3 ms·

It doesn't really solve the problem satisfactorily. Somebody still has to develop the software. The end result is a single download, produced by some form of a

It doesn't really solve the problem satisfactorily. Somebody still has to develop the software. The end result is a single download, produced by some form of authority. And typically there's one project, with one group controlling it. And miners are more or less the same everywhere, they want to make a profit. You can get somewhat better. You can find a dev group that aligns with your interests. Maybe you can even find a cryptocurrency where there are multiple independent implementations led by multiple groups with none being the clear winner. But that's an unstable situation. The dynamics of a dev team can shift as people come and go. Balance is easily broken, if a project loses steam, or one becomes the default and gains overwhelming popularity, and you're back to having a single group in effective control. You can migrate from one system to another, but any conversion has costs to it, both in terms of money and knowledge, attention and time resources. This libertarian utopia seems very much like some sort of bureaucratic hell to me. To exist in such a society without getting royally screwed you have to invest immense amounts of your time to pay attention to everything and correctly and prompty react to it. You go on vacation for a couple weeks, and oops, suddenly the dev team is changing the rules from under you.

tromp 4y ago
A properly decentralized project has multiple implementations, and a formal process for changing the consensus model. This requires the equivalent of an RFC (Request for Comments, or Bitcoin Improvement Proposal in case of Bitcoin) to be written, motivating and detailing the changes, followed by months long discussion to see if there is overwhelming support for, and few downsides to, the changes. Meanwhile, the changes need to be implemented on all existing clients with extensive documentation and tests and be exercised on test-nets to see if all goes well. Finally, a supermajority of miners as well as economic nodes (such as exchanges) need to be convinced the changes are for the better, and then an activation can be planned on main-net that could take several more months to lock in first. Chains where this process can be bypassed or severely shortened don't deserve to be called decentralized.
- dane-pgp 4y ago
  > a supermajority of miners as well as economic nodes (such as exchanges) Is there a formal way of defining this? Maybe it's possible to at least define "a supermajority X% of miners, weighted by blocks mined in the last Y days" (for values X and Y that have previously reached a consensus), but counting economic nodes in a way that is Sybil-proof and globally consistent, seems like it's just pushing the consensus problem somewhere else.

tromp 4y ago

A properly decentralized project has multiple implementations, and a formal process for changing the consensus model. This requires the equivalent of an RFC (Request for Comments, or Bitcoin Improvement Proposal in case of Bitcoin) to be written, motivating and detailing the changes, followed by months long discussion to see if there is overwhelming support for, and few downsides to, the changes. Meanwhile, the changes need to be implemented on all existing clients with extensive documentation and tests and be exercised on test-nets to see if all goes well. Finally, a supermajority of miners as well as economic nodes (such as exchanges) need to be convinced the changes are for the better, and then an activation can be planned on main-net that could take several more months to lock in first. Chains where this process can be bypassed or severely shortened don't deserve to be called decentralized.

dane-pgp 4y ago
> a supermajority of miners as well as economic nodes (such as exchanges) Is there a formal way of defining this? Maybe it's possible to at least define "a supermajority X% of miners, weighted by blocks mined in the last Y days" (for values X and Y that have previously reached a consensus), but counting economic nodes in a way that is Sybil-proof and globally consistent, seems like it's just pushing the consensus problem somewhere else.