producing GnuPG keys as proof of work

[Mirimir]

On 10/02/2014 02:58 PM, Robert J. Hansen wrote:
>> It struck me that a GnuPG key, produced as proof of work, would be
>> more generally useful than an arbitrary hash. But now I see that 
>> challenge-based proof of work could just be used to authenticate a
>> GnuPG key for signing and message encryption.
> 
> How?  What is there about the proof of work that can somehow
> authenticate a GnuPG certificate?

Sorry, I forgot to explain the context. On the tor-talk list, there's
recently been discussion of Wikimedia's policy to hard-block all Tor
(and other known proxy) IP addresses. It does that to protect against
malicious users who use proxies to evade bans and use sockpuppets.

As an alternative, I suggested that Wikimedia use challenge-response
proof of work to validate accounts created through Tor. Creating an
account might require many hours of computation, with the proper
threshold determined empirically.