Farewell, net neutrality The politics of net neutrality are a bit odd. Big technology companies in California and New York tend to favor net neutrality reflexively. Technologists have an idealistic streak and they believe deeply in the free exchange of information. Since net neutrality prohibits an internet fast lane for content providers willing to pay, it allows for a world in which any idea once online enjoys roughly equal availability. (In practice, gatekeepers still remain.) Digital infrastructure companies like cable providers or voice and data providers, hate net neutrality. There is money to be made in charging content providers to join the fast lane and this allows these innovation-adverse behemoths to return to an older broadcast business model. Since the president-elect won despite the active opposition of technology companies and will need the support of the digital infrastructure companies in order to govern, it’s hardly surprising that his FCC transition team is hostile to net neutrality. Why does this matter? This is a minefield of unintended consequences. On the one hand, this type of model will favor established media companies capable of charging higher advertising fees and paying for faster service. One can debate media bias until the end of time, but Trump did not win based upon support from the established media companies. Trump’s media and content support came from a loosely affiliated group of websites, bloggers and social media influencers collectively known as the alt right. These are precisely the types of poorly funded, grassroots content providers who stand to lose the most when net neutrality goes away. The lesson, as always, people rarely vote their self interest. Next Steps: If I were a smaller publisher or content provider concerned about the effect of net neutrality on my business, I would partner with Google to sell my advertising. Google has an interest in a free and open internet, but they will probably start by protecting their sources of income. Read More Friction and impulse control Apple Pay is trying to make it easier to buy. A single touch on the Touch ID banner of the new MacBook Pro will allow you to instantly make a purchase online. Your Apple Pay account can be associated with either your credit card or your debit card allowing you to make quick and secure purchases. While it is tempting to imagine that this little bit of added convenience will not affect shopping behavior, psychologists and neuroscientists say that it likely will. Any time technology allows for lower friction buying, people seem to buy more. The act of reaching into your wallet gives you time to reconsider the wisdom of your actions. The reason that ecommerce experts are always trying to reduce the number of inputs in a checkout is that each additional piece of information gives the purchaser more time to think better of their buying decision. Why does this matter? Personal debt is a significant issue for a sizable minority of Americans. As a nation, we appear to lack impulse control. While it profits the retailer to encourage individuals to purchase beyond their means, there are larger societal issues to consider. Some studies of the gaming industry indicate that their profit margins depend, not on the casual gambler out for a night of fun, but the problem gambler who loses far beyond their means. Similarly, creating an ecommerce infrastructure that depends on the manipulation of low impulse control shoppers is a morally tenuous position to take. I’m surprised Apple isn’t seeing that. Next Steps: Studies indicate that consumers purchase more rationally when they use their debit card rather than their credit card. If you decide to use Apple Pay, connect it to your debit card. Read More Probable Language When I was in high school, I decide to invent a language. The alphabet was a decimal-based system. The initial letter in each word designated the grammatical category of word and then every additional letter modified it until the total meaning was revealed by the final letter. And, for efficiency’s sake, every word meant both itself and its opposite. It was this last point that caused my very-confused English teacher to give me a C on the project. “A word cannot mean itself and its opposite.” he insisted. “By the time you reached the end of a sentence, the number of possible meanings of the sentence would be almost infinite.” As a fifteen year old, I didn’t have a good answer to that objection, but today I would suggest that infinite possible meanings is a characteristic of all languages. In fact, we know the meaning of words we read or hear only as probabilities. Context supplies much of the meaning in language, even as we wish that words possessed absolute meanings. Machine learning translation seems to succeed where direct translation technologies failed because it engages with meaning as probabilities, rather than definite values. Now the same machine learning technologies that have powered translation appear to have bested human beings in reading lips. (See article below.) The added indeterminacy of lip movement does not change the underlying power of probabilities in understanding language. Machine learning gradually narrows down the possibilities, until the most probable meaning is left. It might be wrong, but it isn’t unlikely. Why does this matter? By slow degrees, technology is turning away from certainty and towards probability. This is a major change in computer science. As we abandon mechanistic thinking, human-machine interface technologies improve. The development of quantum computing will allow us to accelerate this process since the underlying technology will be based on probabilities. Next Steps: Do call it “machine learning.” Don’t call it “artificial intelligence.” Words have meanings after all. Except when they don’t. Read More