What the computer sees. For the longest time, computers didn’t see much of anything. Even when a computer was capable of taking a photograph, it had no way to parse or interpret the content of that photograph. But one of the advantages of machine learning technology is that it has introduced the new field of computer vision. Computer vision is a capability that allows computers to recognize objects and to accurately map them within a three dimensional environment. If you’ve heard technology pundits suggest that the camera will replace the keyboard, they are referring to the incredible possibilities suggested by computer vision. For augmented reality to work even slightly, we require far more advanced computer vision than is currently commercially available. Limited AR games like Pokemon Go rely on approximations of locations based on GPS data. But AR startup Blippar suggests they have found a better way. They use a trove of location-identified images to map people within a three dimensional environment, based on what they are seeing. If the system sees Chestnut Street in San Francisco from a certain angle with certain stores visible, it knows your physical location in the world. This allows Blippar to more-accurately place virtual objects within the physical environment. Why does this matter? To be honest, this feels like a kluge to me. (def. “a software or hardware configuration that, while inelegant, inefficient,clumsy, or patched together, succeeds in solving a specific problem or performing a particular task.”) The dream of computer vision is that it recognizes objects and has depth perception in any environment, the way a human toddler does. This company seems determined to recreate the world at 1:1 scale in order guess at your position. It’s like something out of a Borges story. This approach might (MIGHT) work if you are Google and have spent the last decade taking pictures of every street in the world from every angle. But Blippar is clear that they are specifically not working with Google on their technology. Regardless, computer vision has already moved beyond this. Autonomous vehicles require computer vision technology that is capable of recognizing roads, sidewalks and obstacles despite having never been in an environment before. Blippar may be somewhat useful until that technology is commercially available, but I wonder about their long term prospects. In a nutshell: Computer vision is like Hansel. (So hot right now.) Read More I’m really talking about QR codes. Am I? Am I really going to talk about QR codes? Yes, I am. Even though QR codes are kind of a joke in the United States, they are more widely used in Asia, particularly in China. In a recent blog post, Connie Chan of Andreessen Horowitz writes about 16 ways QR codes are being used in China. Chan claims that “QR codes let you hyperlink and bookmark the physical world.” Respectfully, I’m going to call that statement a bit overblown. QR codes allow you to open a web page on your phone without typing in a URL. That’s nice. But it’s not really world-beating functionality. Still, it is interesting to see how QR codes are being used in a variety of ways in another culture. While a variety of examples are given, there are really two use cases – providing an identifier to an object or person and providing a link to either information or a payment portal. Why does this matter? Superficially, the role of technology in culture looks broadly similar from country to country. We tend to describe differences in the use of technology generationally, rather than culturally. However, there are serious differences in how technology is used between cultures. I don’t think anyone should read Chan’s post and rethink QR codes. They simply aren’t used here and computer vision will soon replace even the limited use cases they have. But it is worth thinking about how our digital culture differs from that of other countries. There is no straight line of technological sophistication from hunter gatherers through to Silicon Valley executive. Rather, technology and culture will fragment and coalesce according the utility and habit. I may find QR codes silly. But they are useful in China. In a nutshell: Technology varies more than you might think. Read More Agile at Scale Using an agile methodology at a startup is comparatively easy. Everything is new anyway, so developers and project managers expect to adapt to a new workflow. Groups tend to be small and entrepreneurial and there are no old habits to unlearn. But how can you change to an agile approach when you work at a large, well-established company. McKinsey has published an interesting interview with Scott Richardson, the Chief Data Officer at Fannie Mae. In the interview, Richardson describes his methods for converting a large and established team to working agile. First, Richardson says that it is important to pick your first teams carefully. You need early wins in order to convince the larger organization that your approach makes sense. Inevitably, there will be crises and people will want to revert to the old method of throwing more bodies at the issue. Don’t. Instead, use a crisis as an opportunity to double-down on your agile approach, encouraging the team to re-evaluate priorities and reconsider the scope of the MVP. Training is crucial to a successful agile workflow, but training dollars may be limited. Constantly rate your teams to assess maturity level (in agile methodology, not emotional maturity.) That way training can be applied where it will do the most good. Above all, agile methodology makes the most sense and works the best in a customer-focused organization because customer needs can define the features of the MVP. Why does this matter? I talk to executives at large companies all the time who want to transform their business to be more digital and more customer-focused. Frequently, they imagine that the right partner, technology, or initiative will allow them to leapfrog the competition and become a technology leader. Actual “digital transformation” is not something you buy or something you hire a consultant to do for you. Digital transformation requires companies to change their way of doing business. It is hard and it always encounters intense resistance. That’s not because your employees are lazy or unwilling to change. The reality is that switching over to an agile methodology (from waterfall or whatever your company currently uses) will initially make things slower. This will be frustrating to your best employees who are used to “getting the work done.” It’s only over time and with training that using an agile methodology improves the work and the output. Short term sacrifice leads to long term gain. In a nutshell: Agile will make you slower before it makes you better. Read More