Month: December 2024

I’m currently reading Yuval Noah Harari’s “Nexus,” where an interesting story about a Romanian computer scientist named George Iosifescu is described (see my previous notes on the topic – #raufnexus). In 1976, Iosifescu came to his office and found a man sitting at his desk. Iosifescu introduced himself, but the man did not respond. He went about his business while the man just quietly sat, watched the computer screen, and made notes. Overall, it was apparent that he had arrived from the Romanian secret police.

It’s intriguing to look at work through the eyes of this agent. Imagine you have breakfast and then go spend your entire day in someone else’s office in silence, jotting things down. And so for thirteen years! Thirteen, Karl!

For thirteen years, Iosifescu came to work, and the man was always there, at the desk, to observe and record something. It all ended only with the fall of the Romanian government. Harari writes that Iosifescu never found out the man’s name.

But if you think about it, by 2024 we essentially have the same situation, only our agents are digital, numerous, and the data they collect are somewhat scattered and disconnected. Harari quite rightly notes that we are approaching the moment when a pervasive computer network will be able to monitor the population of entire countries around the clock.

Harari also has an interesting story about the mass collection of handwriting samples in Romania. When the Romanian regime discovered anonymous letters criticizing them, sent to Radio Free Europe, Ceaușescu ordered handwriting samples to be collected from all 20 million citizens. Students wrote essays, and adults wrote autobiographies, which were then passed to the Securitate archives. Even typewriter owners were required to register them and provide sample texts.

It’s interesting—they collected them, but how were they supposed to search through such a volume of information? It was probably not intended at all. Perhaps this database was used to create evidence—if someone was suspected, data from various databases were collected about them, and attempts were made to use it in their case. If something accidentally or not coincidentally matched, then that’s it, case closed, person imprisoned.

* * *

I also liked the idea of information networks.

In tribal societies, where there were no written documents or bureaucracy, the human network consisted only of two chains:

1) “person-person” and

2) “person-story” (forming the chain person-story-person).

Power belongs to those who control the nodes connecting different chains. These nodes are the tribe’s foundational myths. Charismatic leaders, orators, and creators of myths were adept at using these stories to shape identity, create alliances, and impact emotions.

In human networks associated with written documents and bureaucratic procedures, society is partly based on interactions between people and documents. In addition to the chains “person-person” and “person-story,” such societies are held together by the chains “person-document.”

Thus, there are three types of chains:

1) person-person

2) person-story

3) person-document

This has led to changes in the distribution of power. Since documents became an important node linking many social chains, significant power was vested in them, and experts in the complex logic of working with documents became new authoritative figures. Administrators, accountants, and lawyers mastered not only reading and writing skills but also the ability to create forms, organize archives, and manage bureaucratic processes.

The person-person chain can be related to the family, the person-document—to the church.

But there are two more chains that have emerged quite recently:

4) computer-person

5) computer-computer

An example of “computer-person” includes social networks like Facebook and TikTok. These chains differ from traditional “person-document” chains because computers can use their power to make decisions, create ideas, and deeply manipulate proximity to influence people in a way that no document could. The Bible has had a profound influence on billions of people, even as a silent document. Now imagine a sacred book that can not only speak and listen but also recognize your deepest fears and hopes and continuously shape them. Indeed, algorithms on social networks (and not only), according to Harari, can influence human behavior on their own, not because the authors of these algorithms embedded such behavior in them. A good example is the accusations against Facebook for inciting hatred in Myanmar, as the social network promoted posts that gained popularity, which were actually about genocide.

Secondly, Harari writes, “computer-computer” chains are emerging, in which computers interact with each other autonomously, and this interaction is rapidly gaining complexity, which can no longer be untangled without the help of other algorithms. Algorithms of this interaction were developed and debugged on volumes that were still understandable and could be troubleshooted, but then it starts “magic”. An example is trading bots and algorithmic trading in general, and for instance, the “flash crash” of 2010, when the U.S. stock market crashed by trillions of dollars within minutes, only to recover just as quickly. The causes of this flash crash were later found, but it was not easy (and there were several reasons).

#raufnexus