<?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 KPCB Edge on Medium]]></title>
        <description><![CDATA[Stories by KPCB Edge on Medium]]></description>
        <link>https://medium.com/@kpcb_edge?source=rss-6878ba49cb9c------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*VDnB1ajBuRHHKS84e9SuhQ.png</url>
            <title>Stories by KPCB Edge on Medium</title>
            <link>https://medium.com/@kpcb_edge?source=rss-6878ba49cb9c------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 12:07:02 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@kpcb_edge/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[Looking Beyond]]></title>
            <link>https://medium.com/@kpcb_edge/looking-beyond-e39aef4148d2?source=rss-6878ba49cb9c------2</link>
            <guid isPermaLink="false">https://medium.com/p/e39aef4148d2</guid>
            <category><![CDATA[venture-capital]]></category>
            <dc:creator><![CDATA[KPCB Edge]]></dc:creator>
            <pubDate>Mon, 31 Jul 2017 20:44:49 GMT</pubDate>
            <atom:updated>2017-07-31T20:44:49.298Z</atom:updated>
            <content:encoded><![CDATA[<p>The last two years have been some of the most exciting and unique times in the world of technology and venture capital. The unprecedented rise of cryptocurrencies, step function improvements in computer vision from deep learning, and massive investments into VR and AR are just some of the changes that we suspected, but could never have anticipated the sheer magnitude of.</p><p>We started Edge to invest in these emerging areas of technology, and have been lucky to work with an incredible group of founders who are all working at the edge of these advances. It is, however, difficult to not want to participate more directly in the incredible progress that’s taking place around us, and to that end, we are each moving on to building new projects in these spaces.</p><p>For now, new investments in seed stage companies will come from our venture team at KP, and one of our very talented developers in residence, Will, has stepped up to take the baton in continuing to build the software tools that are used daily by our founders and investing teams.</p><p>To the entrepreneurs, co-investors and everybody else who helped bring Edge to life over the past few years, you know who you are, a huge thank you! We’re excited to share more about our new projects in the coming months.</p><p>- Anjney, Ruby and Roneil</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e39aef4148d2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Enabling Analysis of Private Data]]></title>
            <link>https://medium.com/@kpcb_edge/enabling-analysis-of-private-data-a6229c7d72bb?source=rss-6878ba49cb9c------2</link>
            <guid isPermaLink="false">https://medium.com/p/a6229c7d72bb</guid>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[encryption]]></category>
            <category><![CDATA[privacy]]></category>
            <dc:creator><![CDATA[KPCB Edge]]></dc:creator>
            <pubDate>Thu, 06 Oct 2016 16:00:42 GMT</pubDate>
            <atom:updated>2016-10-06T16:00:42.224Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1004/1*lFrYkGM72BYyI9x0IVOqjw.png" /></figure><p>Since Neal Stephenson published Cryptonomicon in 1999, cryptography’s usage has spread far beyond hacker and cypherpunk communities through HTTPS, end-to-end encrypted messengers (ex. WhatsApp), and other easy-to-use technologies. However, for most companies, encrypting user data is fundamentally at-odds with their business interests. Facebook, Google, and many other consumer tech companies depend on user data to target ads and provide recommendations.</p><p>Average consumers care more about privacy today than they did in 1999 too. Telegram, the encrypted messenger app, <a href="https://techcrunch.com/2016/02/23/encrypted-messaging-app-telegram-hits-100m-monthly-active-users-350k-new-users-each-day/">reached 100 million monthly active users in February, who sent more than 15 billion messages a day</a>. <a href="https://telegram.org/">Telegram’s marketing</a> promotes its privacy, security, and encryption features prominently as selling points, and those features are arguably its <a href="https://telegram.org/faq#security">primary differentiators as compared to other messaging apps</a>. 89% of surveyed consumers said they would avoid companies that do not protect their privacy:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/908/1*RVxB4jp3zQF16dCDsnVlrg.gif" /><figcaption>Excerpted from <a href="https://www.truste.com/resources/privacy-research/ncsa-consumer-privacy-index-us/">https://www.truste.com/resources/privacy-research/ncsa-consumer-privacy-index-us/</a></figcaption></figure><p>This creates a fundamental conflict: users want more privacy, but companies, in the interest of building better products and increasing their defensibility against competitors, are incentivized to save everything they can. <a href="https://www.eff.org/deeplinks/2016/09/googles-allo-sends-wrong-message-about-encryption">The EFF recently called out Google’s Allo</a> for not enabling encryption by default, saying “A more responsible messaging app would make security and privacy — not machine learning and AI — the default.”</p><p>However, <strong>we don’t believe consumer privacy and data access / analysis are mutually exclusive.</strong> Using cryptographic techniques and clever system design, it’s possible to enable computation and analysis on data without revealing identifying information or, in some cases, any data at all.</p><p>The techniques we’ve seen broadly fall under three categories: differential privacy, client-side computation, and homomorphic encryption / secure multiparty computation.</p><p><strong>Differential privacy</strong> allows for formal analysis of what an attacker could learn from a data set, and helps reduce what can be learned while preserving accuracy of aggregate analysis. Randomizing user IDs, slightly perturbing numerical values, and injecting noise into data sets are all examples of differential privacy techniques. However, privacy issues have come up when this is done carelessly: in 2007, Netflix released a data set with anonymized user IDs, but <a href="https://www.cs.utexas.edu/~shmat/shmat_oak08netflix.pdf">researchers were able to de-anonymize some of the data by cross-referencing public IMDB ratings</a>.</p><p><strong>Client-side computation</strong> and edge computing, when used correctly, can also allow data analysis while preventing access to plaintext data. Edge computing is not a new idea, but using it as a means to prevent data disclosure is more recent. Apple, for example, <a href="http://highscalability.com/blog/2016/6/20/the-technology-behind-apple-photos-and-the-future-of-deep-le.html">trains neural networks on external/public data sets, then deploys those to phones to be run on local data</a>, giving users valuable insights without Apple being able to see their data. Of course, this works for computation on a specific user’s data, but this can also be extended to enable aggregate analysis; user devices could be asked to produce summary information about their local data, which can then be merged server-side with summary data from others, <em>a la</em> mapreduce. This idea has been <a href="https://www.microsoft.com/en-us/research/wp-content/uploads/2013/12/Accelerating-RNN-training.pdf">used to speed up neural network training</a> in the past, so one could conceivably train a single, aggregate neural network without being able to access the training data set on user devices.</p><p><strong>Homomorphic encryption</strong>, despite being mathematically daunting, is conceptually simple: it allows computation to be done on encrypted data, producing an encrypted result, which can then be decrypted using the original encryption key(s), producing a result equivalent to doing the calculation on plaintext data. <strong>Secure multi-party computation</strong>, a closely related area, can enable similar results for multi-party problems, with the aim of computing a joint result from private data (ex. <a href="https://en.wikipedia.org/wiki/Yao%27s_Millionaires%27_Problem">Thomas Yao’s millionaire’s problem solution</a> allows the inequality a &gt;= b to be computed without revealing a and b). A fully homomorphic system <a href="http://crypto.stackexchange.com/questions/3512/homomorphic-encryption-special-case-of-multi-party-computation">can be used to perform secure multi-party computation</a>.