Re: You Might Be An Anti-Spam Kook If ...

On Mon, 08 Sep 2003 18:38:03 +0800, Shelby Moore said:

> Imagine if we engineered all things in real life by first expending effort
> publishing all the ways not to engineer them.
Umm.. We don't.  The first documented spam was sent over 25 years ago,
and you tuned in late.

How to design a bridge:

1) Build one out of dirt.  Realize that's a dam not a bridge, when it washes
away.  Make a note of it.

2) Build one out of rocks.  Find out the hard way that it works better if you
use mortar too.  Make a note of it.

3) Build one out of rocks and mortar.  Find out the hard way that setting the
piers on mud causes washouts.  Make a note of it.

4) Build one out of rocks and mortar, and set the piers on good solid rock.
Find out the hard way (multiple times) about how thick it has to be to support
a given load.  Make a note of it.

5) Discover the hard way about resonance effects.  Post a sign for the
marching troops and make a note of it.

6) Bundle up your notes and publish the Civil Engineering Handbook, First 
Edition.

7) Discover that wood makes OK bridges too, make lots of notes about strength,
insect prevention, and the like.  Publish the Second Edition.

8) Discover metal.  Make notes about fatigue and rust.  Publish Third Edition.

We can't help you if you think that we sat down and wrote the Third Edition
before doing anything else.  And we *really* can't help you if we hand you a
copy of the Third Edition, and you insist on trying to build a bridge in a way
that we pointed out on page 37 *just won't work*.  

On Mon, 08 Sep 2003 19:25:22 +0800, Shelby Moore said:
> Amazing that Vernon says working on spam is "Kook"y, yet that is exactly what 
> he does (DCC).
No.. What he *said* was "totally disregarding the things others have already 
tried is kooky".
Which line item from his note is the DCC in violation of?  (In particular, note 
that Vernon
has never claimed that the DCC is either perfect or a complete solution - it's 
a work-in-progress
of one tool for one aspect of the problem).

> 2. Vernon wrote, "...it is impossible, because no two people (or at least
> organizations) think the same streams of bulk mail are solicited and
> unsolicited.".  I can easily find 2 streams that 2 people/orgs can agree are
> solicited and unsolicited.  If you give me 2 streams and 2 people/org at
> random, then I can not guarantee every time (but eventually and never
Right. You can't guarantee *every time* for 2 people at random.  That's his 
point exactly.  

Any scheme that requires any 2 random people to come up with the same answer on
this judgment call *will* break with either a false positive or false negative 
for each
object in the stream there's a disagreement on.

> 4. "A fundamental problem is that spam is unsolicited bulk mail, and not IP
> addresses...no mater how highly coorelated with spam.".  The first part is
> true, the second part is not because we don't live in 3 dimensions.  I will 
> not
> tell you why beyond that, but I know why.
Actually, 3 dimensions has absolutely nothing to do with it.  And the fact that 
you
think it does indicates that you really don't understand the difference there.

The real problem is that there exists a non-empty subset of the IP address
space for which the predicate '((sends spam) XOR (sends non-spam)) ' is zero,
even for an actual SMTP source.

pgp4K1cYcuVQs.pgp
Description: PGP signature