<?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 Andrew Kemendo on Medium]]></title>
        <description><![CDATA[Stories by Andrew Kemendo on Medium]]></description>
        <link>https://medium.com/@andrewkemendo?source=rss-8a0d3c1ad4b4------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*lwRq1tcG25hmSo-zFdgwmg.jpeg</url>
            <title>Stories by Andrew Kemendo on Medium</title>
            <link>https://medium.com/@andrewkemendo?source=rss-8a0d3c1ad4b4------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Fri, 20 Nov 2026 16:22:41 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@andrewkemendo/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[Tyler Cowen almost makes a good case against Mega Corporations, unfortunately falls short]]></title>
            <link>https://medium.com/@andrewkemendo/tyler-cowen-almost-makes-a-good-case-against-mega-corporations-unfortunately-falls-short-3aee34802e0f?source=rss-8a0d3c1ad4b4------2</link>
            <guid isPermaLink="false">https://medium.com/p/3aee34802e0f</guid>
            <category><![CDATA[sharing-economy]]></category>
            <dc:creator><![CDATA[Andrew Kemendo]]></dc:creator>
            <pubDate>Wed, 15 Aug 2018 01:14:03 GMT</pubDate>
            <atom:updated>2018-08-15T01:14:03.946Z</atom:updated>
            <content:encoded><![CDATA[<p>Tyler Cowen recently wrote an opinion article for Bloomberg titled:</p><h4><a href="https://www.bloomberg.com/view/articles/2018-08-12/american-ownership-society-is-changing-thanks-to-technology">Americans Own Less Stuff, and That’s Reason to Be Nervous</a></h4><p><em>What happens when a nation built on the concept of individual property ownership starts to give that up?</em></p><p>Cowen’s primary claim seems to be that property ownership gives individuals a stake in the system, and decreasing rates of property ownership in the US is eroding our collective “stakes” in the system.<strong> </strong>Here’s the quote emphasized in the larger context of the paragraph:</p><blockquote>Each of these changes is beneficial, yet I worry that Americans are, slowly but surely, losing their connection to the idea of private ownership. The nation was based on the notion that <strong>property ownership gives individuals a stake in the system.</strong> It set Americans apart from feudal peasants, taught us how property rights and incentives operate, and was a kind of training for future entrepreneurship. Do we not, as parents, often give our children pets or other valuable possessions to teach them basic lessons of life and stewardship?</blockquote><p>This conclusion, that property ownership gives individuals a “stake in the system,” reflects a historically inaccurate analysis of the relationship between property owners and the larger community they live within. Cowen seem to unknowingly refute his own position throughout the piece.</p><p>Take these examples:</p><blockquote>We used to buy DVDs or video cassettes; now viewers stream movies or TV shows with Netflix […] Music lovers used to buy compact discs; now Spotify and YouTube are more commonly used to hear our favorite tunes.</blockquote><blockquote>The great American teenage dream used to be to own your own car. That is dwindling in favor of urban living, greater reliance on mass transit, cycling, walking and, of course, ride-sharing services such as Uber and Lyft.</blockquote><p>By definition if an individual doesn’t outright own the property they are using, like in his examples, then they are relying on some “system” to ensure their access to it. In other words, you are heavily bought in, or have stakes in, some “system” when you pay for Netflix, Spotify, YouTube or Uber.</p><p>This is partly a problem of definitions, as Cowen’s “System” that we’re implored to have stakes in is poorly defined. Is the “System” the one that is controlled by states, like public transport or roads that facilitate trade? Is it the “system” that is controlled by mega-corporations like our personal computing hardware, modern agricultural producers or medical providers? Is it some commingled public-private “system” like fiber optic network infrastructure, water utilities or energy production, where subsidies and private companies are working in conjunction?</p><p>I think really, what Cowen is trying to do is to use the growing popularization of the “sharing economy” as a harbinger of some nefarious ideological trend toward collectivization.</p><p>Take this quote for example:</p><blockquote>[T]he more commonsensical, broad libertarian intuitions of the American public encapsulate a more brutish and direct sense that some things we simply own and hold the rights to.</blockquote><blockquote>Those are intuitions which are growing increasingly disconnected from reality, and no one knows what lies on the other side of this social experiment.</blockquote><p>Aside from the No true Scotsman-ing of “Libertarian,” it seems like a pretty clear worry that as our commercial services go, so goes the nation’s institutions. Said another way, Cowen is worried that the more collectivist our services seem to be, the more collectivist our base impulses-and thus organizational demands-will be.</p><p>Cowen is right though, there is a power imbalance between lessors and owners in today’s world. However the argument that Uber, Spotify, AirBnB etc… offer even a patina of collectivization couldn’t be further from the truth. In fact it’s largely the inverse.</p><p>Workers provide their own capital (cars, homes) and property (music) to intermediaries which then monopolize trade between users and providers through an onerous terms of service. Thoreau would recognize such a state as being a “slave-driver of yourself.”</p><p>On the user side, you’re beholden to the owners of the platform to allow your paid access to the producer’s work, which they can revoke at any time for any reason, with little recourse from the user.</p><p>The sharing economy taken as a whole, looks a lot more like a command economy than a free market, with a cabal of corporations and financiers following each other’s lead in how they implement the commanding. All with consent and choice by users and workers alike.</p><p>The problem however is not that our impulses are becoming more collectivist in nature. It’s that the number of property owners is shrinking, and the remaining ones are not accountable in any real sense to workers and consumers. Cowen’s own examples show that power is massively imbalanced in favor of Corporations, not collectivist groups or governments generally. These massive private organizations create increasingly consolidated barriers to access to what were previously privately owned property, and their management of such access remain beyond scrutiny or reach of the user because of their outsized legal capabilities and regulatory capture.</p><p>For example, if you are an up and coming solo artist and want your music to reach a huge audience, releasing your music for digital download on your own site or CD, isn’t a pathway to success because the majority of listeners use Spotify or Apple Music. The market has spoken and it doesn’t support being outside of the mega-corporate distribution system.</p><p>Cowen is so close to making a cogent anti-mega-corporatist argument, but either doesn’t want to believe that “free-enterprise” has led to a massive power imbalances or doesn’t see it. There’s a case to be made that crony capitalism has led to this state which I would suppose he is more apt to make.</p><p>Entrenchment, capital accumulation, information asymmetry and consolidation of services and products into a handful of companies, has effectively removed any ability for consumers to “vote with their feet.” Possibly competitive products or services are absorbed or killed before they mature as competitors.</p><p>To give credit where due, this state of affairs is the result of supremely effective and efficient organizational management by corporate heads. They used the tailwinds of their massively successful products to invest and diversify wisely, hire strategically and prevent being disrupted by consistency disrupting themselves. It’s genius. It is however devastating to a robust and flourishing distribution of economic power.</p><p>Cowen’s promotion of “Empire Builders” that seek to monopolize industries, is actually what America was built on, and what we’re seeing now is simply an extension of hundreds of years of rent seeking. Only now, this growth of consolidation is moving so fast, our organizations for collective action (Unions, Government Organizations, Advocacy Groups) are so weak, and the smartest people in the world are joining these massive companies to such an extent that there is nothing to stop them. Not even other potential empire builders, as investors and lenders are increasingly less willing to invest in new companies competing in those verticals because they are afraid that major corporations will kill them.</p><p>So Cowen is right, there is a problem, but that problem isn’t an increased impulse to collectivize in the sense that a right-libertarian is worried about — though it might seem like it from the way some young people talk today. The problem is that the “system” is forcing us collectivize around inscrutable and increasingly powerful corporate systems just to live and get along in modern society.</p><p>Cowen should be promoting shared ownership of firms, which promotes the distribution of gains to employees. Those employees will then have the capital to start their own firms and the cycle of creative destruction and socially distributed ownership can continue.</p><p>Promoting the concept of “own more of the stuff you use, don’t be a collectivist” is just a distraction. We should be moving to more intentional distributed ownership of firms to correct this imbalance and it’s possible to do that without coercive structures in a way that is fair to firms and individuals. It’s important to honestly identify the right boogeyman here first though.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3aee34802e0f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The startup clock is ticking: You must raise a Series A within 2 years to be a Unicorn]]></title>
            <link>https://medium.com/@andrewkemendo/the-startup-clock-is-ticking-you-must-raise-a-series-a-within-2-years-to-be-a-unicorn-7435c57680f?source=rss-8a0d3c1ad4b4------2</link>
            <guid isPermaLink="false">https://medium.com/p/7435c57680f</guid>
            <category><![CDATA[unicorns]]></category>
            <category><![CDATA[series-a]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[raising-capital]]></category>
            <dc:creator><![CDATA[Andrew Kemendo]]></dc:creator>
            <pubDate>Wed, 18 Apr 2018 19:08:23 GMT</pubDate>
            <atom:updated>2020-09-14T22:17:48.086Z</atom:updated>
