...

digital signatures for cancellation might allow "uncancellable" messages, which has some attraction in the current climate.

What's uncancellable? If someone starts abusing things with "uncancellable" messages, then the USENET backbone gets together and stops passing these "uncancellable" messages from site to site. The messages will die very quickly, I assure you. If you change the names/message ids/whatever, then *all* such messages will end up getting nuked. Our protection here is that the backbone sites have proven in the past to be very strong supporters of speech. If that changes, we have worse problems.

...

(armm,

USENET would fall apart without the ability to cancel messages.

...

macpgp muscled off of archives

Huh? I don't care how many and whose signatures you put on something. When the guy who owns the disk wants it to go away, it will.

...

the loss of pax and now penet.)

Same as above. When the network provider decides to remove the connection for whatever reason, it's gone. I'm all in favor of technological solutions when appropriate. And I think that the use of ARMM on sci.* when there was no substantial reason was irresponsible and wrong. However, I don't think problems like volume abuse can be solved by purely technical means. When some bonehead starts putting megabytes of noise on sci.crypt, I want to be able to cancel his messages. Don't give me lines about user filtering; The best user agent in the world isn't going to make my net connection any bigger. I don't have the net bandwidth at my site to suck over megabytes of trash in order to ignore it until it expires. Marc

I thought the whole point of alt.* was for sites that didn't mind carrying a truly anarchic hierarchy... (We're losing alt.* here at Ames, it appears, along with soc. and talk.)