Re: [Help-gnunet] "Abandoned" content problem?

[Benjamin Paul Kay]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hopefully I didn't miss this when skimming the Documentation and Papers.

What evil, bad thing would happen if a node could lookup all the
keyword/description blocks in its database corresponding to an arbitrary
URI? If a node delete such blocks out of its own database? Are these
things technically impossible?

I ask this only because it would seem that one obvious albiet not
complete solution to abandoned content would be to delete old keys. Keys
could be assigned a priority: each time a node successfully retrieves a
block for a URI, it increases the priority of all its corresponding
keys; each time a node fails to retrieve a block for a URI, it decreases
the priority of all its corresponding keys. Keys whose priorities reach
zero are deleted, since chances are that that content has fallen off of
the network.

This idea has a few detractors apparent to me. Decreasing the priorities
for all keys corresponding keys on a node could be a costly operation in
terms of CPU and database access since most URI's tend to have a lot of
keys associated with them. Also, this might not address the content
abandonment problem as directly as some would like.

I'm probably missing something obvious here, though. Why won't this idea
work?
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFBYLDuCA2gPtQP+LURAmG5AKDEDa41+u2Fywo1qSyqMYwqcM8gkwCfWcXm
nDF8OOGiOpJ53aw+X8pkeb0=
=e6Sv
-----END PGP SIGNATURE-----