</p><p>Unfortunately fully homomorphic encryption, which enables any computation that could be done on cleartext to be done on ciphertext, is not yet fast enough to feasibly be used on larger data sets. However, there are quite a few examples of partially homomorphic encryption schemes (which enable only some operations to be done on encrypted data) that could be used today for specific use cases. For example, homomorphic encryption could enable <a href="http://security.stackexchange.com/questions/35813/homomorphic-encryption-used-for-e-voting">vote tallying without revealing individual votes</a> or <a href="https://medium.com/@Numerai/encrypted-data-for-efficient-markets-fffbe9743ba8">open up private data sets for use in machine learning</a>.</p><p>These ideas can also be trivially combined: <a href="http://highscalability.com/blog/2016/6/20/the-technology-behind-apple-photos-and-the-future-of-deep-le.html">Apple uses differential privacy to gather data to train machine learning models, then deploys them to run locally on phones</a>. To shift storage/computation burden onto the server from user devices, one could combine edge computing and homomorphic encryption; user data could be analyzed server-side, and the analysis results could be sent to the client to be decrypted and returned to the server.</p><p>We believe there are significant opportunities to use privacy-protecting techniques to enable aggregate queries, machine learning, ad targeting, and much more on private data. Consumer and enterprise users have shown that they care about privacy, and that they’re willing to use services whose primary selling point is privacy and security.</p><p>While these techniques can be used by any company storing user data, we’re keenly interested in folks using these techniques to tackle previously-unsolvable problems (ex. analysis of data sets which are classified, HIPAA-protected, etc.) If you or anyone you know is working on a product using privacy-protecting techniques similar to the ones described, we’d love to chat!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a6229c7d72bb" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Xiaomi Mi Band & the rise of affordable wearables]]></title>
            <link>https://medium.com/@kpcb_edge/xiaomi-mi-band-the-rise-of-affordable-wearables-76caa3ab791a?source=rss-6878ba49cb9c------2</link>
            <guid isPermaLink="false">https://medium.com/p/76caa3ab791a</guid>
            <category><![CDATA[wearables]]></category>
            <category><![CDATA[health-technology]]></category>
            <category><![CDATA[mi-band]]></category>
            <category><![CDATA[venture-capital]]></category>
            <dc:creator><![CDATA[KPCB Edge]]></dc:creator>
            <pubDate>Mon, 18 Jul 2016 15:45:37 GMT</pubDate>
            <atom:updated>2016-07-18T15:45:37.046Z</atom:updated>
            <content:encoded><![CDATA[<p><em>This is the second in a series of posts about Edge’s recent visit to China.</em></p><p>Asked to name the top sellers in wearables, names that roll off the tongue might be Fitbit, Jawbone, and the Apple Watch. Yet the third highest-selling wearable in the world is none of the above — in 2015Q2, sitting just behind Fitbit with 4.4M bands and Apple Watch with 3.6M units sold, the <strong>Xiaomi Mi Band</strong> sold 3.1M devices. So where did the Mi Band come from, what’s driving its growth, and what are the implications of its popularity? We’ll cover our thoughts on these topics in this post.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/0*KWLtumI9y431YzD6." /><figcaption>Xiaomi Mi Band</figcaption></figure><p>Let’s start with Xiaomi, the maker of the Mi Band. Xiaomi was founded in 2010 in Beijing, and was selling 70.8 million smartphones a year by 2015. Now valued at $45B (although perhaps <a href="http://fortune.com/xiaomi-business-china">shakily</a>), Xiaomi didn’t open its first physical store until 2015, relying on its insanely popular online flash sales. Today, their profits are increasingly derived from their app marketplace, phone accessories, smart home gadgets, and many other consumer electronic devices.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ASiD6nZBza6hbuco." /><figcaption>The array of products offered inside a Xiaomi store we visited in Beijing.</figcaption></figure><p>Xiaomi released the Mi Band in July 2014. The Mi Band has a silver core tracker that is inserted into a wristband and comes with the standard features consumers have come to expect in wearable devices: activity tracking, sleep tracking, and heart rate measurement. (Only the slightly more expensive version of the Mi Band has non-continuous heart rate measurement, meaning the user chooses when to measure heart rate in the app.)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*nV7a39TWNfTyeHIj." /><figcaption>The standard feature set offered by the Mi Band.</figcaption></figure><p>So then why is the Mi Band so popular? Where the Mi Band differentiates itself is on <strong>price </strong>and <strong>battery life</strong>. Without heart rate sensing, it costs 69 RMB or roughly <strong>$10.66</strong>, and with heart rate sensing, it costs 99 RMB or roughly <strong>$15.29</strong> (now listed as $14.99 on their <a href="http://www.mi.com/en/miband/">site</a>). The battery lasts on average <strong>30 days</strong>. You can see how this compares to other popular wearables in this chart:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ab_kC9uDARjGScwUjYIs8A.png" /></figure><p>This June, Xiaomi announced the next generation of the Mi Band, with a 0.42” OLED screen that navigates between three different screens for time, steps, and heart rate data. The Mi Band 2 is priced out 149 RMB or roughly <strong>$23.28</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/0*M4BXjC19BIGd7luR." /><figcaption>Mi Band 2</figcaption></figure><h4><strong>Takeaways</strong></h4><p>We believe the Mi Band and Mi Band 2 are important because they make it increasingly affordable for individuals to access their own personal wearable data stream. Xiaomi has now opened up online sales of the Mi Band in the US (although their website is currently out of stock) and plans to do so in other countries. And although not as top-selling, there are other similarly featured devices at sub-$30 prices, like the <strong>Misfit Flash</strong>, which tracks activity and sleep, running on a rechargeable 6 month coin battery for $29.99.</p><p><strong>So what happens in a world where basic wearables are so affordable that they could be given away for free?</strong> What types of businesses could be built with the strategy of giving consumers wearables in order to do something useful with the data? And of course, what will incentivize consumers to keep wearing them? We’ve already invested in an amazing company called <a href="http://www.datamindedsolutions.com/"><strong>Data Minded Solutions</strong></a> that pulls in data from wearables and other connected medical devices and apps into dashboards with insights for providers treating chronic patients. If you’re working on a company answering these types of questions, we’d love to <a href="https://www.kpcbedge.com/team">chat</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=76caa3ab791a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Our focus areas: Robotics]]></title>
            <link>https://medium.com/@kpcb_edge/our-focus-areas-robotics-4ed4afe761a?source=rss-6878ba49cb9c------2</link>
            <guid isPermaLink="false">https://medium.com/p/4ed4afe761a</guid>
            <category><![CDATA[tech]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[robotics]]></category>
            <dc:creator><![CDATA[KPCB Edge]]></dc:creator>
            <pubDate>Wed, 08 Jun 2016 15:07:27 GMT</pubDate>
            <atom:updated>2016-06-08T15:07:27.521Z</atom:updated>