            <content:encoded><![CDATA[<p>The following is a list of <a href="https://www.cbinsights.com/research-unicorn-companies">US based, name brand technology Unicorns</a>* and the time between when a startup company launched it’s core product and when it closed a Series A or significant venture financing, post product launch.</p><p>Lyft: 3 Months</p><p>Uber: 7 Months</p><p>Slack**: 8 Months</p><p>Robinhood: 9 Months</p><p>Stripe: 11 Months</p><p>WeWork: 1 year</p><p>Pinterest: 1 year 2 Months</p><p>Snap: 1 Year 5 months</p><p>Dropbox: 1 year 5 Months</p><p>AirBnb: 1 year 7 months</p><p>Houzz: 1 year 9 months</p><p>Twilio: 2 years</p><p>Admittedly this is a really simplistic conjecture and my data analysis isn’t exactly rigorous. Unlike the <a href="https://fleximize.com/unicorns/">Speed to Unicorn metric</a>, I’m much more interested in the first major cliff for a startup: Launch to Series A. That’s where the most attrition and that speed to Series A is a better predictor of unicorn success than speed between A and the $1B valuation.</p><p><strong>My gut tells me that if you aren’t being chased down by money within the first two years on the market, then you’ll never hit unicorn startup territory and IPO trajectory, no matter how great your company is.</strong></p><p>I’d love to see someone do an in depth analysis to prove this hypothesis wrong.</p><p><em>*I excluded Palantir and SpaceX because they were heavily funded by their founders, and do not fall into the same category of Venture Backed Startups as the others, and excluded Magic Leap because they remain pre-launch.</em></p><p><em>**Slack was a hard pivot and had raised money previously for a different product, however their biggest raise was the one after pivoting to Slack</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7435c57680f" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Artificial Intelligence and Privacy are Incompatible]]></title>
            <link>https://medium.com/@andrewkemendo/artificial-intelligence-and-privacy-are-incompatible-5375035f15c0?source=rss-8a0d3c1ad4b4------2</link>
            <guid isPermaLink="false">https://medium.com/p/5375035f15c0</guid>
            <category><![CDATA[consumer-behavior]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[deep-learning]]></category>
            <category><![CDATA[privacy]]></category>
            <dc:creator><![CDATA[Andrew Kemendo]]></dc:creator>
            <pubDate>Thu, 08 Mar 2018 18:29:27 GMT</pubDate>
            <atom:updated>2018-12-31T17:54:21.186Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/656/1*35wMzW3h5Fux6cnnSVVZrw.jpeg" /></figure><p>The Robot Butler has been a trope of science fiction since we could dream of computing. Today, Siri and Alexa serve as disembodied ancestors to the future robotic personal assistants we’ve always dreamed of. And boy do people like them. <a href="https://www.voicebot.ai/2018/03/07/new-voicebot-report-says-nearly-20-u-s-adults-smart-speakers/">43 Million Americans own some form of dedicated virtual assistants.</a> That’s almost 20% of US households.</p><p>Besides hands free commands such as setting a timer or checking the weather, some of the fastest growing command categories are recommendations. Things like wine pairings, recipe suggestions and movie suggestions are all easy tasks that can inform your decisions in a friendly and conversational way.</p><p>And wouldn’t you know it, the best way to increase the accuracy of personalized recommendations is to give these assistants more information about your wants, likes, needs and behaviors. Doing so helps Amazon or Apple build a better profile that can “learn” from and serve the user better in future interactions, just like human personal assistants do. Everyday, 43 Million Americans are providing <a href="https://www.wired.com/2016/12/alexa-and-google-record-your-voice/">Petabytes of this private data </a>to these companies through their personal assistants.</p><p>So maybe it’s worth thinking for a moment about what a “personal assistant” is anyway.</p><h3>Putting the Person in…personal assistant</h3><p>A personal assistant at their apex is a worker who has a deep understanding of every relevant attribute of their client, in order help the client improve their effectiveness and efficiency. Lets use two fictional examples of personal assistants:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/606/1*nrTx50SbYITAA4Q59qvzNA.jpeg" /><figcaption>Scene from The Devil Wears Prada Courtesy 20th Century Fox</figcaption></figure><p>In the movie The Devil Wears Prada, Andrea (Anne Hathaway) works as a personal assistant for Miranda (Meryl Streep) and spends every waking hour trying to meet her overly demanding needs. The amount of personal, private data that Andrea requires to predict and satisfy the needs and wants of Miranda is nearly equivalent to that of a spouse. Not only does she need to know buying preferences, but also medical needs, allergies, sexual preferences, work and sleep patterns, food likes and dislikes, family composition and personalities, location data, preferred activities, the list goes on and on. Without this information she can’t do her job successfully and the value of the relationship is diminished.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/615/1*SYhLp-EOB2H4vVqg-lLAeg.jpeg" /><figcaption>Scene from The Fresh Prince of Bel Air Courtesy Warner Brothers</figcaption></figure><p>In the Fresh Prince of Bel Air, Jeffrey the butler interacts with each family member in a unique, loving and personalized way because he has developed a personal relationship with each member over years of service. Jeffrey tailors his interactions with each person based on what he knows of them and the situation they find themselves in. He understands and internalizes their individual personalities, histories, even their secrets, and how they will respond to different styles of suggestion to make their lives easier or more productive.</p><p>In both cases, the assistant and butler have a deeply intimate relationship with their clients, one that goes well beyond anything that most people would have with anyone other than a family member. They are part of our private life.</p><p>If we expect that Amazon’s Alexa or Google’s Assistant will fulfill our personalized needs and desires in the same way a human assistant would, then aren’t we required to build a similarly personal relationship with Amazon or Google? That means sharing our preferences and behaviors with them is a required step to create the feedback loops and deep understanding that is necessary to provide personalized value. Can you trust Amazon or Google with the same information you would give Jeffrey the butler? Should you bring Google into your private life?</p><h3>You can Trust me</h3><p>The number one disqualifier when looking for a personal assistant, butler, maid, plumber, babysitter — really any job come to think of it — is a lack of <strong><em>Trust</em></strong>. If you think a housekeeper will steal your jewels or your babysitter will ignore your child, you’re not going to hire them.</p><p>It’s not unreasonable to assume that a big part of why Amazon is leading in the smart home market is because <a href="https://www.forbes.com/sites/karstenstrauss/2017/03/28/americas-most-reputable-companies-in-2017-amazon-knocked-out-of-the-top-spot/2/#">Amazon is the second most trusted brand in the US</a>. People trust Amazon already, so it’s easy to build a stronger relationship with them. However this assumes that the users are aware of the data they are giving away, how it is being used, and have made a calculated decision on how much personal information to share based on their level of trust with Amazon. Probably a bad assumption.</p><p>It’s unknown to what extent users understand how much data they are giving away everyday, so it’s hard to know if they are wittingly trusting companies with their data or if they are unwitting participants in the data game. <a href="https://www.amazon.com/gp/help/customer/display.html?nodeId=201809740">How Amazon uses your data is all there if you want to read it</a>, however most people don’t read it. Even if you did read it, it’s not exactly clear what is happening with your data, so most people either go with gut feel, or simply don’t think about it at all.</p><p>The <a href="https://www.eugdpr.org/">European Union’s General Data Protection Regulation</a> is taking a hard stance on this with the goal of having companies explicitly show users exactly how their data is being used, but I’m not convinced this will have the effect they intend it to. They want to force the question of trust into the spotlight, and that might work, but just like couples therapy you can only force the conversation so much before one side loses trust.</p><p>So trust is really what it comes down to. Can you trust a major company with your most intimate secrets? Chances are you probably already are.</p><h3>Long term relationship with AI</h3><p>Whether you know it or not, you are creating this personal private relationship with companies just by virtue of living in the modern world. Just search for “[Company] tracking/spying” and you’ll find hundreds of articles expressing concerns about privacy and how much data a company is collecting from users.</p><p>This behavior is not restricted to the “smart” tech corporations by the way. Even if you don’t have a smartphone, your phone <a href="https://wrongfulconvictionsblog.org/2012/06/01/cell-tower-triangulation-how-it-works/">carrier knows (generally) where you are at all times, because knowing your location is part of how cell service works</a>. The Safeway/Kroger discount card you use, or the mileage rewards program you enroll in, or any other “loyalty” program you’ve had for two decades, all of these are trying to do one thing:</p><blockquote>Predict your preferences and behaviors so that they can put the coupon/sale/product/content that matches your preferences in front of you at the right time and right place.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/432/1*1942_-EGDU59_rSws3yp4A.jpeg" /><figcaption>The Early Days of big data</figcaption></figure><p>The difference between now and 20 years ago is that users provide orders of magnitude more specific and persistent data in an easily digestible way. With advances over the last decade with Machine Learning and blended prediction systems, companies can process this data and get more and more accurate behavioral predictions. For example <a href="https://static.googleusercontent.com/media/research.google.com/en//pubs/archive/45530.pdf">YouTube made their recommendations system magnitudes better using Deep Neural Networks</a>. Amazon published their <a href="https://github.com/amzn/amazon-dsstne">Destiny engine which uses Deep Neural Networks to build better recommendations from user behaviors</a>. Both of these are in the category of “AI” if you aren’t familiar.</p><p>Better tools and predictions create better services for users, more tailored product offerings, more accurate recommendations and more efficient markets that keep you coming back to their services. As more companies use increasingly precise behavioral models to predict user actions with “AI,” the deeper these relationship will get. Organizations will seek to build ever more personal relationships with their users because they want to serve your preferences.</p><p>By default the recommendation systems that people are responding positively to <em>require</em> your personal input to be successful. Just like personal assistants do.</p><h3>Wait, I don’t want to have a relationship with AI!</h3><p>Actually, you probably do.</p><p>Remember those 20% of Americans that literally have an always listening voice based assistant in their home? They don’t keep those because they are forced to, they do so because the assistants really do provide value that people want.