If you have concrete suggestions for what the XSF should be doing, and/or
how servers could defend themselves against spam and DDoS, I'd be
interested in hearing them.

My understanding is that they're both difficult problems to tackle without
a lot of data processing and analysis, but that a key issue is that freely
available old-school IBR means that - I'll quote something Philip Hancke
told me this morning - '"public server" means "open relay" most of the
time.'

Whilst it's possible to mount an XMPP-based DDoS without IBR, it's much
harder, and forcing people to manually create each account *will* slow them
down.

The other thing would be to set up central services for verification of the
usage of email addresses, which we could do. I'm thinking loosely in terms
of a Bloom table for email address hashes. Needn't even do more than Bloom,
actually, which makes it quite cheap, in terms of resources - since if
there's a collision, we can either reset the bloom, or the admins could
manually accept.

But I'm thinking aloud, and somewhat pre-empting your comments.

Dave.

Reply via email to