kickouts done the Cypherpunks way...

Hi,

Suppose Mr. X, owner of foobarpunks mailing list, wants to kick out Mr. Y,
for his obnoxious letters to the mailing list.

Mr. X, however, is concerned that Mr. Y would subscribe through some
proxy address and would continue replying to messages to foobarpunks.

It is assumed that the only person out of the whole universe, Mr. Y, 
cannot be trusted. The problem is that X does not know which of the 
subscribers is Mr. Y.

The question is, is there a technical way to disable Mr. Y from
reading the list, or detect which subscription address is a proxy for Y?

If we assume that, at the moment when Y was kicked out, he was not
subscribed through any other addresses, the solution becomes simple:
for any new subscription request we require a letter of recommendation 
from some other subscriber. Since other subscribers are presumed to
be trustworthy, their recommendations would be sufficient. It is 
actually being done in some of the mailing lists.

The problem becomes more complex when Mr. Y is already presumed to have
infiltrated the mailing list, possibly through several proxy addresses.

Is there any way to detect/find which if the subscriber is Y? One of
the simple-minded solutions is to _mutate_ mailing list messages
so that all readers get slightly different copies of mailing list messages
for each recipient. (Such mutations may include common misspellings,
inserting spaces, etc)

If the mailing list bot keeps track of what changes were made in 
messages to which individual, and if we assume that Mr. Y has to quote 
significant parts of messages he replies to, finally the variations
in messages may be reconciled with variations in quoted parts.

Mr. Y is not stupid, and may go as far as comparing letters, received
through different proxy addresses, in order to detect "variations"
and avoid quoting them.

The question is, is there a strategy of making variations and detecting
them in quotes to finally catch Mr. Y?

	- Igor.

The problem is that a nobody wanting to join the mailing list to learn
the subject matter of the list probably will not know anyone to sponsor
him.The second problem is that some isps go down occasionally and the mail
bounces back to the list which results in a subscription stop. How than
would one know to whom to go and ask for a recommendation. These  are
certainly things to consider .



On Sun, 24 Nov 1996, Igor Chudov @ home wrote:

> Date: Sun, 24 Nov 1996 19:34:09 -0600 (CST)
> From: "Igor Chudov @ home" <ichudov@algebra.com>
> To: cypherpunks@toad.com
> Subject: kickouts done the Cypherpunks way...
> 
> Hi,
> 
> Suppose Mr. X, owner of foobarpunks mailing list, wants to kick out Mr. Y,
> for his obnoxious letters to the mailing list.
> 
> Mr. X, however, is concerned that Mr. Y would subscribe through some
> proxy address and would continue replying to messages to foobarpunks.
> 
> It is assumed that the only person out of the whole universe, Mr. Y, 
> cannot be trusted. The problem is that X does not know which of the 
> subscribers is Mr. Y.
> 
> The question is, is there a technical way to disable Mr. Y from
> reading the list, or detect which subscription address is a proxy for Y?
> 
> If we assume that, at the moment when Y was kicked out, he was not
> subscribed through any other addresses, the solution becomes simple:
> for any new subscription request we require a letter of recommendation 
> from some other subscriber. Since other subscribers are presumed to
> be trustworthy, their recommendations would be sufficient. It is 
> actually being done in some of the mailing lists.
> 
> The problem becomes more complex when Mr. Y is already presumed to have
> infiltrated the mailing list, possibly through several proxy addresses.
> 
> Is there any way to detect/find which if the subscriber is Y? One of
> the simple-minded solutions is to _mutate_ mailing list messages
> so that all readers get slightly different copies of mailing list messages
> for each recipient. (Such mutations may include common misspellings,
> inserting spaces, etc)
> 
> If the mailing list bot keeps track of what changes were made in 
> messages to which individual, and if we assume that Mr. Y has to quote 
> significant parts of messages he replies to, finally the variations
> in messages may be reconciled with variations in quoted parts.
> 
> Mr. Y is not stupid, and may go as far as comparing letters, received
> through different proxy addresses, in order to detect "variations"
> and avoid quoting them.
> 
> The question is, is there a strategy of making variations and detecting
> them in quotes to finally catch Mr. Y?
> 
> 	- Igor.
> 
















       xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
       x   No success can compensate for failure in the home.  x
       x                                                       x
       xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

ichudov@algebra.com (Igor Chudov @ home) writes:

> Hi,
> 
> Suppose Mr. X, owner of foobarpunks mailing list, wants to kick out Mr. Y,
> for his obnoxious letters to the mailing list.
> 
> Mr. X, however, is concerned that Mr. Y would subscribe through some
> proxy address and would continue replying to messages to foobarpunks.
> 
> It is assumed that the only person out of the whole universe, Mr. Y, 
> cannot be trusted. The problem is that X does not know which of the 
> subscribers is Mr. Y.
> 
> The question is, is there a technical way to disable Mr. Y from
> reading the list, or detect which subscription address is a proxy for Y?

The answer is no.  Plenty of sites gate mailing lists to local
newsgroups, and allow open or relatively open NNTP access.  It's also
silly to assume every other person in the universe is trustworthy.

