/user-memory-design-how-to-design-for-experiences-that-last">high-stakes interaction</a>, and there are so many places where it could go so wrong. Poorly-written instructions may very well end with the customer either throwing up their hands or throwing a hammer at the piece in frustration.</p><p>IKEA understands this and carefully crafts their instructions to guard against rage-fueled furniture destruction. They are written (or drawn really) to guide a novice through the complex process of creating functional furniture out of a pile of otherwise inscrutable panels and hardware. But they don’t just instruct. They are devised to positively shape the assembly experience, hopefully culminating in a feeling of success or accomplishment. This is very similar to an <a href="http://www.useronboard.com/how-slack-onboards-new-users/">onboarding process</a>, and we, as UX designers, can borrow some of the strategies that IKEA models in their instructions to guide our approach to onboarding design.</p><h3>Set an Expectation</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/360/1*O38lYrIIxC--L0-aKEN90g.png" /></figure><p>Every set of instructions starts with a bold picture of the finished product. Here they are making a promise and keeping the customer aware of the end goal: if you follow these instructions, this will be the result.</p><p><strong>When you are onboarding a new customer, concisely restate the purpose of the product. </strong>Demonstrate what a customer can expect to gain by setting out on the sometimes frustrating journey of learning a new app. That promise will help keep your new users motivated.</p><h3>Prepare Your Users</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/269/1*jqg9jdpqc_YUGNKTPig2TQ.png" /><figcaption><em>Instructions all begin the same way with a list of tools you’ll need, a note to protect your working surface, a suggestion to enlist a friend to lend a hand (and possibly moral support), and a reminder to call with any questions.</em></figcaption></figure><p>Preparatory notes offer reassurance. They are also intended to prevent some of the interruptions or frustrations that could slow a customer’s momentum or sour the entire experience. If you know up-front that you will need two types of screwdrivers, you won’t need to break away to dig up those tools later. Proper priming keeps the customer’s attention focused.</p><p>In an onboarding process, there are similar considerations. Is there 2-step authentication where a user will need to have their phone handy? Or, will a user need content to test an email-building application?<strong>First, consider whether these preparatory steps are really necessary.</strong> (Sometimes you can redesign the onboarding process to smooth potential impediments, like providing dummy content to play around with.) <strong>Then, give your users a heads up.</strong> This will help ensure that they continue to progress through the onboarding flow.</p><h3>Connect All the Dots</h3><p>In the past, I’ve encountered instructions that felt either incomplete or overly complex. Multiple steps were combined into one, or transitional steps were missing. This seemed to be in an effort to keep the instruction pamphlet brief. When I began to doubt my understanding of the instructions though, I would hesitate. Why? Because I would hate to realize 15 steps later that I had put a piece together backwards and now have to backtrack to fix it. If a user grows unsure or confused about too many steps, they too will be hesitant to proceed, hindering progress.</p><p>IKEA, on the other hand, doesn’t shy from length. They favor clarity over brevity. While the overall task of putting together a piece of IKEA furniture can be complicated, each step is broken down into a discrete action and clearly and simply articulated. I felt confident after each step that I was putting the pieces together properly.</p><p><strong>Break down your tasks into discrete actions, make fluid connections between these actions, and reveal the appropriate amount of detail to assure your users that they are making progress. </strong>Giving too much detail can be overwhelming, while a lack of detail can be confusing.</p><h3>Anticipate Problems</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/308/1*9QeNRNvGrHaQ5u3cyzb9aQ.png" /><figcaption><em>“Should I be using this peg or that one? Oh, I see, the slightly smaller peg for this step.”</em></figcaption></figure><p>I also noticed how often IKEA anticipated my questions. They anticipated confusion and provided the appropriate guidance. They knew their product through the lens of the uninitiated.</p><p>While this may seem obvious, it is often hard to do. Features and terminology can seem self-evident to the designers who are working on a project day in and day out. But, they may be rather opaque for a new user. <strong>Test the onboarding process with new users to ensure your language and guidance makes sense and that there is appropriate explanation for more complex features or actions.</strong></p><h3>Acknowledge Incremental Success</h3><p>IKEA purposely designs their instructions to make progress apparent to the customer. While putting together some drawers, I realized it would probably be faster to screw in all the screws to all the fronts at once, instead of creating a full drawer, and then another. However, this isn’t how the instructions are written. IKEA would prefer the customer proceed a little slower and use clear visual progress to prod them forward. Recognizable pieces (like a drawer or a headboard) materialize and, then those pieces are assembled into the whole. Each large task is broken down into smaller accomplishments.