This reminds me a lot of the rise of Git in the early days, when decentralised version control made the spectacular promise of allowing people to share diffs directly with each other without going through a horribly centralised system like Sourceforge or any other SVN based service.
Of course, nobody wants to send diffs to each other, that would just be silly. We only want the option to do so. So now that we can mirror and switch to GitHub or Bitbucket to our own Git servers, we ask for centralised systems to handle our decentralised version control.
I think crypto will go the same way. It’s nice to have the option to be completely decentralised and run your own node and network and what not, but the number of people running the networks are likely to be the same proportion as the number of people running their own git servers.
A system that is capable of easy decentralised is better than one that isn’t, but we choose central hubs or bases either way for day to day use.
> It’s nice to have the option to be completely decentralised (…)
I think "nice" downplays the importance of the possibility of self-custody. Despite the majority of people perhaps not opting to control their own keys, the mere possibility has interesting game theoretical implications that will keep neobanks like Coinbase in check.
I think you're underestimating the number of people that run their git server. given the fact that running a "git server" can be as easy as running "git init --bare" on an empty directory on a server for which your team has ssh access
In my country, at least, once upon a time, upper middle class people would use a person/business called "dhobhi" to wash and iron 90% of their clothes. Reasons included convenience, unreliable supply of water at home, low quality consumer detergents etc. For small communities, the dhobhis formed an oligopoly of centralized services.
When washing machines became available, people didn't buy them because they were expensive, unreliable, and required a lot of repair and maintenance. Ultimately, the washing machines became cheap, reliable, and low maintenance, and now most people use washing machines.
I think the analogy to github and personal servers is clear.
I'm not sure this analogy holds, exactly, for project hosting.
Washing isn't inherently collaborative, and the more one looks at the specifics, the more divergent the situations seem.
For example, Git(hub/lab) offer free project hosting while running one's own server would incur costs. This can be a big deal for open source, because paying bills or handling money tends to force a more complex structure on your "organisation".
For project hosting, there are network effects in terms of project discoverability, user account management, operational convenience (data durability, patching, ensuring uptime), consistency of interface and third party integration which tend to make using a centralised service more attractive than running your own, all else being equal.
It seems to me that only quite bad mismanagement of the major project hosting services could reverse this trend.
You are right that the analogy fails in a number of ways.
My point is that, besides legal issues that you point out, personal adoption of these services will grow a lot more, if hosting your server was actually easy. Right now, it is not, not by a large margin.
Of course, nobody wants to send diffs to each other, that would just be silly. We only want the option to do so. So now that we can mirror and switch to GitHub or Bitbucket to our own Git servers, we ask for centralised systems to handle our decentralised version control.
I think crypto will go the same way. It’s nice to have the option to be completely decentralised and run your own node and network and what not, but the number of people running the networks are likely to be the same proportion as the number of people running their own git servers.
A system that is capable of easy decentralised is better than one that isn’t, but we choose central hubs or bases either way for day to day use.