</p><p>Anytime you check in for your flight on your phone, search for something on google, like a post of facebook, create a board on pinterest, rate a product on amazon or leave a review on yelp, you are dancing with the system, leading with your left and the system is following with their right. You are giving them your private preferences for an increasingly tailored service. Your behavioral data is used to tailor the product to your wants, to personalize your experience, which theoretically keeps you happy and using it — or at least addicted to it, the sinister downside of giving people what they want. This is however what people have been asking for: better personalization. Congratulations, you’re in it.</p><p>I will make a prediction of my own: more deeply intertwined AI based services will increase the decision making capabilities for users over the next several decades to such an extent, that it will be considered irresponsible not to use them.</p><h3>Ok, fine then we’ll do it offline!</h3><p>I can hear you now: “Well fine, I value my privacy but I also agree that these services are beneficial. So we will just make systems that never touch Amazon, Google or Facebook or whatever mega corporation services. We’ll own our own data and I’ll just keep all my private data locally in my own home and have my open source Smart Speaker totally off the grid. Or maybe just send out relevant data where necessary. I’ll build my own <a href="https://arxiv.org/pdf/1509.02971.pdf">DDPG</a> based Deep Learning systems and teach it everything it needs to know!”</p><p>But will you? Will 100 Million Americans put the effort in to tweak and modify their own systems? Will 3 billion users worldwide? Is it feasible to think you can build a “smart” enough network with strictly compartmentalized data just from one person?</p><p><a href="https://www.theguardian.com/technology/2014/aug/24/internet-lost-its-way-tim-berners-lee-world-wide-web">The recent lament of how we “lost our way” </a>with the internet, is the latest proof that humans are great at consolidating power, so why would this time be any different? The cyber-utopian fantasy of egalitarian connectivity is in my estimation low probability, and I think it’s irresponsible to move forward with that as a thesis.</p><p>Practically speaking, robust information networks don’t work that way, they need to exchange information up and down to become better, faster and more accurate. I’m not just talking IT here, I’m talking about plant root systems, dolphin pods and migratory bird flocks. All nodes share information up the branches to make the system stronger, more responsive and more efficient. Whomever owns the most connected nodes has the power. We should recognize that as a natural law and build our social systems to compensate for that.</p><p>So how should we as individuals choose to enter personal relationships with organizations that provide services to us in exchange for our private data?</p><h3>Who should we trust?</h3><p>The first step is to recognize what these relationships look like, what data is being shared, how is it being used and how we can mutually benefit from the relationship as both users and creators without blowing the whole thing up.</p><blockquote>Second, We should acknowledge that you cannot simultaneously have systems which adapt to users behaviors while also keeping their behaviors out of reach</blockquote><p>Third, we need to study the cost vs benefit for these systems. My gut tells me that the tradeoff is net positive, but we need more evidence to show that. This is especially important for engineers and designers as we have a duty to provide value and not just extract value.</p><p>Hopefully this serves as a jumping off point for a larger discussion of this topic. I have my own wacky ideas of how we can have a positive outcome by building personal relationships between users and systems but I’m happy to admit that I’m not confident with any answer. I also recognize the risks and dangers inherent in how we’re building these systems which are really collections of human preferences.</p><p>I want to have a discussion here because I think it’s a critically important topic, so please send me your thoughts in comments here or elsewhere.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5375035f15c0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why Facebook’s AR Studio blows the AR market wide open]]></title>
            <link>https://medium.com/@andrewkemendo/why-facebooks-ar-studio-blows-the-ar-market-wide-open-a267cd5a67eb?source=rss-8a0d3c1ad4b4------2</link>
            <guid isPermaLink="false">https://medium.com/p/a267cd5a67eb</guid>
            <category><![CDATA[computer-vision]]></category>
            <category><![CDATA[augmented-reality]]></category>
            <category><![CDATA[facebook]]></category>
            <dc:creator><![CDATA[Andrew Kemendo]]></dc:creator>
            <pubDate>Sun, 14 May 2017 02:07:04 GMT</pubDate>
            <atom:updated>2017-05-14T02:07:04.822Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/970/1*Vxy_3Wc6IgOTCUegU_98Ng.jpeg" /><figcaption>Not much doubt about where FB is going in AR</figcaption></figure><p><a href="https://venturebeat.com/2017/04/29/facebooks-ar-studio-blows-the-market-wide-open/"><em>Author’s note: This article was originally published in Venture Beat on April 29th</em></a></p><p>Last week at Facebook’s F8 conference, Mark Zuckerberg went <a href="https://venturebeat.com/2016/04/12/zuckerberg-future-vr-and-ar-gadgets-will-look-like-ordinary-glasses/">all in on augmented reality</a> (AR). Since then his announcement has been praised as a great thing for the AR developer world and panned as an <a href="http://mashable.com/2017/04/18/facebook-augmented-reality-ar-platform/">attack on Snapchat</a>. See it here in full:</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FCchpWBIA_S4%3Ffeature%3Doembed&amp;url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DCchpWBIA_S4&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FCchpWBIA_S4%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/71c70f86ebc0edfa5b4890843fd4a48b/href">https://medium.com/media/71c70f86ebc0edfa5b4890843fd4a48b/href</a></iframe><p>As a longtime AR developer, I can tell you the announcement didn’t come as a surprise for its content but for the speed and breadth of its potential capabilities: What Facebook announced is a vertically integrated AR development and distribution platform at Facebook scale.</p><p>So why does that matter?</p><h3>It’s exceptionally hard to build mobile AR technology</h3><p>Mark made a great point: AR developers have traditionally had to first build a camera and get a lot of people to use it.</p><p>When he says “build a camera,” he’s not talking about physical hardware. He’s talking about the software behind the mobile camera that can understand the environment through depth, tracking, and object recognition.</p><p><a href="https://virtualrealitypop.com/heres-what-it-s-going-to-take-for-augmented-reality-to-take-over-the-world-275db4aabe7">Here’s What It’s Going to Take For Augmented Reality to Take Over The World</a></p><p>To do that you need a deep understanding of computer vision and machine learning techniques and the ability to optimize those techniques to work on smartphones (or heads-up display systems).</p><p>You also need to create applications and tools that draw a lot of users into them and keep those users engaged.</p><p>Saying those are hard things is really an understatement, and when we look briefly at what goes into those efforts it’s pretty clear why …</p><h3>What’s (probably) in the box?</h3><p>If we analyze for a moment what was introduced during the talk, the technologies that are inside this new platform are probably the following:</p><p><strong>Cross platform monocular SLAM. </strong>This is what gives you the ability to walk around an object placed in the environment.</p><p><strong>Monocular structure from motion.</strong> This builds a 3D representation locally of some portions of the real world inside the phone, allowing you to do things like occlusion (putting virtual things behind real things).</p><p><strong>Robust loop closure. </strong>When Mark talked about “leaving notes” for others, robust loop closure is what you need to enable that. Effectively it lets the system “remember” where these notes are left simply by looking at the environment.</p><p><strong>Object recognition.</strong> Understanding what objects are in the real world and then allowing you to give contextual data around them.</p><p><strong>Authoring and development tool.</strong> This is the biggest piece of the puzzle and would potentially allow anyone to create AR experiences with a very low bar of entry.</p><p>For any group of developers, any one of these tasks is monumental but tractable with the right team, funding, and partnerships. But rolling out all of these tasks in a seamless package that can work at scale is nearly impossible.</p><h3>Impact on AR developers</h3><p>If Facebook can deliver what it announced at F8, then it will have effectively taken the hardest part out of creating AR experiences: the tough engineering. That’s great for new creators but potentially devastating for anyone who specializes in developing these VERY difficult technologies.</p><p>More than likely, Snap, Apple, Android, and Microsoft will follow suit with their own consumer mobile AR developer platforms.</p><p>Mark is wrong, though, that “no one has built [an AR platform for developers] yet.” <a href="https://techcrunch.com/2011/10/27/metaio-adds-gravity-to-their-augmented-reality-platform/">Metaio did</a>, <a href="https://www.layar.com/documentation/browser/layar-platform-overview/">Layar did</a>, we at Pair have. But what he was getting at is right: No single AR platform has captured a significant portion of the market and stayed independent. It takes tens of millions of dollars in development and distribution to do that — and even if you have it, timing is everything.</p><h3>Brands and consumers will flock to FB AR</h3><p>Everyone in the industry feels that AR technologies are going to reshape the consumer experience for nearly everything, and we’ve seen some indications of that happening already. But the Facebook AR platform could supercharge that thesis with rocket fuel.</p><p><a href="http://insights.pair3d.com/2016/01/12/promoting-mobile-augmented-reality-applications-of-today-will-encourage-adoption-of-smartglasses-tomorrow/">I have been saying for a long time that smartphone-based AR will beat heads up display AR in the consumer market for the foreseeable future</a> while laying the groundwork for heads-up based AR. It seems Facebook agrees with this.</p><p>If Facebook is the primary platform for AR experiences, then Facebook entities are going to be the dominant places where people first experience AR shopping, games, navigation, etc. After all, that is the company’s intention. It’s hard to overstate the pent up demand in commerce for this level of integration, and I expect it will push retailers and brands into AR faster than ever.</p><p>Not only that, the information Facebook will get from evaluating a user’s environment visually from AR experiences will outstrip everyone else. It’s well known that Facebook has the best computer vision team on the planet. So retargeting, focusing content, and feeding users ads will become magnitudes more precise and tailored.</p><p>That also means it will be increasingly hard to compete with Facebook in the AR and commerce space for any type of interaction or behavior, as the company has a significant first mover advantage and the ability to iterate quickly.