On Distributed Systems and Catallaxia

In the introduction to “Law, Legislation and Liberty”, FA Hayek discusses about the nature nurture debate. He notices that most institutions are not part of nature, but that they are no longer the product of a single mind. No one woke up one day and set the grammatical rules of english. It is the product of exchanges between many people and across many generations. In his view society emerges from interactions and should not be confused with enterprises in which there is a clear goal. Typical example being an army trying to win a war. Both of those human constructs have their own merits, but they do have to follow distinct rules in order to be efficient. Mostly everything turns around how information is spread. Something like top-down vs bottom-up. New information technologies have changed the landscape and new modes of organization emerge. The first one that comes to mind is the distributed system framework. The fact that we do have systems like Bitcoin, beyond the hype and all the non sensical financial games, working is going to change society in depth.

Nudges and Gravity Fields

There is no magic trick: one needs to receive information one way or another if one wants to participate in a large project. The traditional ways to organize communications are the one on the left and in the center of the image. Information can go top-down (the general in chief gives orders) or bottom-up (information gathered at the periphery flows upward, think polls, elections etc.). Distributed systems are relatively new. The fundamental difference with centralized and decentralized is the equivalence of nodes in the system. From the point of view of the network any node is equivalent to any other node. It is not the case for both of the other ways to organize the information broadcasting.

Social media is a good example of a, seemingly, distributed system. On Twitter, theoretically at least, every tweet can be seen by anyone. Of course there are servers and an AI to determine what your TL is, some nodes may block other nodes or restrict who can see their tweets, shadowbanning etc. But from a functional point of view the default is that the whole world can see my tweets in the same way that the ones of the POTUS. This is fundamentally different than a centralized system like CNN for instance where there is an actual boss who decides what is going to be displayed (even if he doesn’t control who is going to access the information) or not.

The strategies to organize societal projects are more or less known in centralized/decentralized systems: there are the authoritarian ways - the chief gives an order and soldiers execute - and the market ways - the price is used as signal by actors who independently decide what their next action is going to be with relation to their own goals. Both of these approaches do not work for distributed systems. On one hand you can’t force people to tweet about a subject, on the other hand there is no clear equivalent of a “price” signal. Even if likes/retweets, etc. try to play this role, they are attributed after the action was done, and can’t be compared to prices in the real world. Moreover the action in itself is free: post, like and retweet are free. It is therefore hard to see an action from a return on investment point of view.

Nevertheless there are huge trends on social media. Where do they come from? Are they *organized*? And if yes how? It is very hard to answer these questions because the back end of Twitter is not public. Yet we can make some educated guess about those questions. There are myriads of technical means to consider but they basically boil down to a single general idea: playing the numbers and relying on “Pre-suasion”. The raw idea is that if there is italian music displayed in a wine store, you are more likely to buy italian wine. This is simply because the music reminds you the holidays you spent in Venice and one thing leading to another you end up buying Chianti. If you only *see* tweets talking about X, naturaly you are going to think about X and next thing you know you are part of the buzz around X. No one forced you. You could have tweeted about the music you are listening to or about the weather, but statisically you will be sucked in into the buzz. From the commanding side it is also a bit disturbing. The most efficient way to lead is no longer to give direct orders but to set up a sort of gravity field, or to *nudge* users as it is said today. A major feature of this gravity field is that it has to be invisible otherwise the spell is broken. Hypocrisy as a service…

Hydra Vulgaris

Resiliences and growth

Resilience is one of the most interesting feature of distributed systems. Because nodes are interchangeable by design, the removal of a node does not impact the network. This has been highlighted many times for Bitcoin. Even when China emited a ban on mining, it only took few hours for the system to adapt. On one hand many other places took the job and on the other hand many chinese miners found ways to hack the surveillance of CCP and returned back to mining. This is not nothing. Makes you think at the Hydra of Lerna.

Linked to this resilience property, another feature of distributed systems is paramount: the network infrastructure grows organically along with its acceptance. Said differently: the more people are using the system the more efficient it becomes. It is the polar opposite of centralized systems that have to scale up their infrastructure at the same speed as their use. Think about the computer farms of Google or Amazon. They are gigantic investments. On the other hand Satoshi Nakamato presumably never bought more than laptop and a standard access to an internet provider, yet the bitcoin network is now using more electricity than some nations. The fact that it can be justified or not is beside the point. The point is: there were 0 meeting nor investment from “Bitcoin corp” to achieve this. Miners did this on their own.

This appears to be positive remarks, but they can be flipped. Surveillance can become distributed and acquire similar properties. The best example is the use of smartphones to enforce security policies. As the experiment done during the covid crisis you can set up a checkpoint society in a matter of days. It is just enough to set up rules, emit QR-codes and an application of checking the QR-code that produces a green or a red light. The database is centralized but the security system is distributed: anyone with a smartphone will become part of it. From the waitress to the bank employee passing by your local baker. Everyone of them will check that your credentials are ok. There is no need to have police at every corner of the streets. There is no need to pay the checkers either: negative incentives (fines if they don’t obey) are enough. It is essentially free (in terms of funding not in societal terms) and also much more robust than any police only enforced policy.

The Distributed Delusion

Ten years ago E. Morozov published “The Net Delusion - The Dark Side of Internet Freedom”. In this book he deconstructed the myths that internet was only for the good and was going to liberate humankind by making access to knowledge free. There should be a sequel about distributed systems because I have the feeling that there are too much hope of their liberating consequences. They can also lead to the most efficient prison of all time.