On Thu, Apr 6, 2017 at 7:02 AM, Luv Khemani via bitcoin-dev <
bitcoin-dev@lists.linuxfoundation.org> wrote:

> Could you elaborate on why you consider ASICBOOST to be an attack? Attack
> here implies ill-intent by the practitioner towards the network as a
> primary motivating factor.

I think that it is an attack is a completely unambiguous technical
description of what it is. If a signature is supposed to resist forgery
against 2^128 operations, but you find a way to do it with 2^80 instead,
this is an attack. It is, perhaps, not a very concerning attack and you may
or may not change your signature scheme to avoid it or may just instead say
the scheme has 2^80 security. But there is no doubt that it would be called
an attack, especially if it was not described in the original proposal.

In Bitcoin's Proof of Work, you are attempting to prove a certain amount of
work has been done. This shortcut significantly reduces the amount of work.
It's an attack. Normally it wouldn't be a serious attack-- it would just
get appended to the defacto definition of what the Bitcoin Proof of work
is-- similar to the signature system just getting restarted as having 2^80
security-- but in it's covert form it cannot just be adopted because it
blocks many further improvements (not just segwit, but the vast majority of
other proposals), and additional the licensing restrictions inhibit

The proposal I posted does not prevent the technique, only the covert form:
That is, it doesn't even attempt to solve the patented tech eventually will
centralize the system problem. It is narrowly targeted at the interference
with upgrades.

Taking a step back-- even ignoring my geeking out about the technical
definition of 'attack' in crypographic contexts, we have a set of issues
here that left addressed will seriously harm the system going forward for
the the significant monetary benefit of an exploiting party. I think that
also satisfies a lay definition of the term: Something someone does, that
none one expected, that makes them money at everyone elses expense.

- Bryan
1 512 203 0507
bitcoin-dev mailing list

Reply via email to