</p><p>There is one major caveat to all of this, though: These experiences will only be available through Facebook-owned properties. And there will likely be people who don’t want to work with Facebook for personal reasons.</p><h3>Leading to AR glasses: Bootstrapping a global map</h3><p>Some critics of this move by Facebook say that it’s a <a href="https://www.vrfocus.com/2017/04/facebook-and-ar-why-its-already-gone-wrong/">half measure for AR</a> and <a href="https://www.usatoday.com/story/tech/columnist/2017/04/25/augmented-reality-disappointment-facebook-snapchat-microsoft/100836274/">that it’s a disappointment</a> because it doesn’t do everything the Sci-Fi world told us AR would do. I think that’s a silly way to look at it, and doesn’t take into account the reality of incremental progress, even though Facebook’s move is a pretty big leap in and of itself.</p><p>Even then, it’s well known that <a href="https://www.recode.net/2017/4/19/15364674/facebook-augmented-reality-glasses-recap">Facebook is quietly developing consumer AR glasses</a>. However there is a lot of ground work that needs to be laid to make such a head-mounted display work. Everything from user awareness and social comfort to hardware improvements.</p><p>A seamless AR device <a href="https://virtualrealitypop.com/heres-what-it-s-going-to-take-for-augmented-reality-to-take-over-the-world-275db4aabe7">must have a map of the world around you</a> to effectively enable AR everywhere. If Facebook is successful with its mobile AR platform rollout, it will be able to bootstrap this map through mobile AR usage and use its robust loop closure system to provide a seamless experience by the time its glasses launch.</p><p>This is the biggest potential advantage Facebook has right now, and it shouldn’t be taken lightly. The company has more opportunity to own this market than anyone else.</p><h3>Challenges remain</h3><p>Facebook and the AR community still have a lot to do to prove that people will buy into the AR ecosystem for commerce, entertainment, and productivity at the billion user scale.</p><p>The company is grabbing low hanging fruit by helping developers create and users interact with the popular “lenses” effect.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KsQIamdzXSQVuGyq9NQjEQ.jpeg" /><figcaption>Call it Lenses, Filters or whatever, it’s the same thing</figcaption></figure><p>That’s a low threshold and easy to get into for most people. Certainly not far behind, however, are all of the applications that go beyond silly things and into productivity, commerce, navigation, health, and entertainment.</p><h3>New opportunities in the AR market</h3><p>While some may think the number of billion dollar startup opportunities in AR is going to shrink as a result, I think the opposite. What has traditionally been a fairly wide-open playing field (which has scared corporate and venture money to date) is now one where there is serious competition, validating the model.</p><p>New opportunities for studios and publishing will sprout from this announcement, meaning that content creators will boom and will inevitably look for other platforms to deploy onto. The pace of innovation for consumer head-mounted displays will increase, and adoption of AR technologies will accelerate. Importantly, 3D content creation will also boom, which enables AR for all kinds of activities.</p><h3>Bottom line</h3><p>I’m cautiously optimistic about the impact Facebook AR Studio is going to have on the AR world. While it might have a negative short-term impact on hard AR developers, in the end it will serve as impetus for the AR market as a whole to grow. We should be cautious about one player having dominance in AR and all that goes with it but excited about what such a big move signals across the market.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a267cd5a67eb" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Smartphone Camera Focal Length Reference for Computer Vision and AR applications]]></title>
            <link>https://medium.com/@andrewkemendo/smartphone-camera-focal-length-reference-for-computer-vision-and-ar-applications-cd4c932046b1?source=rss-8a0d3c1ad4b4------2</link>
            <guid isPermaLink="false">https://medium.com/p/cd4c932046b1</guid>
            <category><![CDATA[android]]></category>
            <category><![CDATA[development]]></category>
            <category><![CDATA[iphone]]></category>
            <category><![CDATA[augmented-reality]]></category>
            <category><![CDATA[computer-vision]]></category>
            <dc:creator><![CDATA[Andrew Kemendo]]></dc:creator>
            <pubDate>Wed, 26 Apr 2017 19:35:29 GMT</pubDate>
            <atom:updated>2017-04-26T19:35:29.478Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/919/1*lfhpDkHsIub7MR_hmerTlQ.jpeg" /></figure><p>For developers working in the field of computer vision, augmented reality, and to a lesser extent 6DOF tracking for VR, it’s really important to understand the camera properties of the devices that they will be running on.</p><p>While there are many important aspects of these cameras, focal length is arguably the most important because even small differences in focal length effect how you interpret depth. This is especially important for recovering scale from the camera.</p><p>I wanted to create a <a href="https://en.wikipedia.org/wiki/35_mm_equivalent_focal_length">35mm equivalent</a> reference for the top used phones in the US, because they are easiest to work with.</p><p>I figured I’d share to others could benefit. These are sourced from around the web, so please let me know if you have any updates.</p><h4>iPhone</h4><p>Original iPhone (2G): 37 mm</p><p>iPhone 3G: 37 mm</p><p>iPhone 3GS: 37 mm</p><p>iPhone 4 (rear-facing main camera): 30 mm</p><p>iPhone 4s: 35 mm</p><p>iPhone 5: 33 mm</p><p>iPhone 5s: 29 mm</p><p>iPhone 5C: 33 mm</p><p>iPhone 6: 29 mm</p><p>iPhone 6S: 29 mm</p><p>iPhone 6+: 29 mm</p><p>iPhone 6S+: 29 mm</p><p>iPhone 7: 28 mm</p><p>iPhone 7+: 28 mm</p><h4>Android:</h4><p>Samsung Galaxy S7: 26.6 mm</p><p>Samsung Galaxy S7 Edge: 26 mm</p><p>Samsung Galaxy S6: 27 mm</p><p>Samsung Galaxy S5: 31 mm</p><p>Samsung Galaxy Note 3: 31 mm</p><p>Samsung Galaxy Note 5: 28 mm</p><p>Google Pixel: 28 mm</p><p>Motorola Nexus 6: 28 mm</p><p>HTC One (M8): 27 mm</p><p>HTC One: 28 mm</p><p>Shout out to <a href="http://www.gsmarena.com/">http://www.gsmarena.com/</a> for about half of the numbers.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cd4c932046b1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Here’s What It’s Going to Take For Augmented Reality to Take Over The World]]></title>
            <link>https://medium.com/virtual-reality-pop/heres-what-it-s-going-to-take-for-augmented-reality-to-take-over-the-world-275db4aabe7?source=rss-8a0d3c1ad4b4------2</link>
            <guid isPermaLink="false">https://medium.com/p/275db4aabe7</guid>
            <category><![CDATA[human-factors]]></category>
            <category><![CDATA[privacy]]></category>
            <category><![CDATA[virtual-reality]]></category>
            <category><![CDATA[mapping]]></category>
            <category><![CDATA[augmented-reality]]></category>
            <dc:creator><![CDATA[Andrew Kemendo]]></dc:creator>
            <pubDate>Fri, 24 Feb 2017 23:50:19 GMT</pubDate>
            <atom:updated>2017-05-23T04:09:41.975Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/880/1*DmI1lkyZfBlHacpmaAQOMw.jpeg" /><figcaption>Tony seems…troubled. Image courtesy Marvel Comics</figcaption></figure><p>With <a href="http://www.marketwatch.com/story/tim-cook-sparks-predictions-of-augmented-reality-in-apples-iphone-8-2017-02-13">rumors swirling about Apple’s plans in Augmented Reality</a> (AR), as well as recent <a href="https://www.cnet.com/news/magic-leaps-mixed-reality-future-is-blurrier-than-ever/">teeth gnashing about the state of Magic Leap</a>, it’s beyond time for the AR community to have a real discussion about what it’s going to take for Augmented Reality to become the primary computing environment worldwide.</p><blockquote>The question is, what will it take for an AR system to replace the smartphone and the desktop as the ubiquitous computing device for everyday use?</blockquote><p>I’ve <a href="https://www.youtube.com/watch?v=m_0aoE9nrqM">spoken in the past about what I call the “AR stack.”</a> That is the set of technologies that are necessary to get to wide functionality and applicability for the average consumer. However I wanted expand on that and build a very basic reference that identifies key concepts necessary for AR, that people can easily cite when trying to understand this topic.</p><h4>Form Factor</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/965/1*Ab1JmtTJ6OKFw6WPNTrwyg.jpeg" /><figcaption>I’d wear those</figcaption></figure><p>One of the key assumptions that we in the AR community can generally agree upon, is that consumers will demand AR displays that look as close as possible to the existing form factor for eyeglasses. That means that the user is looking through a transparent display. These are called Optical See-Through Head Mounted Displays. The general and most commonly used term which will be used from here on out is, Head Mounted Display (HMD). HMDs also include the category of Virtual Reality displays, which are completely opaque. That does not mean that there won’t be significant adoption of AR before we reach that form factor, for example through smartphones, but AR HMD is arguably the idealized version.</p><p>I’ve heard on multiple occasions that even this wouldn’t be good enough. The reasons most commonly cited: You can easily lose or break glasses, they don’t work well while lying down (where many people use smartphones) and some people just don’t like wearing glasses.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/1*aDjeKfbQPYxT0Iu3uaaX7w.jpeg" /><figcaption>Wishful thinking</figcaption></figure><p>While I generally agree that there are constraints with the eyeglass form factor, I don’t think that the contact lens will be where we see widespread adoption — partly because it’s <em>probably</em> impossible in the coming decades.</p><p>Hopefully it will be clear after reading the following section, that it’s hard enough to pack all the required elements into a pair of glasses. Adding the requirement of doing it on a contact lens, is not practically possible in the near future.</p><p><strong>The Technical Requirements</strong></p><p><em>Field of View (FOV)</em>: The idealized AR display will conform to the <a href="https://vrwiki.wikispaces.com/field+of+view">combined binocular and peripheral field of view of the human eye </a>— which includes motion detection on the far edge of the FOV. That is approximately 200 degrees horizontal and 140 degrees vertical. Within this field, only about 140 degrees horizontal are binocular and the remaining 30–35 degrees are monocular peripheral vision.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/640/1*qtQmt4HrTcypqSIefU_t-Q.png" /><figcaption>Human Horizontal FOV</figcaption></figure><p><em>Pixel Per Degree (PPD)</em>: The idealized display would have equivalent pixel per degree resolution as the human eye at any depth of focus. That means a human can’t discern a difference between rendering elements (ie. pixels) at that resolution. While PPD and human eye “resolution” is not a one-to-one comparison (we don’t have pixels — though you could argue the point with respect to rods and cones), we can consider 60 PPD as the human equivalent necessary for a convincing display. This is the first key element required to eliminate the problem of eye fatigue with digital display systems.