<?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 Jayanth Mysore on Medium]]></title>
        <description><![CDATA[Stories by Jayanth Mysore on Medium]]></description>
        <link>https://medium.com/@jayanth.mysore?source=rss-fc0c7c044c12------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*q51tiVzDee-DJ8ZF.</url>
            <title>Stories by Jayanth Mysore on Medium</title>
            <link>https://medium.com/@jayanth.mysore?source=rss-fc0c7c044c12------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 02:47:58 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@jayanth.mysore/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[Betting on goodness: One way to build a User Generated Content platform.]]></title>
            <link>https://medium.com/@jayanth.mysore/betting-on-goodness-one-way-to-build-a-user-generated-content-platform-ae757fdf18c3?source=rss-fc0c7c044c12------2</link>
            <guid isPermaLink="false">https://medium.com/p/ae757fdf18c3</guid>
            <category><![CDATA[tech]]></category>
            <category><![CDATA[ugc]]></category>
            <category><![CDATA[maps]]></category>
            <category><![CDATA[user-generated-content]]></category>
            <dc:creator><![CDATA[Jayanth Mysore]]></dc:creator>
            <pubDate>Fri, 30 Sep 2016 09:15:44 GMT</pubDate>
            <atom:updated>2016-09-30T09:15:44.070Z</atom:updated>
            <content:encoded><![CDATA[<h3>Betting on goodness: One way to build a User Generated Content platform. My observations from building one for Google Maps</h3><p>This is the content of a talk I gave at Sequoia Capital (India)’s Chief Product Officer Summit in Sep 2016 and earlier at the CS department of the Indian Institute of Science, Bangalore. Since several people have indicated an interest in the area of UGC and had follow up conversations with me, I am sharing this publicly. If you do find this useful and want to have a conversation, just write to me at <a href="http://mailto:jayanth.mysore@gmail.com/">jayanth.mysore@gmail.com</a>.</p><p>The content is based on my experience while leading the efforts to engage contributing users of Google Maps. We built a UGC system consisting of multiple editing interfaces, an algorithmic moderation system that could auto-moderate edits and a system that engaged users to moderate edits submitted by their peers.</p><h3>Introduction</h3><p>Over a billion users use Google Maps every month ! What most of these users (maybe even you?!) are not aware of is that Google puts the power of updating that canvas in the hands of every one of them. In this writeup, I will cover the overall set of products that come together to make this happen. I’ll explain how these products, along with an Machine Learning driven moderation system orchestrate the maker-checker system needed to keep users engaged while ensuring risky changes are handled effectively (oh well… effectively is a big term)I will then share my observations about users who engage in UGC in this context — their behavioral characteristics that forms the underpinning of a product/technology strategy to engage them.</p><h3>Basics — Where does GoogleMaps get its data from?</h3><p>I’m covering projects that Google has publicly announced and provided links for you to understand these more. The short answer is that for a number of key countries Google runs a project called Ground Truth to map them in great detail, while for a large number of others, users map the countries. In all countries, however, Google exposes a set of tools and features for users to edit maps and local business information.</p><ul><li><a href="http://techcrunch.com/2014/09/03/googles-ground-truth-initiative-for-building-more-accurate-maps-now-covers-50-countries/">Project Ground Truth</a></li><li><a href="http://www.google.co.in/mapmaker">Map Maker</a>, <a href="http://www.google.co.in/maps/@12.9580628,77.5831909,18.27z/data=!10m2!1e3!2e3">RAP</a>,<a href="http://i.ytimg.com/vi/DkzcsUmcXog/maxresdefault.jpg"> Q&amp;A</a></li><li>Business owners: <a href="http://www.google.com/business/">Google My Business</a></li></ul><h3>What are some core principles for engaging users?</h3><p><em>These are my observations. There will be errors as my experience is only about 5 years long and I know I made many mistakes during that time. We were a great team that was willing to learn from these and kept iterating.</em> It took real experience to learn what was wrong and all I can assert is what I have captured here is what I believe to be relevant and accurate as of about a year ago. I have since moved on to solving a different problem in an unrelated domain. Please share counter-points and make this better for others !</p><ol><li>Remember that <strong>you are building a product — including a UGC product — FOR users, not for yourself.