[Standards] Proposed XMPP Extension: Extensible In-Band Registration

Goffi goffi at goffi.org
Mon Feb 13 16:14:02 UTC 2017

Le lundi 13 février 2017, 09:24:47 CET Sam Whited a écrit :
> On Mon, Feb 13, 2017 at 7:28 AM, Goffi <goffi at goffi.org> wrote:
> > - proof of work would be really nice, with a fallback mechanism.
> If by a "fallback mechanism" my understanding is correct and you mean
> "something to fall back too if the client does not support the
> particular proof-of-work function the server is asking it to compute",
> we can't do this.
> If you allow fallback, spammers just won't support the POW function
> and will register a ton of accounts anyways. If a server advertises
> that it requires a POW calculation before registration can be
> completed, it will need to always require support in all clients that
> register on that server.
> —Sam

I was thinking about a human challenge (question specific to user), which need 
manual intervention, while proof of work can be fully automated (but expensive 
for spammers).


POW client request registration => server request POW => client compute => 
server check, OK


NON POW client request registration => serveur request a human challenge 
(captcha, whatever) => client user answer => server check, OK

the later is more annoying because it needs human intervention while the POW 
client can do everything automatically.


More information about the Standards mailing list