</p><p>IKEA also front-loads the assembly of larger pieces, leaving smaller, fussier pieces for the end. For example, the frame of the dresser will come together rather quickly before starting on the more labor-intensive and monotonous task of putting together all the drawers that go inside. This allows the customer to feel satisfied almost immediately with their progress.</p><p>How might this apply to onboarding strategy? <strong>First, encourage your user forward by giving them a sense of their progress.</strong> For example, a progress bar or explicit steps orient a user as they work through a task.</p><p><strong>Second, front-load your “aha moment.” </strong>We recently tested an email builder with users. As users discovered the drag-and-drop feature towards the beginning of their interaction with the builder, they often commented how intuitive and easy that seemed. The perception of that early interaction seemed to shape their impression of the entire app experience, even if they struggled with some of the more complex tasks we asked them to try. Try to position a big accomplishment towards the beginning of an onboarding experience. It will keep your users more engaged and motivated. That initial success will build up your users’ confidence, so that they feel more capable of tackling challenging tasks.</p><h3>Give Your Users Some Space</h3><p>Instructions aren’t for everyone. Some users like to just hop in and get started. The nice thing about a book of instructions is that they are available for you when you get stuck, but you aren’t required to use them. An ideal onboarding process is available and accessible when you need it for support, but it isn’t obtrusive. You can skip it, if you just want to jump in and get going, but you can refer back later if you need help.</p><p>Making connections between real life experiences and digital experiences can strengthen our thinking about design. Putting together furniture renewed my empathy for new users, and it also gave me a deeper appreciation for a good onboarding process. For me, discovering these analogies gives me a more concrete understanding of the problem, fresh perspective, and a starting point to begin developing solutions.</p><p><em>Originally published at </em><a href="https://www.viget.com/articles/youre-good-at-your-job"><em>www.viget.com</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=68fb068796b2" width="1" height="1" alt=""><hr><p><a href="https://medium.com/viget-collection/what-ux-designers-can-learn-from-ikea-68fb068796b2">What UX Designers Can Learn From IKEA</a> was originally published in <a href="https://medium.com/viget-collection">Inspire + Advance</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[No, Amazon’s Latest Patent Won’t Disrupt Apparel Manufacture]]></title>
            <link>https://medium.com/@kaolvera/no-amazons-latest-patent-won-t-disrupt-apparel-manufacture-76098ee57431?source=rss-527573d8209c------2</link>
            <guid isPermaLink="false">https://medium.com/p/76098ee57431</guid>
            <category><![CDATA[amazon]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[fashion]]></category>
            <category><![CDATA[product-development]]></category>
            <category><![CDATA[fashion-design]]></category>
            <dc:creator><![CDATA[Katherine Olvera]]></dc:creator>
            <pubDate>Wed, 24 May 2017 19:25:07 GMT</pubDate>
            <atom:updated>2017-05-31T04:15:13.681Z</atom:updated>
            <content:encoded><![CDATA[<p>Over the last few weeks, I’ve read a few <a href="https://www.recode.net/2017/4/18/15338984/amazon-on-demand-clothing-apparel-manufacturing-patent-warehouse-3d">articles</a> that review Amazon’s latest <a href="https://www.nytimes.com/2017/04/30/technology/detailing-amazons-custom-clothing-patent.html">patent</a> for a process to automate apparel manufacture. Most of them seem to focus on how Amazon’s patent to make custom clothing could revolutionize clothing manufacture or even “<a href="https://seekingalpha.com/article/4068709-amazons-demand-manufacturing-patent-sounding-death-knell-apparel-manufacturers">sound the death knell</a>” for the apparel industry. I’m here to say that I just don’t see it.</p><p>For one, most of the writers have tended to be tech and business experts. I wanted a different take, so I decided to share it around with some of my friends in the fashion industry, people who understand the ins and outs of construction and production of everything from apparel to handbags to see what their thoughts were. I wanted to see if I was missing something. Turns out I wasn’t.</p><p><em>To me, this patent demonstrates very little understanding of the complexities and nuances of how clothing actually gets made</em>, especially at scale, and what parts of the process are best suited to optimization. So, let’s dig in.</p><h3>Amazon is a retailer.</h3><p>First, let’s talk about Amazon’s business model. Amazon is first and foremost a retailer. That means they are very good at predicting what customers want, where those customers are, and sorting out how most efficiently to get those products to the customer’s door. They are experts at data and analytics and logistics.</p><p><em>What that doesn’t mean is that Amazon has much experience in actually making products.</em></p><p>For decades, apparel retailers have outsourced the actual production to wholesalers. That means both parties can hedge their risk somewhat. In the most simplistic terms, retailers take on risk in inventory and real estate, while wholesalers take on risk in material liability and design and development overhead. So, retailers buy goods from wholesalers ready-made, which they believe will be appropriate for their customers, determine where to station those goods, and sell them to the consumer.