AI for all I hate the term “artificial intelligence.” I think it betrays a willfully reductionistic understanding of intelligence and contributes to the anti-technological hysteria of people who should really know better. (I’m looking at you, Stephen Hawking!) While the term “machine learning” is also flawed in its way, it is a superior description for the set of technology and methods that are typically referred to as AI. But no matter. Google's DeepMind has staked out a position on the cutting edge of (sigh) AI technology. But because of the aforementioned hysteria, they have come under criticism for not sharing the details of their inventions. Never mind that creating competitive advantage through technology is what successful tech companies do. Bowing to pressure from the tinfoil hat crowd, DeepMind has made their training platform publicly available through Github. Why does this matter? In any immediate sense, it doesn’t matter. The number of people capable of setting up and using this training platform is vanishingly small. Releasing source code that is intelligible to only a small number of advanced machine learning researchers around the world is a good way to mollify the singularity true-believers, without giving away anything actionable. However, in the long term, this reflects a positive trend among leading technology companies. Since many of these companies were founded by academics, they tend to take an academic approach to knowledge-sharing than traditional corporations. Google, Facebook, and even Microsoft have begun to open source more and more of their proprietary technologies. This demonstrates a high level of self-confidence plus an ongoing commitment to innovation. The thinking seems to be: we can share this cutting edge work because we will continue to create cutting edge technologies in the years to come. Next Steps: Technology has always been an uncomfortable fit for the legal concept of “intellectual property.” Innovation is a process, not like “eleven herbs and spices.” Read More Just a cigar There is a popular notion in technology that one way to create and maintain a competitive advantage for your product is to create a network based on the use of that product. This has become known as “come for the tool, stay for the network.” In a recent article on TechCrunch, Marc Bodnick criticizes this approach as unrealistic. His point is that while building a technology product is quite challenging, building a network as a late stage add-on for that product is almost impossible. Networks are fiendishly difficult to create and most products don’t lend themselves to networking. Imagine I have a cigar brand – Sigmund Brand Cigars (trademark pending.) Frustrated by low levels of customer loyalty and poor repeat purchase, I introduce “cigarfancy.net” – a network for lovers of Sigmund Brand to connect with like-minded individuals. How many people do you think would actually sign up for Cigarfancy? And how many of those would be my employees or friends? While this example is absurd, it’s not that far removed from the add-on networks being peddled by some technology companies. Why does this matter? There’s a reason this idea appeals to entrepreneurs. Building market share is difficult. Marketing and customer retention is expensive. And venture capitalists like to invest in companies that benefits from network effects. But you cannot let the long tail wag the dog. (Yes, I am rather proud of that.) Networks are built to be networks, not late stage additions to other products. While it is true that some networks emerge spontaneously (as virality), attempts to reverse-engineer these ad hoc networks have been dismal failures. Next Steps: A network is a network. And a cigar is just a cigar. Don’t get the two confused. Read More The Death and Life of Smart Cities In a “State of the Union” style blog post one year into the founding of Google’s Sidewalk Labs, CEO Dan Doctoroff discusses the projects and possibilities of the smart city powerhouse. (Link below.) As a New Yorker, I am an unapologetic fan of Sidewalk Labs. Their LinkNYC initiative (despite a minor porn problem) speaks to a fundamental optimism about the possibilities of urban centers in this post-Industrial age. Doctoroff correctly identifies the problem of top-down smart city initiatives. Yet, many of the projects he talks about feel like innovation by fiat. This is a natural problem. As Doctoroff points out, cities are complex. Local governments are frequently the only entities that can cut the Gordian knot of conflicting interests. Despite Jane Jacobs ideals, we frequently end up with Robert Moses solutions. Why does this matter? Urban centers are the engines of modern economies. The cheek-by-jowl proximity of teeming humanity inspires and encourages because it challenges our assumptions. Cities are often seen as poor labs for innovation because of their complexity. One of the things I like about Sidewalk Labs’ approach is they understand that cities are monuments to the principle of unintended consequences. Urban planning is no less crucial for being an oxymoron. I would quibble with Doctoroff’s claim that cities have not materially changed since the introduction of the automobile. Computers and the internet have also transformed our landscape. But the changes have been in the private, rather than the public spaces of our cities. Next Steps: Given the increased focus on profitability across Alphabet’s holdings, I would anticipate more product introductions from Sidewalk Labs in the coming years. Read More I’m gonna hurl Virtual reality could be huge. We finally have inexpensive, lightweight headsets and a range of developers and content creators working in VR. Museums, brands, and video game developers are eager to apply VR technologies. But there’s a persistent puking problem. Virtual reality makes people nauseous because you have a visual perception of one kind of movement, but your body feels quite a different set of movements. Every time I read about a new technology that supposedly fixes the puking problem (like in the link below), the punchline of the article is “VR still makes you nauseous.” No one pays for nausea. And no one pays for experiences that only last 20 minutes after which you feel like throwing up. Until virtual reality finds a way to fix the puking problem, this is an incredible technology that is a poor fit for homo sapiens. Why does this matter? Nausea is a large, but not insurmountable barrier to technology. Lots of people get car sick, yet I’m certain this automobile fad is here to stay. The issue I see with a lot of these technologies that supposedly ameliorate VR nausea is they are attacking everything except the root of the problem. Our inner ears are the source of a lot of our sense of motion and balance. You cannot just slap a new experience on the visual sense and expect the inner ear to “get over it.” True virtual reality must provide a virtual sense to the inner ear or we’re going to keep puking. Next Steps: Don’t invest in VR companies just yet. Read More