I just finished reading Nexus. This book is a very interesting and unsettling read. Highly recommended, particularly as we enter the AI summer.

Here are my notes. Some are my ideas, not his (please don’t consider all the notes as things I say he said):

• He’s defining some things as delusions (stalinism, nazism) while implicitly leaving others as non-delusions (democracy, science). But then he argues that the populist view means there is no objective truth and everything’s constructed. What, to him, is non-constructed and true in itself?
• He’s focusing on the networks rather than the information (information is the glue that keeps networks together). The way I’m seeing it now, the information constructs the network, makes the network, even is the network. It’s not just a glue.
• OK, his concept of truth is between complete subjectivity and self-serving (marxism/populism) and the naive view (positivist/capitalist/technocratic).
• Perhaps startups also have mythmakers (the founders) and the bureaucrats (those that come after and make it a normal big business).
• Property as information?
• Happy that Harari’s view explicitly sidesteps the “us against others” mentality.
• I’d love to understand why Harari states that Aronsohn’s death by suicide is an objective fact. What makes it so? Is it the quality of the sources that indicate it, vs the quality of those that say otherwise? What construes a fact socially (and perhaps even in those who experience it first hand) is also a process. To understand the extreme where we say that “x approaches truth” would be very interesting to understand. I think that the less we use the term “obvious” or “objective”, the better.
• His definition of information as an array (“in formation”) is really good and workable. Information is pattern that can create further pattern. If truth is perception (whether first-hand or social), truth is also information.
• What makes “truthful” information more “expensive”?
• A brand as a story.
• The two drawbacks of truth: can be complex, can be negative.
• Bureaucratic lists may be boring, but silly lists can be fun. The question is: can meaningful lists be fun, or by necessity must they be boring if they are useful?
• The alchemy: turn truth into order.
• Tel Aviv: old new land.
• “At the heart of the bureaucratic order, then, there is the drawer.”
• Athenian democracy was installed after Sparta overthrew Athens’ tyrant.
• The Holy Book was assembled by committee.
• The printing press critically fanned the fire of European witch hunts.
• What makes it possible in science to self-regulate/self-correct? Why can science identify errors reliably, where churches cannot? Is it because of the nature of what they study, which makes it ultimately reproducible by anyone?
• Harari’s definition of scientific institutions: those that have systematic self-correction as their main activity. Makes sense why Steve Yegge criticizes those criticizing him for criticizing Lisp.
• Why if governments succeed exposing the mistakes of the previous, there can be still no progress? Is it perhaps because they throw everything away, rather than the mistakes?
• Modern technology made possible both large scale democracies and totalitarian regimes
• The innovation of computers is that they allow communication from computer to computer without a human in the middle.
• 1-1.5% of all human energy usage goes to computers!
• The first two large scale democracies: the Polish-Lithuanian Commonwealth and the Dutch Republic.
• The Dutch invented the newspaper!
• The interesting thing about the secret police is that it bases its power on information it has over everyone, even the rulers. And it uses that information to control and repress, often extremely effectively.
• “Information is not truth.”
• I disagree with Harari. Information is not mere connection, it is representation.
• I agree with his notion of error-correcting mechanisms, but we needs a better operational definition of truth. Because if we don’t define truth clearly, we could argue that the Soviet system was self-correcting in that yesterday’s orthodoxy was purged today. Why are the Stalinist purges not considered as an error-correcting mechanism?
• Is truth objective, intersubjective, or a combination of both? What distinguishes (or creates a continuum) between the truthness of different intersubjective constructions? This is the central problem, and it has to be approached heuristically and operationally.