</p><p>Amazon is an apparel retailer, not an apparel wholesaler.</p><p>This brings me to the first point of confusion. I’ve seen a couple writers mistakenly cite Amazon’s private label brands, like Lark &amp; Ro, as proof that they have entered the apparel manufacturing game. That is not necessarily true. A private label generally means that one company contracts with another company who specializes in design and manufacturing to develop and make clothes for them, which they can then sell under their own name. Typically, these consulting companies are ones that consumers have never heard of, but who are experts in handling all the development of the product. In this case, that company is Li &amp; Fung.</p><p>So, a company, like Amazon, chooses to invest more capital upfront in the manufacture to stand to make more profit from getting both the wholesaler and retailer markup on clothing, but it doesn’t often mean that they actually oversee manufacture of the clothing, or even know the specifics of how that clothing is being made.</p><p>So, why do I think Amazon might not have much experience when it comes to actually manufacturing apparel? A cursory glance at the proposal from any person with apparel manufacturing experience tells them that the proposed system doesn’t look particularly efficient. The drawings would indicate some major gaps in knowledge of the basic construction of clothes.</p><p>They propose a system to take in custom orders and use computers to properly batch and arrange pattern pieces to reduce wastage and increase efficiencies in getting the product shipped to the customer.</p><p>Okay, interesting idea, but it lacks consideration of all the complexity involved in reducing wastage.</p><h3>Amazon didn’t invent the marker.</h3><p>Let’s start with the marker. The marker is the layout of the pattern pieces on the fabric to be cut. Arranging the pieces in a way to reduce waste isn’t novel. It can be done, and is done, very efficiently by highly trained patternmakers and technical designers. Furthermore, many factories and companies already employ computer software to assist in optimizing markers.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/848/1*CaQvzXpV4RFzd5Er1wLcyw.png" /><figcaption>Figure 5-A of Amazon patent (US 9623578 B1)</figcaption></figure><p><em>I think it is also worth noting that the above picture is not at all what actual pattern pieces look like. Sleeves are cut separately from the body, for example. If these were sewn as shown, the consumer would not be able to move their arms up and down.</em></p><p>Now if the lay person were to look at a traditional marker, he or she may see places where there appears to be wastage. That person will want to turn pieces on the fabric to close up all the gaps like a puzzle. But, that would be a huge mistake (and one that Amazon made in its patent drawings).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/848/1*3vlflSCfByn2G6TH0RizRA.png" /><figcaption>Here, I’ve added in the proper grain lines for these pattern pieces on this theoretical marker. In practice all these lines should run the same direction parallel to the edges of the fabric.</figcaption></figure><p>Fabric has a grain, which means that for clothes to hang properly on the body you can’t layout pieces willy-nilly on the fabric to reduce wastage. You will cause all sorts of problems in production. Generally, off grain pieces turn into terribly cheap-looking puckers once they get to assembly, and the end product will tend to stretch and grow in unsightly ways once it gets to the consumer. All the pieces should be laid out to account for this with the grain line parallel to the edge of the fabric. This is Patternmaking 101, and it is essential for garment construction<em> at any scale</em>.</p><h3>An attractive and appealing wardrobe is made of a variety of textiles.</h3><p>Now, this patent asserts that Amazon would be able to manufacture using a variety of textiles, but I really want to know how vast that library would be. And would Amazon lose their economy of scale when it turns out that most wardrobes aren’t made of 10 or even 20 fabrics, but many, many variations of fabrics. I, personally, don’t want to go back to the days of leisure suits.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/450/1*DcU8ZnKwRSetEJtDKG_GkQ.jpeg" /><figcaption>This is not the future I envisioned.</figcaption></figure><p>People’s attraction to clothes isn’t just about color or cut, it is also about feel. Is it itchy or stiff? Is it soft? Is it warm? Is it heavy? Is it light? These all relate to the textile choice.</p><p>What I see in the patent drawings are very different garments cut from the same fabric<em>.</em> There appears to be a blazer and t-shirt (though I can’t be sure, given the inaccuracies of the pattern pieces) made of the same cloth. <em>That is just wrong.</em> Do you own a jersey blazer? Or maybe you have an oxford cloth t-shirt?