To: declan@well.com, gnu
Subject: What to do about spam in general?  Use reader-oriented tools.
Date: Thu, 28 Feb 2002 15:23:03 -0800
From: John Gilmore 

It may be stupid for politicians to send spam, because it is
unpopular, but is it illegal?  In California, it is only illegal to
send unsolicited *advertisements*.  Political speech, including
requests for votes, would not seem to fall into that category.  And if
you have ever voted for a Democratic candidate, or talked with a
Democratic politician, you have had prior contact with the Democratic
party, so a message sent by them isn't unsolicited, by the legal
definition (in California Business & Professions code section 17538.4).

EFF has long advised against anti-spam laws, partly because state
after state makes the same kind of mistakes.  What is objectionable
about "spam" is that it is uninteresting to the recipient and sent in
bulk.  Whether it is an ad, a plea for charitable donations, a call
for political action, a request for votes, a patriotic declaration
during a national emergency, or an incomprehensible rant, people don't
want to see it in their mailbox.  That doesn't mean it should be a
crime or a tort to send it.  Furthermore, singling out particular
categories of messages BASED ON THEIR CONTENT is far more likely to be
unconstitutional, yet the laws invariably distinguish ads from other
bulk messages.  These laws also bring up harsh jurisdictional issues:
if California can legitimately impose rules on what everyone in the
country or world can send to Californians, then Bolivia or Palestine
or France can do the same to American email senders -- of all kinds of
email, not just bulk mail.  And if California can't impose local rules
on non-locals, then those who want to evade the local rules will
simply send their messages from another jurisdiction.

The part that virtually nobody understands is that spam isn't going to
go away.  It's like the drug war -- the more you ratchet up the
penalties against innocent people, the more innocent people are hurt
-- but there's still money in it for the malicious.  People are
clearly sending spam because it works for their purposes: even if it
pisses off 99.99% of recipients, they make back their costs and more,
from the tiny minority who DID wish to receive it.

We have built a communication system that lets anyone in the world
send information to anyone else in the world, arriving in seconds, at
any time, at an extremely low and falling cost.  THIS WAS NOT A
MISTAKE!  IT WAS NOT AN ACCIDENT!  The world collectively has spent
trillions of dollars and millions of person-years, over hundreds of
years, to build this system -- because it makes society vastly better
off than when communication was slow, expensive, regional, and
unreliable.  150 years ago, warriors killed civilians and each other
for months after the combatants had signed peace treaties, because the
news of peace had not reached them yet.  Even a decade ago, a friend
of mine died of a rare cancer, because the existing Japanese research
paper that showed how to treat it wasn't findable in time by his US
doctors.  These are just the tiny tip of an iceberg of problems and
inefficiencies that rapid cheap worldwide communication has solved.

Yet despite this immense value, it should not surprise us that most of
the things that others would want to say to us are not things that we
wish to hear -- just as we don't want to read the vast majority of the
books published, or the newspaper articles.  The solution is not to
demand that senders never initiate contact with recipients -- nor to
demand that senders have intimate knowledge of the preferences of
recipients.  Neither of these "solutions" produces a workable society;
they also violate freedom of association, speech, and privacy.

Nor is it a workable solution to impose liability for unwanted
communications on intermediaries such as ISPs or mail forwarders;
that's just "shooting the messenger", and encourages the
intermediaries to do even stupider things than the endpoints.  As an
example, dozens of my friends who happen to have addresses at
Earthlink simply cannot receive personal emails from me.  When they
send me emails, I can't reply.  Earthlink has decided that as an
intermediary it is going to censor the email its customers can read,
and has also decided that I am a suitable object of their censorship.
I have tried, and Earthlink customers have tried, to get this fixed;
Earthlink refuses.  Perhaps their Scientologist founder takes a
perverse pleasure in censoring an EFF co-founder's personal email.  At
any rate, I have no recourse to this blockage except to use some other
form of communication (like phones or postal mail) to persuade my
friends to stop being Earthlink customers, so I can swap email with
them again.  (I could trivially circumvent their blocking technology,
but that wouldn't solve the social problem -- which is that they feel
justified in deliberately censoring their customers' communications,
even against their customers' wishes.)