</strong> This is a <em>common mistake</em> in the mental model that teams that want to engage users as contributors suffer from (“we<em> need </em>UGC” is not a logical answer to build UGC experiences. The question is <em>do your users need a way to contribute? </em>“How many edits did you get?” is the wrong question to ask;”how many users did you engage?”, “how many users came back?” are right questions</li><li><strong>Consumption is the motivator for contribution.</strong> Power consumers are a prime candidate group to build UGC features for</li><li>There is<strong> far more goodness in people</strong> than the desire to deliberately create trouble. Don’t get stuck because someone says “Oh, but what if people spam”…It’s more fun trying something bold and there is far more goodness in people out there</li><li>A <strong>few trouble makers can slow the entire process down. </strong>Spammers are there…just deal with it :)</li><li>Our <strong>lives are lived in the long-tail</strong> of the physical world</li><li>It’s <strong>impossible to keep up with all changes that matter</strong> to users without engaging them (be it maps, be it wikipedia, be it restaurants in a town or places to stay around the world)</li><li>In any discussion forum, <strong>the</strong><a href="http://en.wikipedia.org/wiki/1%25_rule_(Internet_culture)"> 1–9–90 rule</a><strong> seems to hold</strong></li><li>People join in, <a href="http://www.youtube.com/watch?v=fW8amMCVAJQ">when they see</a><strong>“others like them”join in </strong>and make a difference</li><li>Relying <strong>strictly on intrinsic motivation is key to ensuring data quality. </strong>When you pay or incentivize people to contribute, you are talking to a very different audience. The control system you need to build for inputs from such a source are not the same as what you’d build for an honest UGC system</li><li>Creating a corpus of good content can be parsed into two types of activities:<strong>contribution of knowledge</strong> and <strong>organization of knowledge.</strong></li><li>Knowledge is long tailed. You have to <strong>engage the long tail in contribution</strong></li><li><strong>Power users</strong> seek to do both.<strong> Organize task flows to facilitate both.</strong></li><li><strong>Power users deserve efficient product flows. </strong>Power users are NOT people who will put up with anything because they are passionate- that’s a common, wrong, interpretation. Power users look for keyboard shortcuts !</li><li>UGC is<strong> not the same as paid crowd sourcing</strong>…the latter is more like VGC (vendor generated content :))</li></ol><h3>Architecting the user engagement model</h3><p>This a very important decision that the team has to make. How do you want to engage users? Who are your power users? Do they matter*? How will you develop a trust system? Why will users find it fair?…</p><p><em>*if you found this a weird question, it’s actually not so weird…power users will be that super tiny number of people who deserve a LOT of attention. People, i.e. everybody whom you need to help you build features, don’t understand them all that easily. You’ll need conviction to build features for them instead of 100x the number of non-power users.</em></p><p>The typical trade-off is between the cost of an accidental error versus the benefit of freshness. As an example…mispelling the name of a clinic that’s in the middle of nowhere is a small mistake compared to not having it on the map…ask someone who desperately needed to see a doctor in the middle of nowhere. Incorrectly adding an exit to a highway is a far more expensive mistake. Don’t blindly follow what other UGC products might have done — the constraints of your situation are unlikely to be identical to those of others. The engagement model will inform the mechanics of engagement suitably — user hierarchies, forums etc. In the case of Google Maps, our model took advantage of our unique characteristics of there being 100s of millions of users actively using the map, a few millions that take the time to complain when we let them down and an even smaller number relishing the freedom to create maps. We intersected this with the inherent heterogeneity in the data to construct adaptive feedback loops. Practically, here are examples of the forums we ran</p><ul><li><a href="http://support.google.com/mapmaker/answer/6268380?hl=en">Regional Leads</a></li><li><a href="http://productforums.google.com/forum/#!forum/map-maker">Map Maker user forums by region</a></li><li><a href="http://sites.google.com/site/mapmakerpedia/">Map Makerpedia</a></li></ul><h3>How do you strike the balance between quality and coverage of data?