[Asrg] (Position): Successful anti-spam techniques must avoid software patents

[David Wheeler <[e-mail]>]

Hi - I didn't see a clear position on the ASRG main page
(http://www.irtf.org/charters/asrg.html) on software patents.
I believe any useful ASRG result MUST avoid software patents and/or identify
reasons why any patent claims are, almost certainly, invalid.
I believe the ASRG work should be based on this position.

At the very least, any ASRG work should avoid any
software patent landmines if possible, include a patent search to
identify them, and warn potential developers of the results of the search.
Any organizations need to be REQUIRED to explicitly reveal any patents
that might apply (including those that are pending), or everyone will
risk being exploited (I believe RAMBUS is an example of this).

Rationale: For any anti-spam technique to be widely effective,
it needs to be widely deployed.  Software patents will significantly
constrain such deployment.  In particular, most email is handled
at least part of the time by open source software, and open source
software cannot implement patented concepts in the countries where
the patents are valid.

For example, the mail server survey described here:
  http://cr.yp.to/surveys/smtpsoftware6.txt
found that at the time the survey was made,
sendmail was still the most common SMTP server (42%),
followed by Windows Microsoft Exchange (18%),
Unix qmail (17%), Windows Ipswitch IMail (6%),
Unix smap (2%), UNIX Postfix (formerly VMailer, 2%)
and Unix Exim (1%).  Sendmail, qmail, smap, postfix, and exim
could not practically implement a patented technique, so
64% of deployed systems couldn't be upgraded to a newer version
and implement the technique (if the numbers are the same).
And that presumes that the proprietary systems would all jump
to implement a patented technique -- which is rather unlikely.
The key here isn't the exact numbers, the issue is that
if you allow patented techniques, you may end up with undeployable
ideas.

There are also many open source mail clients, including the
one in Mozilla (repackaged as Netscape Mail), and open source
mailing list software (including the one used here!).
In short, email has a significant presence of open source software,
and it would be extremely unwise for this group to set standards
that cannot be met by open source software and would require
licensing fees to be paid by some

So-called "Reasonable and Non-Discriminatory" (RAND) licenses
aren't enough in this case.  In all cases, RAND licenses discriminate
against those who don't own the patent.  Besides, to be used
in open source software it has to be completely royalty-free,
at least.  Even then the popular license used by open source software
(the GPL) will find this unacceptable unless the idea can be used
for any purpose (not just to implement the standard).

I believe this concern is quite consistent with the ASRG charter:
  "ASRG will not pursue research into legal issues of spam, other
  than the extent to which these issues affect, support, or
  constrain the technology."
Software patents constrain the technology, so identifying and
countering any such legal problems is critical for this group.

It's true that there are patent claims for challenge-response
systems, but that primarily shows how poorly the patent offices
screen patent applications.  Mailblocks of Los Altos, California,
has been suing people based on patents it owns,
6,199,102 (filed in 1997) and 6,112,227 (filed in 1998)
However, it's clear that there's extensive prior art; for example,
Cynthia Dwork and Moni Naor of IBM described in 1992
a challenge-response system in which the sender
would be asked to process a particular solution before
the receiver would accept the email
(http://research.microsoft.com/research/sv/PennyBlack/junk1.pdf)
Thus, it's reasonable for the ASRG to work on challenge-response,
since there's good reason to believe that this nonsense patent
will be overturned, and indeed the ASRG could work to collect
the data on why it should be overturned.

If the ASRG wants anti-spam technology to be widely deployed, it
needs to make sure that its solutions don't have any serious
legal impediments.  And that includes software patents.

Thank you for listening.

--- David A. Wheeler
(Speaking for myself)


_______________________________________________
Asrg mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/asrg