</p><p>And, while Amazon is planning on printing the fabric, printing a fabric doesn’t essentially change the character of the fabric. The variety of fabrics needed to manufacture most wardrobes would be vast. For a super basic, casual wardrobe alone, here’s the most concise list of <em>just some</em> of the fabrics necessary:</p><ol><li>Jersey for t-shirts</li><li>Shirting for button down shirts</li><li>Pique for polos</li><li>Twill denim for jeans</li><li>Fleece for hoodies or sweatshirts</li><li>Nylon for rain-proof jackets or coats</li></ol><p>But even this list is extremely top-level; it doesn’t account for varying weights and hands of fabric. Textiles are more than their surface pattern. They each have a unique content and construction that gives them and their resulting products very different character. Limiting a customer’s choice to even 1 jersey, 1 shirting, or 1 denim as a base doesn’t feel very custom. And, this process also doesn’t account for the fact that many clothes have other finishing processes during and following production that make the clothes feel and look wearable (like distressing and washing for jeans). But, there is a deeper problem here.</p><h3>Do you know how many materials it takes to make just one garment?</h3><p>In this fantasy world, fabric is cut from one piece of cloth and goes onto a conveyor belt to be delivered to an attendant to be sewn. That’s a big problem. If done this way, most garments would be pretty incomplete, especially a shirt or a blazer.</p><p>While most people may not know this, our clothes aren’t just made of the fabric you see on the surface. There are various support fabrics inside to give the garment structure. A collar stays nice and stiff around your neck by having an extra piece of fabric called an interfacing sewn in. Even the most basic t-shirt has ribbing around the neck that isn’t cut from the same jersey used for the body. I don’t see where in the process those pieces are being cut and added to the bin.</p><p>Or buttons, or zippers, or a million other trims. Yep, I don’t see any of that stuff there. So, for right now, I’ll give them the benefit of the doubt that somehow all the requisite pieces are put in tubs and arrive to be sewn at their appropriate stations. And here’s where the patent gets a little crazy because…</p><h3>Robots DON’T sew your clothes.</h3><p>I want to make it abundantly clear to everyone that there are <strong>no robots </strong>sewing your clothes. This seems to be a fairly common misconception.</p><p>The dexterity needed to put together the pieces in the right order and move them through the machine is too complicated (and therefore expensive) to have robots do it. There isn’t even an incentive to develop a robot to do it as in other industries because the end product is so low-price and relatively low-margin. Yet, Amazon’s patent states that “an attendant, or automated sewing machine” will assemble the garment according to instructions delivered to them.</p><p>But, I think that everyone should know and appreciate that for every piece of clothing in your wardrobe, for every seam, someone somewhere sat at a machine and fed that cloth through the machine by hand, <em>one at a time</em>.</p><p>Making clothes is labor-intensive. If you ever have the opportunity to watch someone make a garment at a factory, it is awe-inspiring. These “attendants” are highly-skilled laborers. They are able to ease around curves without skipping a beat and keep the machine humming at 100mph, moving piece after piece through the machine. Trust me, that’s hard. I wouldn’t be able to do it.</p><p>But one of the reasons they are able to move pieces through so quickly at scale is because they are concentrating on one type of garment at a time, possibly even one seam in one garment. They aren’t making one pair of pants. Then a t-shirt. Then a blouse. Putting together garments isn’t like putting together IKEA furniture (which also isn’t as easy as it looks). It is hard work, and each new piece takes thought and practice.</p><p>So, while you may reduce wastage by laying out pattern pieces a certain way, you are likely to lose efficiency on the actual production line.</p><p>Even if computers made sure that each attendant was only making one type of clothing, there is another problem. Thread. Thread is matched to the color of the garment being made. If you have custom orders, and they are printed on the bolt, as shown in the drawing, when is the thread being delivered to the attendant? That seems like it would be a huge waste of time to rethread the machine for every garment.</p><h3>Real Optimization in Fashion</h3><p>Introducing these factors, there are now some more complex decisions to make about optimization. Do you optimize to reduce waste of materials? Or do you optimize to reduce time spent on actually making the garment?</p><p>There seems to be a general assumption that the current fashion industry is wasteful in its production. It isn’t. How else can you find a t-shirt that costs $10 or less? It is most wasteful in its distribution — that is getting the right assortment of products to the right regions. With a sized product, like apparel, that is especially complicated. But,<strong> <em>that </em></strong><em>is the area where Amazon is best equipped to disrupt, </em>not in production. They know where all the consumers are and what they want better than anyone else.</p><p>There are a million ways that custom orders might be handled, but this patented production method looks half-baked, inefficient, and unlikely to deliver on bespoke-quality custom clothes, which is what I as a consumer would expect for a “custom” apparel product. (If, however, it turns out that this is actually some sort of sample-making or prototyping process, which is what looks more probable, I’d love to make some stuff! Though that wouldn’t change its core inefficiencies or problems. Trust me.) If Amazon does want to go this route, I highly recommend that they find some experts at patternmaking and production. I’d be happy to point them in the right direction.</p><p><em>*Rant ended.*</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=76098ee57431" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>