If Mr. Y sends lots of obnoxious mail to a mailing list or news group,
the proper thing to do is to put Mr. Y in your killfile and encourage
others to do so.  That way you don't get bothered by his annoying
messages, and if enough people follow suit, people stop responding to
Mr. Y's messages.  This can be even be extended to cover anonymous
posts using NoCeM-like systems.

If you try to boot Mr. Y off the mailing list using technical means,
several bad things will happen:  First of all you will fail, which
will give Mr. Y a great deal of satisfaction.  Second of all, you will
drive Mr. Y to start posting under different names, making him
considerably harder to killfile.  Third of all, you will double the
traffic on the mailing list by starting flamewars about whether this
failed booting attempt was ethical, legal, intelligent, homosexual,
scatological, or just plain useless.  Since at this point tons of
people will be replying to threads, a killfile becomes even harder to
manage.

So don't look for convoluted technical solutions to Mr. Y's
personality problems.  Just use a little basic common sense.  If you
don't like the way someone behaves on a mailing list, just ignore the
damn person.  Anything else is just going to make matters worse, as
recent history clearly demonstrates.

On Mon, 25 Nov 1996, lcs Mixmaster Remailer wrote:

> ichudov@algebra.com (Igor Chudov @ home) writes:
> 
> > Hi,
> > 
> > Suppose Mr. X, owner of foobarpunks mailing list, wants to kick out Mr. Y,
> > for his obnoxious letters to the mailing list.
> > 
> > Mr. X, however, is concerned that Mr. Y would subscribe through some
> > proxy address and would continue replying to messages to foobarpunks.
> > 
> > It is assumed that the only person out of the whole universe, Mr. Y, 
> > cannot be trusted. The problem is that X does not know which of the 
> > subscribers is Mr. Y.
> > 
> > The question is, is there a technical way to disable Mr. Y from
> > reading the list, or detect which subscription address is a proxy for Y?
> 
> The answer is no.  Plenty of sites gate mailing lists to local
> newsgroups, and allow open or relatively open NNTP access.  It's also
> silly to assume every other person in the universe is trustworthy.

Process of elimination.
If the document that goes to a newsgroup is document #5, then you can
concentrate on the path after #5.  Which newsgroup is it?  Easily
determined by selectively seeding documents to different newsgroups.  You
get the idea.  Message pools complicate the process, but are not
impossible to deal with, particularly when output to the newsgroup can be
controled, as it can here, either by killing the feed the the group, or to
the party posting there.

This all of course begs the question as to whether this is even a good
idea, you address it below:

> 
> If Mr. Y sends lots of obnoxious mail to a mailing list or news group,
> the proper thing to do is to put Mr. Y in your killfile and encourage
> others to do so.  That way you don't get bothered by his annoying
> messages, and if enough people follow suit, people stop responding to
> Mr. Y's messages.  This can be even be extended to cover anonymous
> posts using NoCeM-like systems.

Agreed.  But not because it is technically impossible.

> If you try to boot Mr. Y off the mailing list using technical means,
> several bad things will happen:  First of all you will fail, which
> will give Mr. Y a great deal of satisfaction.  Second of all, you will
> drive Mr. Y to start posting under different names, making him
> considerably harder to killfile.

Not really.  Simply continue to seed and watch new subscriptions to the
list.  That narrows down the leak quite well.

Same thing as winding up agent nets that use dead-drops.  Identigy one
step at a time.

> Third of all, you will double the
> traffic on the mailing list by starting flamewars about whether this
> failed booting attempt was ethical, legal, intelligent, homosexual,
> scatological, or just plain useless.  Since at this point tons of
> people will be replying to threads, a killfile becomes even harder to
> manage.

Agreed.

> So don't look for convoluted technical solutions to Mr. Y's
> personality problems.  Just use a little basic common sense.  If you
> don't like the way someone behaves on a mailing list, just ignore the
> damn person.  Anything else is just going to make matters worse, as
> recent history clearly demonstrates.

Mostly agreed.


--
Forward complaints to : European Association of Envelope Manufactures
Finger for Public Key   Gutenbergstrasse 21;Postfach;CH-3001;Bern
Vote Monarchist         Switzerland

On Sun, 24 Nov 1996, Igor Chudov @ home wrote:

> The question is, is there a strategy of making variations and detecting
> them in quotes to finally catch Mr. Y?

It's not a new concept if thats what you mean.  Has many names.  "Canary
trapping, imposed distribution attributation" and others.

Generally either a single non-spelling error is introduced (generally with
punctuation), or several errors are introduced in different patterns
throughout the document.  Word order is another common variation to use.  
At the most basic level, a script could easily introduce e.g., a
semicolon in place of a comma once in a document, and in a different
place for each document. The "phantom serial number" of a given document
is obtained by counting the commas encountered before a semicolon is
found.  

If a correlation between serial number and original recipiant is
maintained, the first tier of the avenue by which the document is escaping
is easily identified.

Of course this method assumes that the documents are distributed onward
without alteration or summarizing.

More complicated variations can be introduced as is needed.
(Paragraph/subject order, names of participants, etc.)

Coderpunks might have a harder time.  For obvious reasons technical
writings are much more sensitive to even minor alterations.

> 	- Igor.
> 

--
Forward complaints to : European Association of Envelope Manufactures
Finger for Public Key   Gutenbergstrasse 21;Postfach;CH-3001;Bern
Vote Monarchist         Switzerland