            <content:encoded><![CDATA[<p><strong><em>“The world of A.D. 2014 will have few routine jobs that cannot be done better by some machine than by any human being. Mankind will therefore have become largely a race of machine tenders.” — Isaac Asimov, 1964, ‘Visit to the World’s Fair of 2014’</em></strong></p><p>Asimov’s <a href="http://www.nytimes.com/books/97/03/23/lifetimes/asi-v-fair.html">vision</a> may still be a few years away, but he was uncannily correct about the direction of <strong><em>routine</em></strong> jobs in the future. Where Asimov and we differ, however, is on our view that this will be <strong><em>liberating</em></strong> for millions of people employed in mundane, repetitive tasks. We believe the coming wave of automation driven by companies working on robotics will allow a larger number of humans than ever before to engage in more <a href="https://medium.com/@kpcb_edge/our-focus-areas-the-future-of-work-c4f4e02819b4#.4issl1ft2">creative, skilled work</a> by making us more productive, and by allowing small and medium businesses to grow faster. This is the future we’re excited about investing in, and hope to lay out in this post.</p><p><strong><em>Quick History</em></strong></p><p>Four decades ago, an important labor trend began emerging in the American economy. Up until 1973, gains in productivity grew in lockstep with wages. This implied that workers were sharing in the gains from productivity, likely because they were the ones driving productivity increases, through increased education, technical abilities, general efficiency, etc. Then around 1973, wages began stagnating, while productivity continued to rise at a good clip. Between 1973 and 2013, labor productivity grew 74.4%, while wages only grew by 9.2%.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/499/0*1cuW5u4j7pkvBC-N." /></figure><p>So around 1973, workers stopped sharing the gains in productivity that employers were seeing. While it’s always hard to pin-point one exact reason, a likely suspect is that since the 70s, the bulk of productivity gains in several (not all) industries has come from labor becoming more productive due to advancements in technology and tooling, not because of intrinsic human skills improving.</p><p><strong><em>What’s the story?</em></strong></p><p>But these are just correlations, you might say. What is actually happening in the field? What’s the story behind the numbers? There are too many different ones to list here, but there are two points of view we’ve heard several times from employers during the course of our investments:</p><ol><li>Human labor productivity has plateaued — In many low skill sectors, humans are basically at peak efficiency, and any future increases in productivity have to come from automation. Automation is often a result of investment by employers, not by employees, the end result being unchanged compensation for employees.</li><li>Rising employment costs — While average wages aren’t increasing, other employment costs, such as benefits and compliance costs, have increased faster than employers expected. This is especially true in some sectors, such as agriculture, where the US government has <a href="http://www.wsj.com/articles/on-u-s-farms-fewer-hands-for-the-harvest-1439371802">become stricter</a> around <a href="https://www.uscis.gov/working-united-states/temporary-workers/h-2a-temporary-agricultural-workers">areas like migrant labor.</a></li></ol><p>So to summarize, it’s costing some employers more to keep workers whose productivity isn’t necessarily increasing. What do you do?</p><p><strong><em>Enter the bots</em></strong></p><p>Imagine you are such an employer, and one day a scientist comes by and offers you a new kind of worker. This new worker not only saves you costs today by performing better, faster and more accurately than some of your other employees, but also opens up entirely new sources of revenue. This is because the new worker can learn and record data about your business at a speed and scale that you haven’t seen before. As this data set grows, you can expand into new areas, such as selling insights from that data to other people in your industry. This new worker promises to begin as a low skilled, cost saver, but over time, to grow into a high skilled revenue driver.</p><p>But in addition to saving you costs and growing your revenue, this new worker comes with a caveat. It needs to be taken care of by some of your other employees, who can stop performing routine, mundane tasks and become more focused on helping this new worker as technicians. Some of them can now focus on new areas of business that this new worker has opened up. Others leave your business to join a competitor who wants to hire employees who have worked with this revolutionary new worker. And at every step, the new worker helps your other employees achieve far more, in less time, and with less tedium.</p><p>You can see what we’re getting at here. We’re excited by teams that will make this future possible, where robots not only enable higher productivity, but improved outcomes for the people and workers who will no longer need to perform the same, robotic tasks. The crucial point here is that unlike most human employees, robots can and will change from being cost savers to revenue drivers for hundreds of different industries.</p><p><strong><em>Playing in an Industry Near You</em></strong></p><p>One of our first companies in this vein is Abundant Robotics, a team bringing automated harvesting to apple orchards, that over time can enable growers to sell high precision yield data, expand the kinds of fruit they grow with radically cheaper overhead, and employ/train more educated farm workers. Einsite, another company we’ve backed, is bringing similar monitoring and automation to the construction industry by turning traditional field equipment into intelligent machines that can track and update material usage, pilferage, and movement automatically, allowing project managers to focus on analysing insights instead of managing tedious ERP processes.</p><p>Examples of other similar use cases are becoming increasingly common: in the world of power plants, robots are inspecting field equipment instead of humans; in port terminals, they’re sorting containers and loading them onto the right trucks instead of human operated chassis loading; in retail, they’re monitoring stores as security guards, while generating important retail insights over time. In each of these cases, the visual and time series data they collect in the long term turn them from basic human replacements to important revenue drivers.</p><p>The sketch above might seem like an intellectual thought experiment. But we’re lucky enough to meet with founders everyday who are working on making this vision of the future possible — where robots can help us liberate thousands of people from routine, mundane tasks to focus on more skilled work. Our hope is that by backing the best founders working to build robots that complement and elevate businesses via the data they collect, rather than simply saving costs, we are playing our part in enabling a more productive human future. If you’re a founder working on this too, <a href="https://www.kpcbedge.com/team">let’s chat</a>!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4ed4afe761a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[A look back on 1 year of Office Hours]]></title>
            <link>https://medium.com/@kpcb_edge/a-look-back-on-1-year-of-office-hours-12f6fec0068b?source=rss-6878ba49cb9c------2</link>
            <guid isPermaLink="false">https://medium.com/p/12f6fec0068b</guid>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[enterprise-software]]></category>
            <category><![CDATA[venture-capital]]></category>
            <dc:creator><![CDATA[KPCB Edge]]></dc:creator>
            <pubDate>Thu, 02 Jun 2016 17:51:13 GMT</pubDate>
            <atom:updated>2016-06-02T17:51:13.032Z</atom:updated>
