Anonymous server encryption

[L. Detweiler]

faust at cd.chalmers.se brings up some interesting points about remailer
encryption, suggesting the capability to re-encrypt a message at the
remailer point instead of a-prior by the sender. This was proposed to
allow replies to a anonymous user to be encrypted using that user's
public key, avoiding the problem of the final message being transmitted
in cleartext.

The sophisticated server run by D. Clunie around December for a few
months allowed exactly this basic and valuable capability, and so far
has been unmatched. In his system the user sent messages to the server
encoded by the server's public key. Replies to the user through the
server were first encrypted with that user's public key before being
transmitted. This means that security at the remailer-to-`nym' is
protected even if respondents do not encode their messages first. 
Despite being fairly simple to implement (esp. with the availability of
PGP) this is not yet available on anon.penet.fi. J. Helsingius has
repeatedly reaffirmed his intent to put the feature in a New & Improved
Mark II server due sometime in the fall.

I've also suggested some improvements to anonymous servers. In
particular, I would like to see a sophisticated `nym-management' scheme
whereby the user can allocate, redirect, and deallocate addresses at
will. Such flexibility and versatility allows many other neat
unforeseen uses of the anonymous server. For example, many sysadmins
are in the habit of creating new email ID's for each new query posted
on Usenet to aid in organizing simultaneous queries. With
nym-management a user could do this as well. The user could also
receive mailing lists at different addresses. This would be useful
under a very radical new capability: local archives. In fact, the
transition from anonymous servers to personal rented file spaces may be
automatic and seamless in the future.

Unfortunately, while there's been a modest proliferation of cypherpunk
remailers, and quite a bit of general interest (the recent summer
Usenix anonymity conference was attended by a few hundred, see
comp.org.usenix), the only server allowing anonymous responses is still
Helsingius's anon.penet.fi. (I've not heard a lot about Kleinpaste's.)
It seems to me this is an extreme weakness in cyberspace at the moment.
When his server went down recently for a few days (from 90 degree
weather, damage to the system board) many people were `stranded'. In
fact, the only link between a lot of people and their correspondents is
that server and its nym-to-address database mapping file, and if
anything happened to either many would be disconsolate at minimum.

I suspect operators are reluctant to run a `full-fledged' anonymous
server because of all the cyberspatial politics & hassle involved for
virtually no personal benefit and serious drain on personal time,
finances, and cycles. When someone sets up an FTP site they at least
are creating a sort of personal library and BBS for perusal. But the
anonymous server operator actually is ethically restricted from
completely perusing the data spawned by his creation. Hopefully the
field will change when there are economic incentives to run an
anonymous server. For now we will have to rely on selfless heroes and
paternalistic voyeurs. I'll leave it to the reader to sort them out.