What’s the deal with decentralisation?

Penned on the 23rd day of May, 2021. It was a Sunday.

The cutting edge of today’s real politics seems to revolve a lot around new motions towards what people have come to call decentralisation. In finance the term is “DeFi”, short for “decentralised finance”, and it seems like the very thing everyone is hoping for to shake up the existing order and fight off the glaring problems and inadequacies of our incumbent tech platforms. But what is decentralisation, anyway? And what relation does it have to the concept of federation, if anything? Is it really just the end of monopoly and tyranny in the hands of powerful tech overlords, or is it something else altogether?

In fact, decentralisation is a very fluid and multi-faceted concept. It’s a variable amalgamation of a lot of qualities of the software we use that have been deemed desirable for one reason or another. This includes the ability to self-host instances of the software, instead of having to rely on a single corporation that does hosting and keeps the source code to themselves. It also often involves federation, as in the case of Mastodon, where many servers all over the web can interconnect and share content as part of a “fediverse”. Another more dubious angle of decentralisation is the involvement of the Ethereum cryptocurrency, specifically its “dapps” feature that provides a decentralised virtual machine that people can run code on by spending Ethereum “gas”.

Regardless of what technical aspects are being furnished, the political message of decentralisation is almost universal: take back control of your digital life by regaining ownership of your data. This follows logically from the way in which big tech companies keep users on their platform: monopolising the content they post and information about that content. So it stands to reason that if control of data is taken back and put into the hands of users, things can go back to normal and these problems of big tech just go away, right? Well, not quite.

When people push initiatives to decentralisation, they often come with the catch that such initiatives will nominally imply guardianship over people’s data, with theoretical claims or promises that it is totally feasible for users to really own their data if they really want to. An easy example of this is someone promoting a Mastodon server, where custodianship of the data would then be theirs to have, not the users. This is excused by the facilities Mastodon provides to export user data, and network with other servers in the “fediverse”, and Mastodon’s very restrictive GNU Affero GPL licencing. But at the end of the day, user data has just changed hands from one custodian to another. If a user wants to leave and start their own place, they face the same exact problem presented by Twitter, Facebook, or any other supposedly centralised tech platform: moving one’s real social network. The truth is, moving platforms is easy. Getting your friends to go with you is hard.

This same gotcha is also present in efforts that use dapps, where it is even more hidden and pernicious. Proponents of such dapps may know about the catch-22 presented above, and even Mastodon’s failure to solve it, and counter with the claim that through dapps, it truly is solved. You really must own your data, because you can log in to your ship, or domain, or whatever it may be called, and no one else has control over that except you thanks to cryptography. This shoves under the rug the critical weakness of dapps: it is dependent on the currency running beneath it. People who are very wealthy in Ethereum will have vast quantities of gas to use more resources than you could ever hope to afford. And if you have no gas, you can’t get your program to run! That seems pretty silly since we already own computers with ample processing power in our own two hands, doesn’t it? It matters not even how cheap such tasks are made to appear, since economics is still the core mechanic at work. It’s not just code, it’s code over money. Given the dubious nature of cryptocurrency at large, this seems downright suspicious.

Samo Burja once wisely said, “The cloud is just someone else’s computer. Decentralization is just someone else’s center.” With cryptocurrency or not, with federation or not, this seems to hold true. Dapps are a balsa wood stand-in for real digital infrastructure anyway, so it’s pretty natural to find that “decentralisation” is much the same for problems that are fundamentally to do with leadership, not technology. You can’t apply technical solutions to non-technical problems. So, what’s next?

Until next time,
Άλέξανδερ Νιχολί