            <content:encoded><![CDATA[<p>Almost a year ago, we launched <a href="https://medium.com/knowledge-economy/introducing-edge-office-hours-bd36711fd31f#.kkue5ls9a">Office Hours</a>, an app for anyone in the world to chat with us 1:1. Since then, we’ve released an <a href="https://play.google.com/store/apps/details?id=com.kpcbedge.officehours&amp;hl=en">Android app</a> to go with our <a href="https://itunes.apple.com/us/app/kpcb-edge-office-hours/id1018492359?mt=8">iOS app</a> and written about <a href="https://medium.com/@kpcb_edge/building-two-mobile-apps-at-once-with-javascript-bf0b70a56403">how we did so</a> by switching over to React Native. It’s safe to say that we’ve achieved our goal of reaching founders all over the world — to date, we’ve had 2,500 people download Office Hours and we’ve even met with <strong>11 companies</strong> that reached out on Office Hours.</p><p>We’ve been fortunate to host four special guests for their own Office Hours sessions in the past few months — <strong>Mike Abbott</strong>, <strong>Ted Schlein</strong>, and <strong>Bing Gordon</strong> from the KPCB investing team, <strong>Jessi Hempel</strong> of WIRED, and <strong>John Maeda</strong>, KPCB’s design partner and creator of the #designintech report. Today, we wanted to share (with permission) some excerpts from Office Hours hosted by these guests over the past year that we think may be particularly insightful for early-stage founders.</p><p>— — — — —</p><p><strong>Rachel T</strong>: What metrics do you use to evaluate the “success” of a enterprise startup? Does a specific product with growth in ACV with a small set of customers triumph a product supporting a wide breadth of potential paying customers? For the first product, does being “specialized” limit growth? What is a reasonable ratio from your experience of a startup spending on sales team growth to spend on R&amp;D to scale a product?</p><p><strong>Ted Schlein</strong>: i think trying to optimize price initially is not a good idea. would rather have more customers and figure out how to get money dollars out of them over time. you will learn more from more customers. with that said, I do think <strong>right when you get started having 5 or so customers to iterate with and learn from before expanding too fast is important in order to get your product right</strong>. Not exactly sure what ‘specialized’ means but being focused on solving an acute problem so you earn the right by a customer to solve more problems for them is a good strategy. <strong>Your two most important expense buckets are sales and engineering. initially you will be engineering heavy. then once you figure out how to sell a product and you have a repeatable model, you should start scaling sales</strong>. i would expect this ratio in terms of dollars spent to be about 1:1 once you have reached this point. then eventually sales and marketing tend to be almost 2:1 as you are really scaling</p><p>— — — — —</p><p><strong>Nick Arner</strong>: Hi John, I’ve followed some of your tweets regarding seeing in health care. What are your thoughts on the role of sound and acoustics in hospitals? When I’ve visited family recently in the hospital, I was struck by just how loud the environment is, and how it’s probably harder for patients to get proper rest due to the high noise levels. How can we design a more peaceful hospital environment?</p><p><strong>John Maeda</strong>: Hi Nick, There’s a great video by IDEO on this very topic. Where it is here. Ah. Here it is. <a href="https://www.youtube.com/watch?v=cDDWvj_q-o8">https://www.youtube.com/watch?v=cDDWvj_q-o8</a>. It’s an old video, but an example of how designers can find these problems, but historically they’ve never been able to fix them. That’s why I am bullish about the startup universe — there’s at least a fighting chance to change things.</p><p>— — — — —</p><p><strong>Daniel Farrell</strong>: at what point in the life of this product do you recommend we bring-on an advertising / marketing specialist to fuel the content side?</p><p><strong>John Maeda</strong>: I think the lesson today is that once you hire a marketing person, you can suffer because your product may not be good enough to win, and marketing is what is tipping it over the line. <strong>If your product is doing well, and you’ve already tapped out in terms of people who already use it and already love it, that’s a good time to think about how to grow your market.</strong></p><p>— — — — —</p><p><strong>Ivana Oda</strong>: Ted, what are the most important metrics for measuring the performance of an early stage cloud infrastructure company?</p><p><strong>Ted Schlein</strong>: Ivana, assuming you mean business metrics, it will most likely be around customers and customer usage. In other words, how many customers do you have, how fast are they adopting, how often are they using the software or services.</p><p>— — — — —</p><p>If you’d like to chat with us about the topics in these discussions or see which sessions are coming up, check out the <a href="https://www.kpcbedge.com/products/office_hours">Office Hours app</a>!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=12f6fec0068b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Our focus areas: The future of work]]></title>
            <link>https://medium.com/@kpcb_edge/our-focus-areas-the-future-of-work-c4f4e02819b4?source=rss-6878ba49cb9c------2</link>
            <guid isPermaLink="false">https://medium.com/p/c4f4e02819b4</guid>
            <category><![CDATA[future-of-work]]></category>
            <category><![CDATA[freelancing]]></category>
            <category><![CDATA[venture-capital]]></category>
            <dc:creator><![CDATA[KPCB Edge]]></dc:creator>
            <pubDate>Tue, 24 May 2016 15:31:09 GMT</pubDate>
            <atom:updated>2016-05-24T15:31:09.225Z</atom:updated>
            <content:encoded><![CDATA[<p><strong>Our focus areas: The future of work</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/1*vYkGpRKmHkqOKHj2fyXQyQ.png" /></figure><p>The nature of the jobs we hold as a population is constantly changing as a result of technology and sociopolitical movements. Centuries ago, the majority of the workforce — 90% by some estimates [<a href="http://www.pbs.org/wgbh/amex/trouble/timeline/">1</a>] — was employed in agriculture. With new machinery and techniques improving planting, seeding, fertilization, and harvesting, today only 2% of the US workforce are farmers. A similar shift has happened in manufacturing, where the number of jobs has declined by 37% since its peak in 1979 [<a href="http://cnsnews.com/news/article/terence-p-jeffrey/7231000-lost-jobs-manufacturing-employment-down-37-1979-peak">2</a>], the result of globalization and more efficient production lines. During this time, much of the job loss in farming and manufacturing was offset by gains in the service industry (e.g. education, healthcare, banking, and legal services).</p><p><strong>Today, the most interesting labor shift we’re observing is the rise of the contingent workforce</strong>, composed of freelancers, independent contractors, part-time workers, and self-employed individuals. This group has grown from 30.6% of the workforce in 2005 to 40.4% in 2015 [<a href="http://www.forbes.com/sites/elainepofeldt/2015/05/25/shocker-40-of-workers-now-have-contingent-jobs-says-u-s-government/#4c2177492532">3</a>]. In fact, this implies that <em>all</em> of the net employment growth in the US between 2005 and 2015 can be explained by the growth of the contingent workforce, as total employment grew from 140.4 million in February 2005 to 149.4 in November 2015 [<a href="http://krueger.princeton.edu/sites/default/files/akrueger/files/katz_krueger_cws_-_march_29_20165.pdf">4</a>].</p><p>It’s important to note that <strong>the contingent workforce is comprised of far more than those working in the on-demand economy</strong> (Uber drivers, TaskRabbit workers, DoorDash deliverers, etc.). For instance, there’s the masseuse or fitness trainer managing her own client base, the retail worker stringing together part-time jobs, and the electrician or plumber traveling to multiple job sites a day.</p><p>What does it mean to be part of this workforce? How is your life different if you’re an independent contractor or a freelancer versus a full-time employee? Here are just a few areas to consider:</p><ul><li><strong>Finding &amp; scheduling work</strong> — instead of a regular schedule, contingent workers piece together a schedule and/or client base on their own.</li><li><strong>Income volatility</strong> — instead of a regular paycheck, 41% of individuals now experience more than a 30% change in income on a month-to-month basis [<a href="https://www.jpmorganchase.com/corporate/institute/document/jpmc-institute-volatility-2-report.pdf">5</a>].</li><li><strong>Taxes</strong> — instead of employers doing tax withholding on each paycheck, independent contractors are responsible for tracking income and expenses, and often making quarterly estimated tax payments.</li></ul><p>We believe that <strong>there are great opportunities for software and software-enabled services to play a role in serving this growing group of contingent workers</strong>, particularly as people increasingly grow accustomed to managing their work on their mobile devices. In this post, we discuss each of these areas and the problems we’re interested in seeing companies tackle.