On Thu, Jun 30, 2005, Matan Ziv-Av wrote about "Re: A new venture - preventing 
spam":
> I see (also by Nadav's reply) that I needed to elaborate a bit more. 
> Your idea might make the spam situation a lot better. Your idea might 
> make a lot of money (if that is your goal, which was not clear from your 
> message). But it won't "solve the spam problem". Since there isn't even 
> an agreed upon definition of spam, there is no way anyone but the 
> mailbox owner can tell with 100% certainty whether a given message is 
> spam or not.

Like I said, theoretically speaking, you are right. That's why anti-spam
work will not get you any Ph.D. in pure mathematics.

But in practice, here is the definition I use: The "spam problem" is the
problem that causes 90% of the email I get to be mail that I have no reason
to read. I.e., it was not personally written to me, but rather written to
many people for some economic/political/whatever reason. If that "90%"
figure became "9%" then this would no longer be a "problem" - perhaps just
a minor annoyance. The "just hit delete" solution works if you get 100
emails a day and 10 of them are spam, but what if out of the 100 emails,
90 are spam? Speaking from experience, even a human makes mistakes once in
a while and presses "delete" on the wrong message.

So "solving the problem" doesn't necessitate a 100% solution. A 99%, or
maybe even a 90% solution, is already a real solution, because it makes
your email readable again.

Of course a *good* solution (as opposed to a crappy one) will need to be
very sensitive to the problem of false-positives - genuine email mistakenly
thought to be spam. When I wrote "90%" above I meant a 10% false-negative
rate, and a tiny false-positive rate. Again, this isn't rocket science.
It is doable, and if it weren't, I would not have been able to read my
email in the last 5 years.

And to re-iterate the point I made above: a computerized anti-spam system
isn't perfect, whether it is based on filtering (like you appear to assume)
or on completely different techniques (authentication, stamps, confirmation
emails or whatever) it will make some mistakes, lose some good emails and
hand you some spams. But, so what? The same thing will happen if you try to
read all your mail on your own, casually hitting "delete" on a non-spam
by mistake because for a second you didn't recognize the name. I know,
because it happened to me when someone asked me to read his (unfiltered)
mail for a week, and I missed a very important message which I assumed was
spam. So the important thing is that an automated anti-spam system CAN be
better than a human doing it. And I think there is no argument that we
need to do something, unless we want to read advice on how to grow various
organs all day :-)

-- 
Nadav Har'El                        |     Thursday, Jun 30 2005, 23 Sivan 5765
[EMAIL PROTECTED]             |-----------------------------------------
Phone +972-523-790466, ICQ 13349191 |"I'll doublecross that bridge when I come
http://nadav.harel.org.il           |to it" (a politician about the future)

=================================================================
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]

Reply via email to