</p><p><em>Latency</em>: In a similar nature as PPD, the latency of the visual pathway should mimic as close as possible to the human biological visual processing pathway. <a href="http://news.mit.edu/2014/in-the-blink-of-an-eye-0116">Experiments put humans at reliably discerning differences in visual stimuli at around 77 Frames Per Second (FPS) or 13 milliseconds</a> between being shown a new image and recognizing it as such. To take a page from the VR world, if we want to be conservative we should expect that anything worse than 60 Frames Per Second would be unacceptable.</p><p><em>Accommodation</em>: When you want to focus on an object close to you, muscles around your eye, deform your eyeball lens ever so slightly to focus the incoming light (in the form of multiple and varying wavefronts) onto your retina. This process happens in reverse when you want to focus on something far away, and does so without conscious thought. <a href="http://hyperphysics.phy-astr.gsu.edu/hbase/vision/accom.html">This process is called accommodation</a>, and allows us to change our focus simply by refocusing our gaze. Ideally, a display would give us unlimited focal points for which we can focus our gaze. In practicality I expect that an AR system would only need to provide between 5 — 10 different focal depths to be deemed sufficient (too little data to know with much confidence). This is the second key element required to eliminate the problem of eye fatigue with digital display systems.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NqOmpvhhOOpKVyNdHDWQZg.jpeg" /><figcaption>Depiction of wave-front distortion with distance</figcaption></figure><p><em>“Rendering Black”</em>: One of the more contentious topics in AR is how, and even if it is possible, to successfully show black in a see through AR environment. Given that “black” is a relative lack of light coming into the eyeball from a specific vector, it’s not simply as easy as reducing the strength of the emission that you are sending to the eye in a see through transparent display. The visual cueing that humans get from shadows and light effects, are very powerful, and are a key aspect of creating presence of a virtual object in AR. [Note: There are theoretical ways to do this but none have been proven or practical. The most promising technique to render black pixels that I know of is to have an optical element (a lens) that selectively blocks all light waves from a specific direction, regardless of wavelength.]</p><p><em>Power</em>: Ideally this is a device that you can put on in the morning and take off at night without needing a re-charge. Matching iPhone 7 specs would put that between 10 and 13 hours of consistent use.</p><p>There are ungodly number of ways in which the community is trying to address these issues given the current state of technology.</p><p>The primary display method we see in practice, utilizes fixed micro OLEDs or LCDs, which are reflected through a series of waveguides presenting the image into your line of sight.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DSKxAXZM27QIUkaOlebz0g.jpeg" /><figcaption>Moverio BT-300 waveguide and LCD. Image Courtesy Epson</figcaption></figure><p>While nearly ubiquitous, this method has hard limits on FOV and PPD. Further, there is no known way to give accommodation utilizing this method.</p><p>I consider Virtual Retinal Displays (VRD) or Retinal Scan Displays the best hope for achieving consumer level AR display technology. Introduced long ago, but becoming increasingly feasible in practice, these displays can project multiple wavefronts with various curvatures directly onto the retina, allowing you to focus at different depths. Done correctly, this can potentially give you a display with nearly unlimited FOV, accommodation and extremely dense PPD on the order of human acuity limits.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HfsgEQ0FIgWTGkb97vAo0Q.png" /><figcaption>VRD diagram. Courtesy: <a href="https://www.google.com/patents/EP0562742A1?cl=en">https://www.google.com/patents/EP0562742A1?cl=en</a></figcaption></figure><blockquote>BOTTOM LINE FORM FACTOR: For AR to be the primary computing device the form factor needs to be at least a pair of standard eyeglasses, which give human equivalent Field of View, Pixels Per Degree resolution, and Accommodation at a high enough refresh rate and a day worth of battery life . I’ll hold judgement on whether “rendering black” is required for wide scale adoption.</blockquote><h4>Environmental Tracking</h4><p>I wrote about <a href="http://insights.pair3d.com/2016/02/23/environmental-understanding-for-augmented-reality/">environmental tracking on the Pair Blog</a> last year but it is worth discussing briefly why this is important.</p><p>In the real world, physics dictates that objects generally stay in the place where they are set, relative to the objects around them. In the Augmented world we can’t take this for granted.</p><p>In order for you to see an object in an AR environment and have it look convincing, there needs to be a way to “anchor” the virtual object to the real world. In an AR environment, this consists of tracking real-world objects (really image features) and translating their location to a virtual coordinate system. In practice this tells the system things like how far the viewer is to a real object, the height of the viewer with respect to a real object and so on. Implementations of this are generally captured in the discipline of Simultaneous Localization and Mapping (SLAM), though there are less robust ways to do tracking which will not be discussed.</p><blockquote>Think about it like adding gravity to the Virtual World</blockquote><p>Once you combine the spatial awareness of the real world, with a representation of it in the virtual world, then you can add new objects into the real world. As long as you maintain this combination, these worlds will synchronize such that when a viewer moves, the view of the object moves in-sync with the view, the same way as objects in the real world do.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/1*tCAtRF8ECr6_6aSA2fXjaQ.gif" /><figcaption>Monocular SLAM on an iPhone. Courtesy Pair3D</figcaption></figure><p>As mentioned in the previously published article, there are several ways to do this. You can either do it passively, by interpolating a 2D video feed (or feeds) you receive from the real world; or you can do it actively by sending out laser, infrared or ultrasonic emissions to return depth estimates immediately. The distinction between these is in the hardware necessary and the fidelity of depth estimates.</p><p>The best in class systems today utilize active IR or Laser depth systems to track the environment.</p><p>Ideally these systems will work faster than the display systems do, however when we look at size and power requirements of existing best in class systems, they are not yet able to do so in a very compact form that would go into the idealized form factors described above. It may be ideal to use two passive cameras (stereo RGB) for tracking in our idealized form factor, as they are small and need very little power, however advances in shrinking active tracking systems hold significant promise.</p><blockquote>BOTTOM LINE ENVIRONMENTAL TRACKING: An AR system needs to be able to accurately maintain a virtual object’s placement in the real world, such that virtual objects convincingly and reliably stay where placed, regardless of where or how the viewer moves.</blockquote><h4>Content and applications</h4><p>Think of every application that you use today on your desktop or mobile device. Everything from spreadsheets to candy crush. Ok, now add an entirely new category of interactivity that you can have with real-world objects and places. Here’s a brief list of those:</p><p><a href="https://uploadvr.com/augmented-reality-use-cases-list/">The Gigantic List of Augmented Reality Use Cases</a></p><p>Everything from the User Interfaces to the 3D models, need to be created, curated and placed into the world in a way that people using AR will encounter, interact and collaborate with them in the right context. This content needs to be basically everywhere for it to work at scale such that 1. You don’t need to rely on another device for your data and 2. It actually makes your daily life easier or better.</p><p>Oh, and that means that the content can’t be a bunch of spam, AR popups and general noise that overloads your senses.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/936/1*QVLVXclywYGjzFnEFFqwLg.jpeg" /><figcaption>Please no. Image Courtesy Keiichi Matsuda’s Hyper-reality</figcaption></figure><p>Practically speaking it’s probably not necessary to replace every possible application on the planet in order for AR HMD to have smartphone level penetration. However there needs to be enough content and applications to replace most of the core functionalities of smartphones: Messaging, Calling, Search, Navigation, Shopping and Video/Image playback.</p><p>This also means that, like every other computing platform, we need to build a strong community of developers, content creators, designers, hardware engineers and advocates, who will spread the word and build the experiences that people want.</p><blockquote>BOTTOM LINE CONTENT &amp; APPS: An AR system needs enough content and use cases that it replaces or enhances current functionality of digital systems used today on smartphones, desktops and laptops</blockquote><h4>Ubiquitous Mapping</h4><p>Speaking of having content everywhere and accessible to anyone, something that is massively overlooked, and rarely discussed in the AR community, is the need for a common <a href="http://support.esri.com/other-resources/gis-dictionary/term/georectification">georectified</a> AR environment.</p><p>If you remember <em>way back</em> to when PokemonGo came out this summer, you might recall throngs of people going to very specific places to catch certain Pokemon:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3hl1hDd5RkF_ro9MTgQ1ww.jpeg" /><figcaption>PokemonGo users swarming a park</figcaption></figure><p>That’s because the creators wanted you to feel like you had to actually interact with the Pokemon where they were in the “real-world.” The concept of contextual content is certainly not new. Many apps and products only work in certain geographic places (<a href="https://en.wikipedia.org/wiki/Geo-fence">geofencing</a>) and others require physically seeing some feature to activate certain content (most marker based AR). In practice, Niantic used geofencing to give Pokemon their homes, so it was possible to do at scale without big leaps in technology.</p><p>Imagine now, that you put on the perfect form factor AR HMD. It has great tracking capabilities and there is great content out there to be discovered. How then will the AR system know where the content you want to interact with is to high enough positional accuracy? GPS? Unfortunately <a href="http://www.gps.gov/systems/gps/performance/accuracy/">GPS has wide margins of error, averaging about 5 meters (16 feet) of accuracy.</a> It also fails indoors.