From: Adam Back <aba@dcs.ex.ac.uk>
Date: Sun, 19 Jan 1997 15:50:56 -0800 (PST)
To: dlv@bwalk.dm.com
Subject: Dr Vulis's crypto experiment (Re: IMDMP: SOURCE CODE RELEASE ANNOUNCEMENT)
In-Reply-To: <51Rq1D26w165w@bwalk.dm.com>
Message-ID: <199701162142.VAA06080@server.test.net>
MIME-Version: 1.0
Content-Type: text/plain



It's occured to me recently that Dimitri's posts are not being read
correctly, it is not the content which is the point, nor the
superficial lack of crypto relevance.

Dimitri is an intelligent guy, and has a high level of crypto
expertise (he has a PhD on a cryptography topic).  It is my belief
that his posts can only be understood in a "meta" sense -- he is
engaged in a highly complex cryptographic experiment.  People who read
and respond to his individual posts are the unwitting subjects in his
experiments.  His posts and the responses to them are actually the
data-set for a thorough cryptanalysis of mailing list threats.  His
current topic under investigation is Denial of Service (DoS) attacks
on mailing lists.

Denial of service on mailing lists is a complex business, and requires
expert human input to be done properly.

Dimitri has systematically explored these types of posting behaviour:

1. Posting only crypto relevant material.
2. Interspersing crypto relevant posts with hand personalised flame bait.
3. Interspersing crypto relevant posts with bot generated flame bait.
4. Interspersing crypto relevant posts with news stories.
5. Posting only non crypto relevant material.

The alert reader will recall these phases of posting style (currently
we are in sub experiment 5, the other phases have occured over a
protracted period of intensive experimentation, and some newer readers
may have missed earlier phases).

There were other experiments which may or may not have been part of
Dimitri's series of DoS experiments:

6. Subscribing the list to itself (testing list resilience to recursion)

7. Forging posts to carefully selected newsgroups with
"cypherpunks@toad.com" as the sender (this indirectly adds user
"cypherpunks@toad.com" to many direct marketing lists as direct
marketers make use of email addresses scanned from newsgroups).

8. The "UNSCRIVE" and other spelling variations of "unsubscribe" epidemic,
and ensuing instructions and discussion

9. Subscribing the list to other lists

10. Bot generated flame bait posted anonymously (with ascii art)

>From the post I am following up to the reader will observe an oblique
reference to the transition from phase 4 of the experiments to the
current phase, phase 5 (the reader will also note references to experimental
results c and d described below):

Dimitri Vulis <dlv@bwalk.dm.com> writes:
> One memorable "censorship" incident occurred when the lying cocksucker
> John Gilmore (spit) forcibly unsubscribed me from this list because he
> didn't like the contents of my submissions - or did you forget already?
> Look up Declan's disgraceful writeup on Netly News archives.
> 
> At that time most valuable contributors to this list (such as myself)
> resolved not to contribute crypto-relevant content to a censored forum.
  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

More widely read cypherpunks will know that Dimitri has performed
similar experiments in other fora -- his systematic experiments in
numerous USENET newsgroups resulted in the his receiving the
prestigious KOTM ("Kook Of The Month") award.

His main experimental results to date with the cypherpunks mailing
list series of experiments have been:

a) many cypherpunks publically announcing kill filing him

b) numerous cypherpunks suggesting censoring him (a particularly
interesting result considering the libertarian leanings of many on
this list)

c) the outstanding experimental result of being forcibly unsubscribed
from the list, and of being barred from resubscribing by John Gilmore.
(creator of alt.* USENET newsgroup hierarchy, and well know freespeech
advocate)

d) another interesting, incidental experimental result was provided by
Declan McCullagh in his Netly News piece in prematurely, and entirely
unwittingly, publishing some of Dimitri's expermiental data-set.

e) the main experimental result: the list shortly moving to a
moderated form, seemingly at the request of Sandy Sandfort, with
agreement from John Gilmore.

These experimental results are quite significant, when taken in the
context of the anti-censorship, libertarian, pro-freespeech
environment of the cypherpunks mailing list.  Dimitri should be
congratulated on his outstanding work.

I await with interest the last phases of Dimitri's experiment, when
the cypherpunks list becomes a moderated forum.  My suggestions for
interesting experiments during the moderated phase are:

1. Testing the limits of Dale Thorn's anti-censorship sentiments (for
those who don't read Dale, he is subscribed to "cypherpunks-unedited"
in preparation for the moderation).

2. Testing Sandy Sandforts rejection criteria.

3. Testing Sandy Sandforts rejection rate for long crypto relevant
posts interspersed with irrelevant flame bait.

4. Testing Sandy Sandforts rejection rate for posts with flame bait .sigs

I hope Dimitri's selfless efforts in furthering understanding of DoS
attacks on mailing lists is properly acknowledged when he publishes
his findings on completion of his experiments.  I also hope that
Dimitri will document his recommendations for mailing list
configuration and management in light of his experiments.

Adam
--
print pack"C*",split/\D+/,`echo "16iII*o\U@{$/=$z;[(pop,pop,unpack"H*",<>
)]}\EsMsKsN0[lN*1lK[d2%Sa2/d0<X+d*lMLa^*lN%0]dsXx++lMlN/dsM0<J]dsJxp"|dc`