The Federation Fallacy
The Federation Fallacy
An interesting article I saw (from 2019) describing the potential intrinsic tendency for decentralized platforms to collapse into de facto centralized ones.
Author identifies two extremes, "information dictatorship" and "information anarchy", and the flaws of each, as well as a third option "information democracy" to try and capture the best aspects of decentralization while eschewing the worst.
Someone said the link is broken so here it is: https://rosenzweig.io/blog/the-federation-fallacy.html
You're viewing part of a thread.
Maybe I'm not quite following you. Out of 3070 instances, 50% were registered in just the top 3. Whether it's the top 3 or top 1, doesn't this clearly show a tendency to cluster into large centralized monoliths? This type of power-law clustering is ubiquitous in all kinds of human behavior, but this would correspond to a really high exponent value for the distribution.
3 is a different number from 1. If a single instance had over 50% of signups it would be reasonable to describe it as dominated by a single big player. If the biggest instance only has 20% or whatever the reality is, then it is not dominated by a single big player.
Definitely there's a tendency to centralize up from thousands of little shards to a few big professional units - though as we see in every one of these examples, that doesn't mean the little ones have to disappear. You still have plenty of small email clients and small instances. What's important is that if one big one goes down or goes evil the other big ones are there, and that there's always the possibility of new small ones blowing up if they do something better than the big boys.
Then it seems like we're in agreement. The thrust of the article is not that the fediverse is bad, or that it doesn't improve upon the FAANG cartel model, but rather that decentralization by itself is not a silver bullet. It also needs to include the "bottom-up consensus seeking" that e.g. Wikipedia uses for its decision making.
Yes I get what the article was arguing. My critique is that it doesn't seem to have a firm grasp of the fediverse model, since it thinks there's something problematic about the sizes of instances follows a power distribution and refers to "the federated ideal, where all instances are created equal" in the sense of having the same number of users.
They clearly don't think that, since right after they said:
Of course, due to substantial inequality between instance size, we expect to see a power distribution, with a spike on the left that quickly falls and tapers out. Power laws govern much of the real world; many phenomena behave according to this unequal distribution.
But look at that graph. Calling this distribution a power law would be generous to say the least. There is a massive spike corresponding to just a few instances, and the rest of the graph is nearly invisible to the naked eye, so tiny and so overshadowed by just a few giants. Frankly, this distribution is closer to the Dirac delta function than a power law.
All of your comments in this thread seem to have been a strong defence against perceived criticisms of the Fediverse, when the article isn't so much criticizing it as it's proposing that it needs additional safeguards (besides decentralization) to ensure it remains aligned with users' interests:
-
Firstly by alleging that the author advocates everybody running their own instance
-
And now alleging that the author only considers Fediverse successful if all instances are the same size.
These are really disingenuous and uncharitable ways to read the article. I'm sure the author has a perfectly good understanding of the concept of Federation, as they detailed several examples, and they have a lot of experience in highly technical computer science work.
I'm fine with criticisms of the fediverse, my issue with this article is how the author repeatedly makes these negative comparisons of the existing fediverse to some 'dream' of what it is supposed to be like that seemingly exists only in the author's own head. You can see in each of my quotes where the author makes claims about how the fediverse should be much more decentralized than it actually is to live up to that dream, even if he doesn't necessarily claim to agree with that dream himself. As to the "does three equal one" question - clearly having three big instances sharing half the space and a long tail of thousands the other half is a very different scenario from having a single dominant instance.
-
3067 is a lot of ways to slice half a pie. I'd consider even 16.5% (or whatever the top dog of that 3 with 50% has) to be domination.
Hmm domination in what sense? Maybe in terms of winning the competition for biggest instance, but clearly that's not big enough to impose their will on the whole.
It really depends. If you're in a smaller instance and you look at the global view, you're going to see more of Mr. 16.5% than one of the smaller ones.
Though I suspect usage patterns and the way users interact with instances beyond theirs will play a role. But, in an immediate sense, I could see larger instances having a bigger voice (so to speak).
And now I'll waffle and say it's all a crapshoot because people are unpredictable and social media platforms even more so.
I agree large instances have a bigger voice proportional to their larger size, but I don't think that's really an issue as long as there are plenty of instance options and no single one is so powerful it can force the system to conform to it rather than conforming itself to the system.