Longreads

• Jon Stokes on feeds and context. This is a very interesting refinement of the filter bubble thesis—it's not just what you read, but the order in which you encounter things. This is most obvious when something viral happens out of context; yes, a Senator did ask Facebook to "commit to ending Finsta," which makes it sound like he doesn't know what "Finsta" means, although earlier in the hearing he'd defined it, so it was just an inaptly phrased question, not a demonstration of ignorance. If you watched the entire proceeding, the line wouldn't have been newsworthy, but out of context it was a big deal. Since there are plenty of things that are more interesting when they're taken out of context, there's continuous selection against context in any popularity-weighted feed.

• Rohit of Strange Loop Canon on outliers, with details on how competitive tests do badly when they're known to be competitive. This doesn't mean people can't be ranked, but it means that the rankings have to be applied judiciously, and that they mean a lot more before they're prestigious. A good way to test this will be to track Y Combinator's hit rate on getting funding the startups that turn out to be the most valuable ones a few years later; YC does seem to do a great job at assessing quality, so if their hit rate goes down it's a good measure of a fundamental limit.

• A fun piece on where Tether's money is, and where it isn't. This does not end up solving the longstanding mystery of Tether, though the piece does end up finding one bank that claims to have $15bn of Tether's money. One ongoing theme in Tether's history is that some of its managers are gun-shy about risk, and the other managers buy them out. So there's an evaporative cooling effect, accelerated by the fact that Tether managers who are willing to use its funds aggressively naturally have more buying power than the ones who don't.

• Chris Dixon on Why Web3 matters: the life cycle of most successful companies starts with a period of usage growth and then moves on to more monetization, i.e. they begin by creating value and end by capturing more of it for themselves. This does make them more fundable, but it's not the only model, and token-based funding can spread that equity around in a way that keeps everyone aligned for longer. There are probably some categories where that model is less effective than a more concentrated one—in an industry with lots of execution uncertainty, like biotech, it makes more sense for there to be a short list of investors who have done substantial research. But for many consumer-facing products, a tokenized model is promising.

• This FT profile of Farouk al-Kasim, one of the architects of Norway's oil policies, is worth reading. It's partly a look at how contingent history is—he moved from Iraq to Norway to get better healthcare for his son, and wasn't aware that Norway was looking for oil at the time. And it's also a look at how nonrandom it is: for many countries, discovering natural resources is a net negative for the average standard of living, because of the ensuing economic distortions, but Norway managed it well.

Open Thread

• Drop in any links or books of interest to Diff readers.

• What are some other good examples of Norway-like outliers, where a country had exactly the opposite of the expected impact from a given economic change? And are there any good examples in the private sector?

Leave a comment