</p><h3><strong>Finding &amp; scheduling work</strong></h3><p>Without customers, shifts, or job orders, there is no work. And this is the defining characteristic of contingent workers — unlike somebody working one full-time job, they are piecing together a schedule on their own. In the 2015 1099 Economy Workforce Report [<a href="https://gumroad.com/l/rfsreport">6</a>], <strong>‘finding enough work’ was the most commonly reported pain point by 49.2% of respondents</strong>, and ‘optimizing my schedule to maximize earnings’ was the third most commonly reported by 31.5% of respondents. (‘Understanding tax or legal obligations’ was the second most common, which we talk about later.)</p><p>Because there are many different types of work ‘units’ for different types of workers — e.g. client appointments, work orders, or shifts, depending on what exactly you do — we see several different types of opportunity here. For example, for people who build their own client bases, we’re excited about marketplaces that connect and schedule them with new customers. For people who need to choose shifts, data that can help them optimize earnings across multiple platforms is extremely helpful. Despite existing tools in this space (39% of contract workers report using schedule optimization tools), many still see it as a pain point.</p><h3><strong>Income volatility</strong></h3><p>One of the defining characteristics of the contingent workforce is income volatility. Earlier this year, JP Morgan published a study that showed <strong>41% of individuals experience more than a 30% change in income on a month-to-month basis</strong> [<a href="https://www.jpmorganchase.com/corporate/institute/document/jpmc-institute-volatility-2-report.pdf">5</a>]. This means if you’re making the US median wage of $2,336 a month [<a href="http://www.mybudget360.com/how-much-do-americans-earn-in-2015-household-income-wages-real-income-gdp/">7</a>], your income could easily range from $1,635 to $3,037 a month, a $1,402 difference. And this can mean the difference between making a home/car/insurance/etc. payment on time, versus missing the payment and accruing further debt in fees and interest.</p><p>The study also found that particular opportunities to save arise during five-Friday months and between December and March. The extra paycheck for part-time workers paid weekly or biweekly usually means positive cash flow, although the timing of these five-Friday months is usually unanticipated. And at some point between December and March, 61% of individuals experience a 5%+ increase in earnings, which is partially tied to increased hours during the holiday season.</p><p>Interestingly, there’s also evidence that “gig economy” and “sharing economy” platforms like Uber, TaskRabbit, and Airbnb are beginning to be leaned upon. Only 1% of the 1 million people in this study were making income from such platforms in any given month, but data shows that earnings from “labor platforms” (defined as platforms where workers perform discrete tasks) are used to offset dips in income, while earnings from “capital platforms” (defined as platforms where participants sell good or rent assets) are used to supplement income.</p><p>Overall, we feel this means there’s need for software and software-enabled services to play a role in tracking and predicting income, suggesting what’s safe to spend and what needs to be saved, and perhaps even recommending time to spend working on a labor platform in a given month.</p><h3><strong>Taxes</strong></h3><p>Independent contractors are responsible for their own taxes. For a full-time salaried employee, tax withholdings are taken out of every paycheck, typically leading to a more balanced outcome at the end of tax filing season. This doesn’t happen for independent contractors, which means they need to estimate tax payments to put aside to avoid surprises. <strong>On top of this, independent contractors who expect to pay over $1,000 in taxes need to make quarterly estimated tax payments to avoid an additional penalty to the IRS.</strong></p><p>During tax filing, independent contractors file separate Schedule Cs tracking income and expenses for each type of “business or professional activity” they have participated in, in addition to the standard 1040. These business expenses need to be tracked continuously, as they can include home offices, miles driven, equipment depreciation and repairs, insurance premiums, and education expenses.</p><p>Here, we see opportunity for software to help independent contractors keep track of their expenses, find the right deductions, and predict and pay their quarterly tax payments, likely with integration into bank and debit/credit records and machine intelligence sitting on top of this data stream.</p><h3><strong>Concluding thoughts</strong></h3><p>Overall, <strong>this post covers only some of the areas that we feel are important when thinking about the tools and services that will become essential as the future of work evolves</strong> (there are so many more — e.g. education, training, insurance, retirement benefits, etc.). In each section, we mentioned a few use cases where software could help — some of these might be great opportunities standalone, while others would make more sense as a feature of a larger offering. If you’re a founder interested in shaping the future of work, we’d <a href="https://www.kpcbedge.com/team">love to chat</a>.</p><p><strong>By the way</strong>: at the beginning of this year, our team decided to add two new focus areas to our <a href="https://www.kpcbedge.com/about">existing six</a> representing where we see the world going. They are the future of work and robotics. Post on robotics coming soon!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c4f4e02819b4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Making small data sets more useful]]></title>
            <link>https://medium.com/@kpcb_edge/making-small-data-sets-more-useful-c68a3e660ac0?source=rss-6878ba49cb9c------2</link>
            <guid isPermaLink="false">https://medium.com/p/c68a3e660ac0</guid>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[machine-learning]]></category>
            <dc:creator><![CDATA[KPCB Edge]]></dc:creator>
            <pubDate>Fri, 06 May 2016 16:08:06 GMT</pubDate>
            <atom:updated>2016-05-06T16:08:06.779Z</atom:updated>
            <content:encoded><![CDATA[<p>Given our <a href="https://medium.com/@kpcb_edge/our-focus-areas-computer-vision-cf8d733f5046#.t6l5v4x8t">focus on computer vision</a>, we’ve spent a lot of time looking at the deep learning ecosystem as a whole too. Within AI, deep learning has received a lot of attention in recent months, and for good reason — its success across a large variety of problems is impossible to ignore. Most of the promises of deep learning hinge on the use of convolutional neural networks and unsupervised or semi-supervised feature learning, which require more data than typical machine learning algorithms with hand-tuned feature extraction.</p><p>Deep learning is very well-suited to a specific class of problems. However, we’ve seen a lot of folks use deep learning in contexts where it may not be the best fit. We believe there’s also a large opportunity to create value for end users by applying new, less-supervised techniques to <em>smaller</em> data sets. These could be data sets generated by single individuals, like health sensor readings, email correspondence, or location data. Time series data is often a good example of such a smaller data set, because the rate of new data collection is constrained by time.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*meOLKuBOvamhqdleQDMVzg.png" /></figure><p>With new academic work being done, there may be better ways to apply machine learning in highly data-constrained situations. There are a few directions that researchers are pursuing that look promising. One angle has been to use probabilistic methods to increase neural network accuracy on small data sets, such as <a href="http://ieeexplore.ieee.org/xpl/login.jsp?reload=true&amp;tp=&amp;arnumber=4021401&amp;url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D4021401">this paper</a> defining a technique using posterior probabilities. Another technique, called Bayesian program learning (BPL), was recently used to recognize and generate handwritten characters based on a single example. The BPL system researchers built was able to pass a turing test of sorts, where the handwriting samples it generated from one example could not be separated from human-generated samples. Full paper can be found <a href="http://web.mit.edu/cocosci/Papers/Science-2015-Lake-1332-8.pdf">here</a> (<a href="https://www.technologyreview.com/s/544376/this-ai-algorithm-learns-simple-tasks-as-fast-as-we-do/">secondary</a>).