</h3><p>Here are three guiding principles I’ll be able to argue strongly for:</p><ul><li><strong>A well run UGC system is self-regulating — entities unfamiliar with the information should not interfere with the content (</strong>so before you decide you can “hire” a moderation team to decide what is “right”, think hard….it can very well turn out to be what kills the system)</li><li><strong>A committed user who contributes never intentionally commits a mistake </strong>(so don’t spend too much time second guessing inputs from a user who edits a lot of good content…and DEFINITELY read the next principle !)</li><li><strong>A committed user does not want to be over-ruled by someone or some algorithm who(that) doesn’t know as much </strong>(somebody who patiently draws miles of roads and names them doesn’t want to be told by a person who does not even live in the same continent or an algorithm that she is wrong)</li></ul><p><strong>Truths we embraced in constructing a moderation model</strong></p><p>Less than 10% of users who edit, moderate</p><p>Heterogeneity is your friend</p><ul><li>All data is not created equal</li><li>All edits are not created equal</li><li>All users are not created equal</li></ul><p><strong>So, how do we solve the moderation problem?</strong></p><p>We adopted a four pronged approach outlined below</p><ol><li>An algorithmic moderation system that takes advantage of (a) the heterogeneities in data, edits and users and (b) the guiding principles above, to adaptively determine what’s safe to apply vs what’s not. By keeping extrinsic motivations strictly off, the algorithms are able to take a more trusting stand. For edits that it’s not able to make a decision on, we relied on the next three approaches…</li><li>Engage power users in a region: Regional Leads</li><li>Engage the long tail of users: Crowd sourcing votes</li><li>Never, ever let an edit sit forever: Use Google moderators to decide</li></ol><h3>Interesting stories I get to tell</h3><p>(aka, you need to stay the course…you’ll upset everyone from users to the government to other engineering teams whose life will be so much simpler if this UGC thing was just not there !!)</p><ul><li><a href="http://www.theatlantic.com/technology/archive/2011/05/osama-bin-ladens-compound-already-mapped-on-google/238116/">Abbottabad</a></li><li><a href="http://sites.google.com/site/mapmakerpedia/community-projects/slum-mapping">Slum Mapping in Kampala, Uganda</a></li><li><a href="http://www.huffingtonpost.com/entry/google-maps-north-korea-prison-camps-nuclear-complex-gulags_n_2573238.html?section=india">North Korea</a></li><li><a href="http://timesofindia.indiatimes.com/tech/tech-news/FIR-filed-against-Google-over-Mapathon-2013/articleshow/40194009.cms">India Mapathon</a></li><li><a href="http://thenextweb.com/google/2015/04/24/theres-an-android-peeing-on-apple-on-google-maps-and-we-cant-figure-out-why/#gref">Android Pee’ing</a></li></ul><h3>Some videos that visually show the scale of UGC</h3><p>Power users treat stuff like this (not money) as reward for their work</p><p>Timelapse videos of cities coming to life on Google Maps — if you have any doubts about how much users enjoy contributing….</p><p>Note the impact of “external reviewers” on the pace and scale of mapping</p><ul><li><a href="http://youtu.be/DFowAeIRaiM">Lahore</a> (almost NO reviewers who were not contributors as well or an algo)</li><li><a href="http://youtu.be/ZqSxDwMeIoU">Kottayam</a> (some reviews from Google moderators — i.e. “external” to Kottayam, from a contributing user’s perspective)</li><li><a href="http://youtu.be/A-GZoHhYfY4">Hyderabad</a> (even more reviews from Google moderators)</li><li><a href="http://www.youtube.com/watch?v=1o1j--EqVk8">Bucharest, Romania </a>(almost no reviewers external — see the level of detail)</li></ul><p><a href="http://www.google.co.in/mapmaker/pulse">Map Maker Pulse</a> — see edits happening right now</p><h3>Some useful talks, documents online</h3><ul><li>LISTEN: <a href="http://www.npr.org/2013/07/13/197986218/why-we-collaborate">Why we collaborate</a> — this is a TED Radiohour discussion by the awesome Guy Raz with luminaries who have created UGC communities or have interesting perspectives on how stuff gets done in a loosely collaborative environment: Jimmy Wales (Wikipedia), Luis Von Ahn (ReCaptcha), Clay Shirky (Ushahidi), Jason Fried (founder &amp; CEO 37 Signals/Basecamp), Jennifer Pahlka (Code for America)</li><li>READ: <a href="http://www.aaronsw.com/weblog/whorunswikipedia">Who runs Wikipedia? </a>— a superb write up by the late, brilliant Aaron Swartz</li><li>LISTEN:<a href="http://www.npr.org/programs/ted-radio-hour/449179937/open-source-world"> Open Source World </a>— another TED Radio hour discussion on Open Source, with very relevant lessons for UGC, as every Open Source project is a big UGC effort</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ae757fdf18c3" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>