</p><p>Well the answer is a ubiquitous re-localization map. Effectively this looks like a copy of the real-world, but with all of the virtual layers and content georectified inside of it. It also needs to be so good that a user simply needs to looking at an object and you will find the associated digital content that lives around it. For an AR HMD system to work seamlessly, it needs to find content both indoors and outdoors and serve it to the user with inch level accuracy.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZAQnCZPVs-bZc0fOvfVcWg.jpeg" /><figcaption>PointCloud Map of UCD optics Lab. Image Courtesy Ruth Kerr</figcaption></figure><p>This is very much tied into the tracking capabilities, but broadens the scope of the localization portion such that every user has a precise <a href="https://en.oxforddictionaries.com/definition/geolocation">geolocation</a> that maps to the georectified map.</p><blockquote>BOTTOM LINE FOR UBIQUITOUS MAPPING: To have AR everywhere, there needs to be a massive consistent and accessible 3D map, consisting of virtual content that is georectified to the physical world down to inch level accuracy. I will note that I think this is the most important piece of the AR puzzle and has the highest potential value.</blockquote><h4>Human Factors Engineering</h4><p>Something the AR community doesn’t have much research or practice in, are the human factors engineering problems for an AR environment. That’s not to say people aren’t thinking about it, but it’s not well worn territory.</p><p><strong>Gaze</strong></p><p>When we look historically at the mechanics of how people consume and create content, some interesting trends are revealed.</p><p>Since the invention of parchment, we have gazed downward at our content. Whether we are writing with pen, typing on typewriter, or reading and watching, this trend has held true forever.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qu2bQrsX_OOag18FG0vxxg.jpeg" /><figcaption>Always looking down</figcaption></figure><p>Certainly we gaze forward and around while walking, talking, driving and passively interacting with the world — for example watching theater or a movie — but nearly all interactions we have with digital or print content is while we are gazing downward. The reason for this is a mixture of social and ergonomic, but not worth getting into depth here.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/728/1*vTYvZcx0sLJWEfrFwbiRqA.jpeg" /><figcaption>Downward gaze is best for Ergonomics</figcaption></figure><p>Should you be checking your twitter feed while you are engaging in conversation with someone? Probably not, and it will be obvious that you aren’t actually listening to them. Nearly all proposed AR HMDs, follow in the wake of Virtual Reality HMDs, in that they assume the user will predominantly be forward gazing — so they put the bulk of information and user interface directly in front of the user.</p><p>This is misguided. It ignores the fact that VR and AR are extremely dissimilar when it comes to how long you are in the environments (a few hours vs all day), and what “natural” actually looks like in the different contexts of day to day life.</p><p><strong>Inputs</strong></p><p>There seems to be this silly idea that gesture input will dominate AR interactions and input.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FaZORupDeNhM%3Ffeature%3Doembed&amp;url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DaZORupDeNhM&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FaZORupDeNhM%2Fhqdefault.jpg&amp;key=d04bfffea46d4aeda930ec88cc64b87c&amp;type=text%2Fhtml&amp;schema=youtube" width="640" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/c02ce9066226d5c485d0d7921a2077d9/href">https://medium.com/media/c02ce9066226d5c485d0d7921a2077d9/href</a></iframe><p>Try this: Get a stopwatch out and start it. Now hold your arms out in front of you and see how long it takes before you get uncomfortable. You can even bend your elbows if you like. How long did you last? Probably not more than a few minutes. Now imagine doing this gesture all day long for everything. I personally don’t think it’s practical, comfortable or reasonable.</p><p>But wait, don’t we deal with real objects all day long with our hands? Well kind of. The actual amount of time you spend reaching, grasping and manipulating really depends on your job type, ability level and age. If you are a miner, construction worker, or other blue collar worker, you probably have years of repetition and strength built into your upper body — not so much for white collar workers, the elderly or the disabled.</p><blockquote>By proposing gesture based interactions as the core input mechanism for AR, we are telling white collar workers, the disabled and the elderly that to use these new computing devices they will have to gain strength in their shoulders, arms and chest.</blockquote><p>There is a lot of <a href="http://skepdic.com/woowoo.html">woo-woo</a> around the idea that forward gazing, gesture based interactions are “natural,” and that all current AR displays are mimicking the natural way we do things. It would be useful for those stating as much as well as future AR UX designers, to take a look at ground truth ergonomics and human factors in current daily life, rather than looking to science fiction for this.</p><blockquote>BOTTOM LINE FOR HUMAN FACTORS: AR needs to adequately factor in how humans naturally interact with digital content and the real-world in different contexts, so that we deliver the most seamless blending of the real and virtual worlds in the most comfortable ways.</blockquote><h4>Privacy and Security</h4><p>What happens at the societal level when everyone has a camera array and depth sensor on their face that is constantly recording and mapping? How do our systems distinguish between private and public spaces? How do we protect private data in a pervasive mapping and tracking environment?</p><p>There are well documented cases of people being attacked for wearing a form of HMDs. Kyle Russell being <a href="http://mashable.com/2014/04/13/google-glass-wearer-attacked/#.QLFKZhmFsqW">attacked for wearing his Google Glass</a> was a well publicized example. Steve Mann’s <a href="https://techcrunch.com/2012/07/16/augmented-reality-explorer-steve-mann-assaulted-at-parisian-mcdonalds/">assault at a McDonalds for wearing his EyeTap</a> was also well publicized. It’s clear that people aren’t broadly ready to interact with such systems.</p><p>This is not a new concern, nor is it intractable. There are solutions proposed, but practically, we will probably see technologies deployed to the public before we have robust solutions. We shouldn’t ignore these concerns completely, but there is a trend where we “figure that part out later” once we have sufficient penetration of these technologies. I generally agree with that approach, as we have adapted so far (citation needed) to the massive growth of social media and all that has come with it. It’s certainly worth us having a public conversation about what the risks and benefits really are as AR is potentially orders of magnitude more impactful.</p><p>This is probably the hardest problem in the group of problems, as it takes a shift in how we collectively view this kind of technology and the trade offs in the broader scope of society. I fear that we will rush into this and see a backlash if not done correctly, which could set us back for widespread adoption.</p><blockquote>BOTTOM LINE FOR PRIVACY AND SECURITY: We in the AR community need to understand and explain the benefits and drawbacks to pervasive AR — with the proof being real-world usage.</blockquote><h4>Summary</h4><p>For an AR system to become the ubiquitous computing device, the form factor needs to conform to how humans contextually interact with digital content and the real-world.</p><p>It should be similar in form to common eyeglasses, with display fidelity mimicking the real-world and the ability to be used all day uninterrupted.</p><p>It needs to convincingly display virtual content in the real-world, with a sufficient number of applications and content, accessible everywhere regardless of environment, such that it can replace the most common computing use cases with better quality than alternatives.</p><p>Finally, and critically, it needs to provide so much value that any social drawbacks are outweighed by the overall measurable improvement in quality of life.</p><ul><li>*I intentionally do not use the term Mixed Reality in this context as it is being used specifically by Magic Leap, in an attempt to dissociate itself from the historically poor implementations of Augmented Reality. The term Mixed Reality defines the <a href="http://etclab.mie.utoronto.ca/publication/1994/Milgram_Takemura_SPIE1994.pdf">Virtuality Continuum introduced by Paul Milgram in his 1994 paper</a></li></ul><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fupscri.be%2F1af398%3Fas_embed%3Dtrue&amp;url=https%3A%2F%2Fupscri.be%2F1af398%2F&amp;image=https%3A%2F%2Fupscri.be%2Fmedia%2Fform.jpg&amp;key=d04bfffea46d4aeda930ec88cc64b87c&amp;type=text%2Fhtml&amp;schema=upscri" width="800" height="400" frameborder="0" scrolling="no"><a href="https://medium.com/media/43dd20747f6cad50f2137710ef578e54/href">https://medium.com/media/43dd20747f6cad50f2137710ef578e54/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=275db4aabe7" width="1" height="1" alt=""><hr><p><a href="https://medium.com/virtual-reality-pop/heres-what-it-s-going-to-take-for-augmented-reality-to-take-over-the-world-275db4aabe7">Here’s What It’s Going to Take For Augmented Reality to Take Over The World</a> was originally published in <a href="https://medium.com/virtual-reality-pop">Virtual Reality Pop</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The AI revolution will be different from the microprocessor revolution]]></title>
            <link>https://medium.com/@andrewkemendo/the-ai-revolution-will-be-different-from-the-microprocessor-revolution-cb417d324203?source=rss-8a0d3c1ad4b4------2</link>
            <guid isPermaLink="false">https://medium.com/p/cb417d324203</guid>
            <category><![CDATA[software-architecture]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[startup]]></category>
            <dc:creator><![CDATA[Andrew Kemendo]]></dc:creator>
            <pubDate>Wed, 27 Jul 2016 00:02:35 GMT</pubDate>
            <atom:updated>2016-07-27T00:31:29.214Z</atom:updated>