</p><p>Research in this area is still very early, and the greater use cases for BPL are still unproven, but there are early indications that it could have more general applications. From the work done so far, it’s clear there are vision-related use cases where BPL could learn from far less data than deep learning. BPL could also be used to generate more examples (like the above) to feed into a CNN training process.</p><p>Value provided on top of a small data set also creates an opportunity to aggregate a larger data set across all users of a company’s product, where the initial incentive for users to provide data might not be present otherwise. A good example of this is how Mint or Personal Capital are useful for just a single user, providing an aggregate view of your personal finances, but have built up a much larger multi-user financial data set that allows them to provide more powerful recommendations. Small data AI could be a new way to provide that initial user value, paving the way to aggregate features like benchmarking or more traditional deep learning on the larger future data set.</p><p>More data will almost always lead to better results, but we believe there are many potential use cases for intelligent systems where data is scarce. If you or anyone you know uses machine learning on smaller data sets, we would love to <a href="https://www.kpcbedge.com/team">chat</a>!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c68a3e660ac0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Shenzhen: The World’s Secret VR Supplier]]></title>
            <link>https://medium.com/@kpcb_edge/shenzhen-the-worlds-secret-vr-supplier-5b7de22e3b0e?source=rss-6878ba49cb9c------2</link>
            <guid isPermaLink="false">https://medium.com/p/5b7de22e3b0e</guid>
            <category><![CDATA[shenzhen]]></category>
            <category><![CDATA[virtual-reality]]></category>
            <category><![CDATA[mobile]]></category>
            <dc:creator><![CDATA[KPCB Edge]]></dc:creator>
            <pubDate>Wed, 27 Apr 2016 17:07:29 GMT</pubDate>
            <atom:updated>2016-05-02T22:27:26.783Z</atom:updated>
            <content:encoded><![CDATA[<p><em>This is the first in a series of posts about Edge’s recent visit to Shenzhen.</em></p><p><strong>Looking West</strong></p><p>Over the last three years, the rise of VR in public discourse has been impossible to ignore. During this time, most mainstream discussions have centered on the same few efforts in the US — Oculus, Vive, and Cardboard. But there’s something as important going on roughly 7000 miles west of San Francisco, across the Pacific Ocean in Shenzhen, China. There, in the markets of Huaqiangbei, the factories in Dongguan, and in the high rise offices of Nanshan, is where demand and supply for VR hardware and software are on par, if not outpacing, the US. The Edge team took a few days to go on the ground in Shenzhen in March, and this is a summary of what we saw.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/845/0*bj69_w2OwOrnqfwn." /></figure><p>If you, like us, believe that VR has a bright future, then you likely believe in the ability of most people on Earth to access mobile VR. This is because smartphones are the only devices that have anywhere near the scale, computational ability and price points to bring VR to most of the planet in the near future. This is the story playing out in Shenzhen. Unlike the Rift and Vive, which cost around $1600 all-in for headset and PC, there are a hundred different flowers of mobile VR blooming (pun intended) in the Chinese electronics market, ranging from low tech VR smartphone cases to standalone Android headsets. Not all of these are strictly VR devices (we’ll get into that in a moment), but when it comes to stereoscopic, head mounted displays (HMDs) for consumers, Shenzhen is a leader.</p><p>Broadly, most devices we saw fell into three categories — Mobile VR headsets, PC HMDs, and standalone headsets:</p><p><strong>Mobile VR (a.k.a. Budget VR)</strong></p><p>These mobile headsets involve turning your phone into a VR viewer by adding a couple of plastic biconvex lenses to a frame that holds your mobile device, essentially a nicer version of Google cardboard. The sheer range of these is reminiscent of the explosion in custom smartphone cases since 2010. The bulk of stores we saw dedicated to mobile VR headsets were located in the SEG Communications Market, a multi storey edifice dedicated just to mobile parts and accessories, sold by the sorts of retailers who would otherwise be selling cases, chargers, cables and lens protectors for phones.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*WQUeOVWUvqjpBq8c." /></figure><p>The cheapest of these was the $10 VRCase iPhone case viewer below (prices can drop quickly below this with the number of units you buy) inspired by the Figment Kickstarter, with the most expensive one at $45 for a drop-in style viewer.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*iqTu-AZrzwPC_fLX." /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/928/1*w_oNKP5i_Voh0l7_2vNtlg.png" /><figcaption>The $10 iPhone VR Case — seen here with VR Flappy Bird.</figcaption></figure><p>There are entire retail storefronts, like the one pictured below, dedicated just to these mobile VR headsets. While speaking to the folks manning these stores, we were told it’s common to move up to 1000 headsets a day in retail, and 10k units a day to distributors, with orders mostly from locals or buyers visiting from Korea and Japan.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*M6z5P1tFPm1uDOKw." /><figcaption>One of several dedicated VR retail stores</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*rWgnZFS42w2HDzuR." /><figcaption>More VR retail storefront — this is the one that sells the iPhone case.</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2LLdDr4F6x4MrAi529X6yw.png" /><figcaption>Cases selling by the dozens.</figcaption></figure><p>When we asked the VR salesmen which apps were most popular with consumers in China, we saw a combination of apps from overseas like Homido and Cmoar’s cinema apps, and homegrown first party apps, like Baofeng Mojing and VRCase’s iOS VR players. Watching 2D movies and videos in 3D spaces is by far the most popular use case for these headsets, with gaming and virtual tours as runner ups.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*5LBhuSpRaJFAJ7M4." /></figure><p><strong>PC-only Headsets (a.k.a High End VR)</strong></p><p>Adjacent to the mobile accessories building is the main SEG tower, a 72 story building with 10 floors dedicated to a dizzying range and number of electronics retailers. Here is where we spotted a few Oculus and Vive alternatives, like the AntVR pictured below which offers similar specs to the DK2 for around $280 and comparable build quality, and the Deepoon E2, another DK2 alternative. The lack of enthusiasm for these compared to the mobile headsets from the retailers was palpable, and it sometimes felt like these headsets were displayed more for novelty value than anything else. This is understandable, given the low volume of these headsets that are moving in the market and their relatively high prices for the average Chinese consumer market.