Some simple changes to the tor architecture I believe may greatly improve it (errata)

glymr glymr_darkmoon at
Mon May 15 16:05:17 UTC 2006

Hash: RIPEMD160

I forgot one extra idea - that of generating a number of different
circuits to be used at any one time (generation could be spaced say 5
minute intervals by the client, circuits more than 4 or maybe 5
generations back could be expunged in a fifo type arrangement) that,
whenever possible will select a random route to establish any new
connection from one of the number of circuits that are built. The
purpose of this specifically is to address the reduction of timing
attacks that are possible against a website - if the client sets up say
3-6 different circuits to retrieve the various different components of a
web page, each circuit will have a different time signature (due to the
viscisitudes of the network) and will make it harder to definitively
establish the origin. (and of course every node being both client and
server further obscures this)
Version: GnuPG v1.4.3 (MingW32)


More information about the tor-talk mailing list