            <content:encoded><![CDATA[<p>This post stems from a conversation I had recently on <a href="https://news.ycombinator.com/item?id=12168228">Hacker News</a> about a <a href="http://en.collaboratory.de/w/Power_in_the_Age_of_the_Feudal_Internet">recent article from Bruce Schneier</a>.</p><p><strong>Executive Summary: <em>Reinforcement Learning (RL) is going to reshape how we build software and businesses in the near future. This will make it increasingly difficult to start, compete with and disrupt incumbents the same way the microprocessor revolution did</em></strong></p><p>The past few years of progress in machine learning, specifically reinforcement learning have been exciting with some breathtaking results that have been well publicized. Prior to a year ago however I never really thought Q-learning or Policy Gradients were huge breakthroughs that would put us on a truly revolutionary path; more that they were steps in the right direction. My mind is now changed.</p><p>Stated simply: I believe a Reinforcement Learning Architecture will replace explicit instruction based Architecture for building software within a decade.</p><p>Reinforcement Learning tasks rely on ridiculous amounts of data. Whereas with traditional software architecture, where you accomplish tasks through explicit task instruction, RL trains for tasks based on millions of tests through a reward system. Most importantly once you have trained it to some minimum level, if you deploy it correctly, then it should continue improving — so long as you bake feedback into the UX. Imagine that instead of telling excel what to do, you and every other user will have a conversation with excel, improving the system incrementally.</p><blockquote>Imagine that instead of of telling excel what to do, you and every other user will have a conversation with excel, improving the system incrementally.</blockquote><p>These are simplifications of course, and reducing complex decision systems to a series of reward goals is not trivial but the general idea that software architecture is going to move from “do this exactly” to “improve constantly” I think is valid.</p><p>Perhaps for very simple tasks, small teams will be able to build the data they need to train some early systems. I think however that the more likely outcome is that massive companies with huge data sets and throngs of amazing ML people like Google etc…will be the ones who can take advantage of these systems. Google is already doing it with their power systems and untold other number of projects nobody knows about that would have otherwise been done by a niche company.</p><p>This has some pretty staggering implications on how small companies will be able (or not) to compete with or disrupt existing players going forward. People have been saying that “Big Data” is dead, but they hadn’t considered how critical that data is for RL.</p><p>The microchip revolution democratized computing by giving hobbyists and hackers the power that only massive companies had. RL Software architecture can’t be democratized without massive data sets. And only a handful of companies have such sets. Unless these companies open source their data sets (they won’t — for many reasons, including legal ones) then we as the small timers have to build our own — a challenge much more massive than it looks. By the time we can, the big players will be able to scoop us up or use their existing scale to out compete.</p><p>Disruptive innovation historically been niche players finding gaps in processes (or markets) and then moving faster and more iteratively than big players to patch (or fill) them. If the faster and more iterative path forward is by leveraging Exabytes of data and hundreds of Machine Learning engineers — then the big players are going to have a massive advantage from here on out. And that is a big deal.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cb417d324203" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Startups need to start valuing themselves on impact not valuation]]></title>
            <link>https://medium.com/@andrewkemendo/startups-need-to-start-valuing-themselves-on-impact-not-valuation-8276690ae862?source=rss-8a0d3c1ad4b4------2</link>
            <guid isPermaLink="false">https://medium.com/p/8276690ae862</guid>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[venture-capital]]></category>
            <dc:creator><![CDATA[Andrew Kemendo]]></dc:creator>
            <pubDate>Thu, 21 Jan 2016 20:16:41 GMT</pubDate>
            <atom:updated>2016-01-21T20:16:41.813Z</atom:updated>
            <content:encoded><![CDATA[<p>The increasingly fever pitched gnashing of teeth over the past year about a possible tech bubble doesn’t seem to be slowing down this year.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/862/1*_KTA5wdnEvR4l6fcYQWltA.png" /></figure><h4>Mythical animals get all the press</h4><p>The hot topic of course are the issues surrounding “Unicorns” and whether their valuations have any relation to reality or if they are just funky paper exercises. This was summarized brilliantly with this example:</p><blockquote>VC: We will buy .0000001% of your company for $1. Welcome to the Unicorn club</blockquote><p>Obviously this satirizes the reality, but the overall point that private valuations seem to be divorced from the long term value of a company still stands.</p><p>That begs the question though, by what metrics do we call a Unicorn, or any startup for that matter, successful?</p><p>The answer of course depends on who you ask. Ask their investors and they will say, “When they return X multiple of my money.” Ask the local chamber of commerce and they will say “When they bring X jobs to our community.” Ask their users and they will say “As long as I’m getting value out of it, it’s successful.”</p><p>The problem is that people focus almost exclusively on the VC’s answer.</p><blockquote>The Startup world focuses too much on valuing companies only by their financial worth</blockquote><p>There are a lot of reasons for that probably. For one, it makes great news. Forbes, BI, etc… hang their hat on that metric. This just creates a self licking ice-cream cone where startups focus on valuation to get press, press covers it and you start the process over again.</p><p>It’s also really easy to measure, and takes relatively little effort to understand. It also really impacts fundraising. The more money you have the more money people want to give you. For all these reasons financial value has won the day when evaluating startups - the same as any other company no matter what they are producing.</p><h4>So who is valuable?</h4><p>According to VC and business tech press, the highest financial goal of a startup is a big exit, with an IPO being the ultimate validation. Both GoPro and Square achieved this vaunted status. So is that the end? Not by a long shot and the press is flogging them:</p><p>GoPro: IPO in 2014 at $24. <a href="http://fortune.com/2015/11/12/gopro-shares-fall-below-ipo-price/">Today they are trading at $10.</a></p><p>Square: <a href="http://www.marketwatch.com/story/square-falls-below-its-ipo-price-while-twitter-stock-reaches-record-low-2016-01-20">Trading below its IPO price</a></p><p>Both are “under IPO” price, so according to the financial valuation metric, as of today, they are duds in terms of what they brought the world. You would have been better off investing in the S&amp;P over the same time period. So why should we care about them anymore?</p><h4>Why we should value “startups” differently</h4><p>We care about GoPro and Square because they are totally changing the way people work and play by bringing fantastic new technologies to the masses.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TVt8g4lsSPwPBGsFr5zB0Q.jpeg" /><figcaption>You can’t skydive without a GoPro anymore Source: stohke.com</figcaption></figure><p>That’s the difference between Tech startups and every other company in the world and how we should be valued. We should value ourselves based on our ability to transform for the better how people work, live and play.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/675/1*wMfjhXIG3Abd8JNEGFvF-Q.jpeg" /><figcaption>Square transforms payments for small sellers Source: NY Times</figcaption></figure><blockquote>Tech startups should be valued based on their ability to transform for the better how people work, live and play.</blockquote><h4>Great startups really are different</h4><p>At the same time as all of this talk about valuations and bubbles, there has been lament that “Startups” and “Silicon Valley” itself is no longer comprised of scrappy, small, moonshot companies, creating world changing technologies. Whether or not that narrative is historically true or not, <a href="http://www.scaruffi.com/politics/sil18.html">and there is good evidence it isn’t</a>, this lamenting makes the point that small, privately owned, high growth, high technology companies should be about more than just making money. They/We should be at the forefront of technological commercialization.</p><p>If we aren’t, then what is the point of taking the risk? We as founders and startup employees are probably all better off working at large comfortable mega-corporations with 50+ years of longevity. The risks of starting a high-technology company are massive and 90% of startups die in the first two years. We win by pushing the limits of technology and process in a way that can’t be done through traditional business.</p><p>There are certainly cynical actors that look for a quick buck or take advantage of the startup ecosystem with less than revolutionary technologies and shady business practices. Unfortunately that is just part of business in a fast growth industry. That doesn’t mean we should resign ourselves to being valued primarily by our financial metrics and just makes those that do look bad when they inevitably crash and burn. The trouble comes when those groups are the loudest voices and their implosions affect the rest of the ecosystem.</p><p>Focusing on bringing massive value to a broad range of consumers, quickly and in new and innovative ways that large gregarious companies simply can’t, is how the best startups prevail. That doesn’t mean we ignore revenue or financials, fundamentals are of course necessary, only that those aren’t the most important metrics worth evaluating.</p><p>So for the good of the community, lets start measuring ourselves on how we are tangibly making people’s lives easier and better, and how we move technology forward to bring about the future that we want to live in.</p><p>Andrew Kemendo is the co-founder and CEO of <a href="https://www.pair3d.com/">Pair</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8276690ae862" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Augmented Reality glasses will fail without content infrastructure]]></title>
            <link>https://medium.com/@andrewkemendo/augmented-reality-glasses-will-fail-without-content-infrastructure-dfe35817feb?source=rss-8a0d3c1ad4b4------2</link>
            <guid isPermaLink="false">https://medium.com/p/dfe35817feb</guid>
            <dc:creator><![CDATA[Andrew Kemendo]]></dc:creator>
            <pubDate>Sun, 28 Jun 2015 03:37:28 GMT</pubDate>