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*P6CW7IpNs3Sjq6ZD." /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/684/1*9DaijsYr29Wmr51MTtL6tA.png" /></figure><p><strong>All-In Headsets</strong></p><p>Walking through Huaqiangbei, it became apparent to us that there’s a secondary consequence of the rise of VR taking place — namely, a halo effect on non-VR HMDs. These are essentially stereoscopic media players, which give you the ability to watch your regular 2D content on a display that, when suspended on your face, offers the effect of a much larger screen. While these have been around for years in various form factors and high price points like the $799 Sony HMZ (which launched in 2011) and the $500 Vuzix iWear, we saw a plethora of newer, cheaper alternatives. We were told these have become popular as HMDs like the GearVR and Oculus are making it more palatable for consumers to strap displays to their faces. Most of these were running their own flavors of Android on the headset, and had USB ports for users to plug in their own content. These ranged from $120 video players, to the $617 RoyoleX headset that offers a direct alternative to devices like the $700 Avegant Glyph. While none of these are VR headsets in the strict sense that they didn’t have gyroscopes and accelerometers to allow head tracking, it seems to be only a matter of time before that changes with the coming launches of devices like the <a href="http://www.tmtpost.com/1482900.html">Baofeng Mowang,</a> <a href="http://www.aliexpress.com/item/2015-newest-Letv-super-helmet-3d-VR-glasses-portable-giant-screen-3D-theatre-only-support-Letv/32596835765.html?spm=2114.01010208.3.1.R1DofC&amp;ws_ab_test=searchweb201556_10,searchweb201602_3_10017_10005_10006_10034_10021_507_10022_10009_10020_10008_10018_10019,searchweb201603_2&amp;btsid=64598f6b-8639-4030-8b04-73b3de820149">LeTV’s Helmet</a>, and <a href="http://www.picovr.com/en/neo.html">Pico Neo</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*-16BD_COqXCxMok3." /></figure><p>One of the headsets we bought, the Bijela iTheater (picture), essentially functions as a fully functioning tablet, allowing you to run all the apps you’d typically want to use with a desktop monitor (Gmail, Chrome, Kindle Reader etc.) but with a mobile device that fits in your backpack, costs you roughly $190 and weighs about 40% less than a GearVR with a phone. It’s easy to see why headsets like these that are somewhere between VR headsets and smartphones both in functionality and price points could become popular with young consumers who have never owned a dedicated PC or laptop.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/568/1*ZsuydVuBRSTvs-4phVQctQ.png" /><figcaption>The Bijela iTheater — one of many non VR HMDs seen in Huaqiangbei</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*AsaRaDJWiqvfQgZ5." /><figcaption>Another non VR HMD that runs Android on board.</figcaption></figure><p><strong>Takeaways</strong></p><p><a href="https://medium.com/@kpcb_edge/our-focus-areas-virtual-reality-debbcc2ca40f#.ocvzrof04">Last year, we wrote about a coming age of mobile VR</a>. Shenzhen is where that future is approaching quickly. It’s an exciting time in history, where for the first time we’re seeing the convergence of two unique trends — massive numbers of smartphones in the hands of billions of people, combined with the sheer range of low cost VR headsets described above. Over the next few years, founders working on compelling software experiences in VR will be able to build for an audience of hundreds of millions of users who can access VR for the price of a low cost smartphone accessory. This is the reason we’re excited to work with entrepreneurs working on building the apps, services and metaverses that will bring these headsets to life for consumers across the world.</p><p>If you’re one of these founders, let’s talk!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5b7de22e3b0e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Our focus areas: Mobile marketplaces]]></title>
            <link>https://medium.com/@kpcb_edge/our-focus-areas-mobile-marketplaces-252e65a6f2cf?source=rss-6878ba49cb9c------2</link>
            <guid isPermaLink="false">https://medium.com/p/252e65a6f2cf</guid>
            <category><![CDATA[mobile]]></category>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[marketplaces]]></category>
            <dc:creator><![CDATA[KPCB Edge]]></dc:creator>
            <pubDate>Tue, 26 Apr 2016 17:18:03 GMT</pubDate>
            <atom:updated>2016-04-26T17:18:03.695Z</atom:updated>
            <content:encoded><![CDATA[<p>(Post 5 of 6 introducing our focus areas at <a href="https://www.kpcbedge.com/">KPCB Edge</a>)</p><p>Marketplaces have evolved through several phases as of late. In the pre-Internet days, buyers and sellers met in person and exchanged goods at physical locations. As people moved online, buyers and sellers shifted their transactions to their desktop terminals, using marketplaces like Amazon and eBay. Now, as the world has moved onto mobile devices, buyers and sellers can exchange goods and services in new ways, enabled by the capabilities and interfaces of our phones.</p><p>At Edge, we’re interested in these new marketplaces that are using technology to connect users and providers. For example, <a href="http://usetrusted.com/">Trusted</a>, one of our portfolio companies, is a marketplace that connects parents with trustworthy child care providers. And <a href="https://www.homie.com/">Homie</a>, also a portfolio company, is a marketplace that connects home sellers with potential buyers.</p><p>There are three trends we’ve observed as a result of the world’s shift to mobile that we believe will pave the way for many more successful marketplaces to grow:</p><ol><li>On mobile, marketplaces are able to collect new types of data in greater volumes to improve their recommendations and operating efficiency.</li><li>Mobile developers have access to better APIs and SDKs for building add-on features to drive retention, either by helping individuals and businesses on the marketplace run their operations, or by otherwise augmenting their experience.</li><li>Users increasingly expect seamless experiences on mobile when searching for and interacting with providers, regardless of whether the task at hand is done on a daily, yearly, or once-in-a-lifetime basis (e.g. ordering food, finding a plumber, or planning a wedding).</li></ol><p>In this post, we’ll walk through these trends in more detail and discuss what this means in terms of the companies we’re excited about.</p><h4><strong>Trusting the data</strong></h4><p>As a marketplace company operates, it collects data that continuously improves its recommendations and its operating efficiency. And on mobile, this data can be collected more ubiquitously, relying on sensors built into the device and on notifications that prompt the user for input. One example here is Uber’s rating system which pops up at the end of every ride for passengers to rate drivers and vice versa. This data is subsequently used to better understand the profile of each participant in the marketplace, which feeds into the matching system.</p><p>Another example is DoorDash’s <a href="http://blog.doordash.com/post/141605482666/delivering-our-biggest-tastiest-update-yet">Delight score</a> which ranks restaurants by bundling several parameters it has collected over time, like restaurant popularity, customer reviews, and speed of delivery. By tracking the driver’s location via the mobile app the driver uses, DoorDash can easily calculate delivery time. And by giving users the option to sort by Delight score, DoorDash in turn increases the likelihood that the food it delivers will be on-time and tasty.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/842/1*6oUEPH5W6faLMIeseC-MOQ.png" /></figure><p>We’re excited about companies that can leverage the data generated by their marketplace operations to create even better product experiences.</p><h4><strong>Add-on features</strong></h4><p>Secondly, we believe that marketplaces can increasingly provide value to their users beyond facilitating the match. For businesses, this might mean building SaaS-like tools to get analytics on their transactions flowing through the marketplace, to view insights on their customer base, or to send invoices to their customers. For consumers, this might mean the ability to manage communications with the supplier through the marketplace app or to track spend for tax purposes. These features can not only help keep users on the platform, but be part of the value proposition for getting users to sign up in the first place.</p><p>For mobile developers, these features have become easier to build due to the increasing availability and maturity of APIs and SDKs for services like messaging (e.g. Layer, Twilio), payments (e.g. Stripe), and mapping (e.g. Google Maps SDK, Mapbox).</p><p>Homie is a great example here — for home sellers who join Homie, beyond just helping them find a buyer, Homie will have their homes professionally photographed, setup a custom phone number and screen calls for qualified buyers, help with home pricing with a certified appraiser, and more. For a seller, paying a monthly fee to join the Homie platform instead of paying 3% of the home price to an agent is a clear win.</p><h4><strong>Seamless mobile experiences</strong></h4><p>With US adults now spending over half of their digital time on smartphones and mobile spending now 30% of all e-commerce, it’s clear to us the intent to make purchasing decisions and get things done on mobile is prevalent. Companies like Amazon and Uber have made consumers comfortable and familiar with purchasing on their phones. Successful marketplaces can capitalize on this trend by designing simple and seamless mobile flows to go from zero to bookings.