What makes the Internet so valuable to everyday people is that you can
reach anyone, on ANY email system, through it.  There were many email
systems before the Internet, but they didn't catch the broad public
interest.  If we continue the current process of anti-spam-driven
Balkanization (I can send email to Joe, and he can send to Nancy, but
I can't send to Nancy myself, because Nancy's ISP is filtering me), we
will destroy the value that we created when we linked all these
networks with a common email protocol.  We might as well go back to
having separate un-linked networks, like MCI Mail and Compuserve and
AOL and UUCP and BITNET and FidoNet.  You'd just have to become a
customer of that provider, and use its idiosyncratic interface, if you
want to send mail to its customers.  Remember that world?  If not,
you're lucky.  But your luck is running out, because the "solutions"
that people continue proposing and backing and implementing to "the
spam problem" will result in that.

THE REAL SOLUTION is to build and use mail-reading tools that learn
the reader's preferences, discarding or de-prioritizing mail that the
reader is unlikely to care about.  Every person can choose what
tool(s) they want to use to read their email.  This is a very close
relationship; I spend hours every day with my mail-reading software.
Most of the info is already there about what I prefer, based on what I
do with each message as I see it; the software just has to start
remembering and using it -- unlike the extremely uninformed
relationship between you and everyone who might want to send you
email.  If your software throws away an important message, you have
nobody to blame but yourself (or your vendor), and you have the
ability to fix the problem (perhaps by changing vendors).  If your
software shows you too many uninteresting messages, again you have
both the incentive and the ability to fix it yourself.  Your
preferences are kept locally, under your control, rather than having
your detailed "profile of interests" handed over to people like
advertisers, governments, or your (kind-hearted I'm sure) ISP.  In
such a system, knowledge of your interests will only be used to
benefit YOU, not those third parties.  You don't need to let anyone
else know that you have a fetish for shiny leather boots, or that you
secretly like People magazine.  This is a far better solution than
trying to impose the cost of "filtering out the mail that you wish not
to see" on your ISP, on every other ISP in the world, or on every
other sender in the world.

Also, such mail-reading tools provide far more useful capabilities
than merely filtering out spam.  Have you ever dropped out of a high-
volume mailing list because you really only cared about a fraction of
the messages in it?  A competent interest-based reader would let you
still read the fraction which you care about, while shielding you from
the intermingled irrelevant messages.

Being the sort of person who puts his money where his mouth is, I have
been funding a talented programmer to build such a mail reader, called
"grokmail".  It lets me assign interest rankings to each email
message, applies those rankings to each word in the message, and
combines these word rankings to find new incoming messages I'm likely
to be interested in.  It can keep multiple interest contexts (like
work-related messages, those from personal friends, relating to
hobbies, etc).  It will be free software, once it works well enough
that its author wouldn't spend all his time doing tech support rather
than development.  It is research -- barely a prototype now, and I
don't even know if the current design will turn out to ultimately
select the messages I really want to see.  But at least I'm trying to
solve the real problem -- while meanwhile fending off the rabid
anti-spammers who try to censor me or get me kicked off the Internet
for disagreeing with their approach.

This overload problem is not unique to email; it will come up with
instant messaging, with phone calls, with postal mail, and with any
other medium whose costs drop and whose reach improves.  I'm sure the
intelligence agencies have this problem in spades -- though as with
encryption, they aren't sharing their technology with the rest of
society, even though the benefits to the rest of society far outweigh
the minor problems caused by releasing the technology.

A hundred research labs and companies should also be experimenting
with various approaches to solving similar problems, so that we as a
society can continue to shrink the costs and increase the bandwidth of
our communication capabilities, without drowning ourselves in
irrelevant information.

Instead 99.9% of the energy goes into blacklisting technologies,
censorship laws, and bitching about what other people are "doing to
us" by sending us email.  Redirecting even a tenth of this effort
toward real reader-oriented solutions will invigorate our world
society, by enabling every person on earth to make more effective use
of our existing and upcoming communication tools.

	   John Gilmore