Netcompany Snippets #5

Summer is about relaxing - but it’s also about embracing a bigger picture than you’re used to and asking yourself if you’re working on the right problems. For the fifth edition of Snippets, we add four classic texts to your holiday reading list. They all attempt to broaden the day-to-day perspective of what problems are valuable - and how to solve them.

Classic #1: You should always work on important problems

In 1986, retired Bell Labs scientist and Turing award mathematician Richard W. Hamming gave a talk at a seminar in the Bell Communications Research Colloquia Series, still widely quoted to this day. Hamming asked a simple question: why do so few scientists make significant contributions? His answer was simple: it isn’t a lack of brains. It isn’t a lack of luck, either. It’s failing to continually work on important problems.

Classic #2: Use trendlines to identify important problems

One way to find important problems might be to look at trendlines. In 1965, Intel co-founder Gordon E. Moore did that to observe that the number of transistors on a silicon chip tends to double roughly every two years. Observing such a trend was easy. Guessing whether the trend would continue was harder. That set of problems spanned physics, semiconductor technology & production, energy production, regulation of tech, and more. Looking at trendlines can point to a set of problems, but it cannot solve them.

Classic 3#: Use decentralized intelligence to solve complex sets of problems

In a recent Twitter thread, economist Daron Acemogly renewed an old debate around central and decentralized ways of solving complex sets of problems. With the arrival of AI, might a single mind be able to work out the entire problem set? In 1945, economist Friedrick Hayek anticipated this debate and argued strongly against the idea that a “single min [can] deliberately solve the problem set by [all the world’s] data”. Solving problems isn’t a problem of aggregating knowledge, Hayek thought. Rather, “it is a problem of the utilization of knowledge not given to anyone in its totality.”?

Classic 4#: Decentralized problem-solving requires global coordinationHayek, then, thought that problems (such as price levels) should be solved organically. In a famous blog post, Meditations on Moloch, rationalist Scott Alexander details how this decentralized problem solving process often runs into multipolar traps: what looks good from within the system often looks bad from without. One example is a version of the Prisoner’s Dilemma where two prisoners keep defecting because of an inability to coordinate. More up to date, fierce AI-competition is a multipolar trap: no individual is better off by ceasing AI-production, but the world as a whole might be. The problem is that such an optimum requires global coordination.

Use trendlines to identify important sets of problems, delegate work, beware of multipolar traps, and have a great summer!