</p><p>Trusted is an excellent case study here, shown below. The user is guided through entering information about the children who need care, address of service, and booking time, and then presented with the profiles of available qualified providers and the ability to book instantly.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QqnmhyfWTY0yb2wYkLe5tQ.png" /></figure><h4><strong>What this means for Edge</strong></h4><p>It’s clear to us that there that tremendous opportunities now exist for companies who are able to build seamless mobile experiences for users to connect with some specific product or service, leveraging both the data collected by running the marketplace and the add-ons that keep users coming back. We haven’t touched as heavily on supply- and demand-side acquisition tactics in this post since they often tend to vary more by vertical, but we’re always interested in hearing about what’s working for you. If you’re a founder building a marketplace and some of these pieces resonate with you, we’d <a href="https://www.kpcbedge.com/team">love to chat</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=252e65a6f2cf" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Building Two Mobile Apps at Once… With Javascript]]></title>
            <link>https://medium.com/@kpcb_edge/building-two-mobile-apps-at-once-with-javascript-bf0b70a56403?source=rss-6878ba49cb9c------2</link>
            <guid isPermaLink="false">https://medium.com/p/bf0b70a56403</guid>
            <category><![CDATA[android]]></category>
            <category><![CDATA[ios]]></category>
            <category><![CDATA[react]]></category>
            <dc:creator><![CDATA[KPCB Edge]]></dc:creator>
            <pubDate>Wed, 10 Feb 2016 05:12:01 GMT</pubDate>
            <atom:updated>2016-02-10T05:22:20.922Z</atom:updated>
            <content:encoded><![CDATA[<p>The only programming language that every full-stack developer must know is Javascript. A strong statement, I know. Let me justify it. Two and a half months ago I had never touched iOS or Android environments, and two weeks ago <a href="https://medium.com/p/21c74f3637bd">Edge released both Android and iOS versions of an app I wrote</a>. How did I do it? Javascript and React Native.</p><h4>React Native</h4><p>React Native is a view library, built on top of Facebook’s React project, which uses native iOS and Android components to render components written in Javascript. It leverages React’s powerful properties, like compose-ability, clear component responsibilities, and reactive paradigm. But more importantly, it abstracts the implementation of built-in components.</p><p>Practically, this means I can write “&lt;ScrollView&gt;…&lt;/ScrollView&gt;” and be assured it will (to a surprising degree) work on both Android and iOS. To me, this means I can write an Android app and an iOS app at the same time, with no experience in either platform.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/616/1*824ITeMmWbwomMBiFudYDw.gif" /><figcaption>The ScrollView component worked out-of-the-box with both iOS and Android!</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/768/1*rOeswk7MBoPmae-Ft4fYhg.gif" /><figcaption>Notice the UI cues for overscrolling are native</figcaption></figure><h4>The Process</h4><p>When I started at KPCB Edge, my experience with Javascript was building Backbone apps in the typical jQuery-heavy imperative style, usually by taking mock HTML created by our designer and adding interactions to it. But when I started on the Office Hours project, our designer-in-residence, Brian Tran, had already designed beautiful image mocks. So the first step was to take those images and render them with functional code. This first step, strangely, was the point of greatest friction for me.</p><p>In addition to learning how the iOS simulator worked, I had to think about UI in a radically new way. In React, the component hierarchy is not just a proxy for the layout, it’s also the logical structure of the project. Unlike an HTML document with embedded code (like an ERB template in Ruby), data must flow in a more structured way so React knows when and how to re-render. For example, the first component I implemented was the partner image in the home screen.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/497/1*FJd8PFWvDgpfockknljwYw.png" /><figcaption>A first draft of the summary screen</figcaption></figure><p>Here’s the code to render the partner images:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/b7e78dd175b4e22037d579d1d2d826eb/href">https://medium.com/media/b7e78dd175b4e22037d579d1d2d826eb/href</a></iframe><p>And the code to render the whole summary screen:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/d252a3434cb74c0437813a64dee2769e/href">https://medium.com/media/d252a3434cb74c0437813a64dee2769e/href</a></iframe><p>Eventually each partner image needed to have some data property to indicate which image it should display, and it wasn’t immediately clear which component should be in charge of fetching that data from the server and then passing it through to each of the three image components. This blending of layout with logic also means it can be difficult to get a designer onboarded to contribute to the project. Brian already knew some Javascript, so the process was straightforward, but the key insight was the same for both of us: an interface built in React is architected and designed simultaneously.</p><p>Once the iOS version was working and submitted to the App Store for review, I began work on the Android version. Most of the components I had written still worked! There were a couple cases of components which had more general cousins that I had to replace (for instance, the &lt;NavigatorIOS&gt; component had to be switched out with the &lt;Navigator&gt; component). The relative immaturity of the Android implementation meant inconsistencies in styling and component APIs between platforms. But there were workarounds in both cases. For instance, the swiping carousel component I had used on iOS didn’t work on Android, so I replaced it with simple next and back buttons.</p><p>All said, <strong><em>porting to Android took 10 business days</em></strong> from first draft to app submission, and the code reuse between the two platforms was just about 90%.</p><h4>Conclusion</h4><p>The landscape for mobile development is fundamentally different with React Native. For most applications, there will no longer be an incentive to learn one of Android or iOS when learning Javascript gives you the ability to code for the server (node.js), the browser (React), mobile (React Native), and even the desktop (React Desktop). For the first time, it’s practical to be a truly full-stack developer by knowing just one language and a few frameworks. Even Javascript itself is quickly evolving into a <a href="https://nodejs.org/en/docs/es6/">modern, functional language</a>. As React Native matures, we’ll begin to see more perfect feature-parity with native mobile development. The barriers to writing an app will dissolve for anyone who knows javascript.</p><h4>Thanks for the help!</h4><p>A shout-out to two organizations which made the app possible: <a href="http://www.smooch.io/">Smooch</a> and the React Native community.</p><p>Smooch is the chat solution we used for Office Hours V1, and when we started the React Native version of Office Hours, Smooch had no React Native library, but they were responsive to our needs and built a library with both Android and iOS bindings.</p><p>React Native is a young project — it was open-sourced 9 months ago — so there are rough edges, especially on Android. But the community, especially on Github, was indispensable to us. I can’t tell you how many times I discovered a bug only to realize it had been discovered by someone else 2 days ago and there was already a pull request to fix it. The community builds fast, communicates clearly and politely, and releases often.</p><p><em>This post was written by </em><a href="https://www.kpcbedge.com/team"><em>Andrew Taber</em></a><em>, our Developer in Residence. Learn more about our residencies </em><a href="https://www.kpcbedge.com/roles"><em>here</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bf0b70a56403" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>