[Cryptography] Standards Trolls: Re: Bitcoin is a disaster.

Greg greg at kinostudios.com
Mon Jan 11 22:56:57 EST 2021


> On Jan 11, 2021, at 1:57 AM, Ben Laurie <ben at links.org> wrote:
[.]
> Wikipedia mentions peer-to-peer applications in paragraph 1, so I'm not sure I agree with this view. But that hardly matters because I am not debating the meaning of decentralised vs. distributed but instead pointing out that their definition looks like the latter more than the former. Perhaps you'd prefer federated?

I mean decentralized, and the definition that was given in the paper. Federated can be a form of decentralization, for sure, and the paper goes into that somewhat indirectly in section on the relativity of decentralization.

> I'm not sure how I can be more clear - if a system has no central control (distributed, federated or otherwise)

If it has no central control, then it isn’t distributed. This is the definition of decentralized we’re going with, and definitions are important, otherwise we’re talking past each other, so let’s stick to them. If you do like the definition, that’s fine, but for the sake of being able to understand one another, let’s stick with the definition given in the paper.

> , then there is nothing that can stop it from having two or more partitions that stay partitioned forever. A partitioned system cannot successfully process all authorized messages - assuming that "success" includes, for example, prevention of double spend.

It’s unclear to me where this thinking is coming from. If you were able to partition the system and keep it that way, then it implies there was a central point of control. A contradiction. Again, see the paper's section on the relativity of decentralization.

For example, the Internet is decentralized, but that doesn’t prevent some parts of it from being partitioned off. For the sake of definitions and defining what is System A and what is System B, we say that "North Korea’s Internet" is not the same system as “our Internet”, and consider these to be two different systems.

When you add consensus to it, there is additionally a way of dealing with partitions. So if one system became two, and then one again, or if two systems became one, the process of consensus is a way of resolving the conflict.

For example, in Bitcoin specifically, temporary partitions have a way of resolving, and indeed this happens daily. If a permanent partition is created, it is enforced at a higher level than the level that Bitcoin itself operates at. Bitcoin remains decentralized within both of the sides of the partitions, and continues to remain decentralized if they merge again. The mere existence of a partition doesn’t prevent Bitcoin from doing its job, it just creates two Bitcoins (indeed, we’ve seen that happen a few times already, for different reasons, and they’re given different names).

For bitcoin to not be decentralized, that is certainly possible, and even has happened momentarily in the past, and may happen again in the future. But this is a separate discussion from the notion of partitions, which I feel miss the point.

Cheers,
Greg
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: Message signed with OpenPGP
URL: <https://www.metzdowd.com/pipermail/cryptography/attachments/20210111/466c0521/attachment.sig>


More information about the cryptography mailing list