            <atom:updated>2016-12-09T19:29:25.915Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Enterprise and consumer interest into AR glasses and heads up displays are booming. This is fantastic for the future of computing — but is crowding out investments in application and content infrastructure.</em></p><h4>Applications drive adoption</h4><p>If you look at the history of any major new computing technology, widespread adoption always happened after applications and content became available that proved the value of the technology to the average user. Technologists refer to these applications and content as “<a href="https://en.wikipedia.org/wiki/Killer_application">killer applications</a>” and they drive technologies to cross the boundary between hobby users and wide scale adoption.</p><p>As an example, the personal computer didn’t seriously take off until breakthrough applications such as VisiCalc (<a href="https://visidraft.com/">Visidraft</a> is a homage to that name by the way) and Wordperfect proved the value of the platform for the enterprise. These functional applications, with improved user interfaces, proved valuable to consumers and as a result became more broadly adopted.</p><p>Similarly, it wasn’t until the mobile application infrastructure boomed with app stores that the value of having computers in a compact mobile form was recognized. One would be hard-pressed to find a single killer app for smartphones but I would venture a guess that it was a tossup between robust email/messaging with Blackberry (again, originally adopted primarily by enterprise) and the camera integration with the iPhone which drove wider consumer adoption. The deluge of productivity/connectivity applications, like maps, added fuel to that fire.</p><h4>How AR content is delivered</h4><p>The theoretical use cases of Augmented Reality technologies dwarf the use cases of pretty much all previous computing interfaces. At one extreme, we have a simple 2D overlay of information in our line of sight similar to what google glass delivered. At the other extreme is virtual elements convincingly embedded into our real environment through a “hands free” visual interface.</p><p>Regardless of where your content lies on this spectrum, the one thing all have in common is that the content should be responsive to the real environment in which you exist. That is to say, content should seamlessly add value to your visual landscape by overlaying or embedding content that would otherwise not be accessible.</p><p>To do this, content needs to be tied or linked to the environment in a way that is simple for users. The most utilized process for this real-virtual link to date has been with AR Markers, printed images which are linked to 3D content which “pop out” of the environment. The most successful AR companies have exploited this mechanism through print advertising, giving additional interactivity to print advertising.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/1*ezIYkNxb45JUSmIvN9lIVA.jpeg" /></figure><p>For AR to be widely adopted however, there needs to be a content infrastructure that provides users with real added value — something that makes their lives easier or better, like local data, object recognition or navigation through a dead simple interface. For complex 3D and locational data, we can’t expect users to generate this content on their own, so it is up to developers and companies to build this out for them.</p><p>This content must be “pulled” from the environment for the user interface to work — we cannot rely on end users to place their own virtual elements in the environment. That would be backwards. This is where the marker based approach that is so common fails, as plastering the world with images is neither practical nor desirable.</p><h4>Putting the interface before the applications</h4><p>To do what is described above however, needs robust markerless AR technology. To date, investment in markerless AR has been sparse and limited to companies already well established in the AR market or in government laboratories. Prior to their recent acquisition by Apple, Metaio had been developing a few markerless AR approaches which allowed for embedding content to 3D maps of environmental objects — a great development which is no longer available to developers.</p><p>Markerless AR with the ability to draw data from the existing environment is the only realistic way to bring content to users in a seamless and simple fashion. The content and application infrastructure then is the software, and to some degree hardware, that makes Markerless AR “just work.” These are techniques that lie in the computer vision and deep learning world devoted to understanding the world visually: SLAM, PTAM, Semantic object recognition, etc…that see the bulk of their research coming from universities and larger corporate labs. Largely though, companies investing in these approaches are applying them to self-driving cars, robotics and other autonomous navigation problems, not to the AR content infrastructure problem.</p><blockquote><strong>The content and application infrastructure then, is the software, and to some degree hardware, that makes Markerless AR “just work.”</strong></blockquote><p>The hardware to support this middle layer is already available. Tablets and smartphones can already support what is needed to make these interactions possible. At Visidraft, we are proving that you can run real time monocular SLAM to build dense geo-rectified point clouds from mobile devices already. These devices are the natural stepping stone to better glasses, especially considering that these devices are still relatively new, having only seen market dominance for less than a decade. Even then, many of the AR glasses on the market can support this content infrastructure. Critically the markerless AR tracking and content creation that would give these devices breakthrough experiences have not seen significant investment.</p><p>The current glut of investment in AR is ignoring this critical middle layer and focusing on making better display technologies.</p><h4>Towards building a robust AR content infrastructure</h4><p>Building the markerless AR content and application infrastructure that just works right out of the box is really hard. Exceptionally hard actually, and needs capital and investments at the level the AR hardware is seeing. Without this layer, all the advances in hardware won’t be able to bridge the user experience needed for broad adoption of AR.</p><p>Investors and companies interested in AR should be taking much more interest in developing the content infrastructure of AR through markerless AR, and divest in the hardware for the time being. Making AR widespread is as much an educational exercise for users and consumers as it is a technological problem. We have the hardware and use cases today to start bridging those gaps with unique and compelling applications. As users demand more AR in their world, the demand for better wearable AR devices will grow, and the transition from phones/tablets to glasses will be that much more seamless.</p><p><em>Andrew Kemendo (</em><a href="https://twitter.com/AndrewKemendo"><em>@AndrewKemendo</em></a><em>) is the founder and CEO of </em><a href="http://twitter.com/pair3d"><em>Pair</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dfe35817feb" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[First time founders and the trouble with founder friendly terms]]></title>
            <link>https://medium.com/@andrewkemendo/first-time-founders-and-the-trouble-with-founder-friendly-terms-e804a9783b5c?source=rss-8a0d3c1ad4b4------2</link>
            <guid isPermaLink="false">https://medium.com/p/e804a9783b5c</guid>
            <dc:creator><![CDATA[Andrew Kemendo]]></dc:creator>
            <pubDate>Sun, 31 May 2015 20:32:58 GMT</pubDate>
            <atom:updated>2015-05-31T20:32:58.615Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Disclaimer: I am a first time founder, not an investor. My data is based on discussions with dozens of other early stage, software startup founders on the east coast. My experience could simply reflect the difference in investing climate between east and west coasts.</em></p><p>I recently had a<a href="https://twitter.com/AndrewKemendo/status/604989011212550144"> twitter conversation</a> with Mark Suster and Arlo Gilbert about Mark’s recent post <a href="http://www.bothsidesofthetable.com/2015/05/30/one-simple-paragraph-every-entrepreneur-should-add-to-their-convertible-notes/">regarding Convertible notes</a>.</p><p>This is the main point,<strong> in my words</strong><em>,</em> that I took from the post:</p><blockquote><em>Founders should not do a convertible note deal with investors who will not agree to language limiting </em><a href="http://siliconhillslawyer.com/2015/04/29/capped-convertible-notes-liquidation-overhang/"><em>liquidation preference overhang</em></a></blockquote><p>Implementing language like this would be fair for investors and fair for founders. Here is the rub: every investor knows about liquidation preference overhang and many of them bank on it being how they make their money. In fact, the angels and seed stage VCs that we have talked to, all explicitly state that this is how they benefit from early stage deals. In their words, the valuation jump from from seed to A round is their payoff from taking such risks early on</p><p>Having been through only one very small, but critical round of funding I can tell you that we fought against term sheet markups with multiple preferences and ratchets. Our well respected and high priced lawyers told us that multiple preferences were standard on a seed convertible note and that most of these terms didn’t really matter in a first seed round. Having done my homework I pushed against those and we ended up getting good terms that everyone seemed happy with — but that doesn’t mean they weren’t going to try.</p><p>This gets to my larger point that first time, or early stage founders do not have the leverage or funding options to push back on terms that are not friendly. Often the BATNA for these folks is, take out a personal loan. So when I read the following quote I had to think that it doesn’t understand the funding environment that we are in.</p><blockquote><em>So here’s what every entrepreneur should add immediately to any convertible note they are trying to raising money against and run from any angel who won’t agree to this.</em></blockquote><p>Run where? When I talk to my fellow early stage east coast founders, the majority aren’t beating away founder friendly term sheets. Even seed stage companies with revenue and traction raising relatively small amounts are giving away board seats and agreeing to multiple preferences because they have nowhere else to go. For founders, these deals can be make or break. For investors they can hold out and the only downside is slightly lower yield. So there is asymmetry in needs which means founders have little leverage.</p><blockquote><em>And please don’t let any angel try to convince you they deserve more for the early risk. The early risk is why they get a cheaper price.</em></blockquote><p>This is exactly what a lot of them say though, and early founders don’t have the history or credentials to fight them on it (aside from pointing to posts like Mark’s). So the options in many cases are, close up, or take worse terms.</p><p>The immediate response to these points is always some mix of “well then you aren’t good enough” or “go back to bootstrapping till you get better terms” or something that points to the founders or the company not being ready for good terms. Granted there are a lot of those, but I am talking specifically about trends I have seen with more than one startup who have bootstrapped themselves to traction/revenue and have teams comparable with those well funded elsewhere.</p><p>The bottom line here for me is that for first time seed stage founders, most don’t seem to have the leverage or the network to be able to push back on terms that aren’t particularly friendly. To say to these founders that all we need to do is fight back on them, is ignoring the ecosystem that they find themselves in. Given the startup to investor ratio, many of us only have personal connections to a few, and if none of them are willing to agree to language that is friendly then we end up being stuck.</p><p>We need something better than pointing to a handful of (very valuable) blogs if we are going to be able to push back against terms unfriendly to founders. I don’t really know how else to prevent taking bad terms other than falling back to a defensive position and hoping you can squeeze enough juice out to not need any outside money before shutting down.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e804a9783b5c" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>