Double Irony! (was Re: [linux-elitists] ruben's stupid filter)

Bulent Murtezaoglu bm@acm.org
Mon Mar 25 12:30:15 PST 2002


>>>>> "KMP" == Karsten M Self <kmself@ix.netcom.com> writes:
[...]
    KMP> One suggestion from me (who happens to _like_ his
    KMP> auto-reporting tools, and is working to improve on their
    KMP> specificity, accuracy, and effectiveness), would be for those
    KMP> of the opposite camp to suggest alternatives.  This I haven't
    KMP> seen.

In my opinion it requires more than simple pattern matching for the
false positives to go down to negligible levels.  I suspect that would
entail some understanding of the semantic content.  Now, I understand 
pattern matching for detection does work well in practice, and has 
been used for years behind closed doors (Kernighan and Pike even
discuss the optimization of what they used in Bell Labs in one their 
books AFAIR).  I think it is fine for flagging, or sorting for
personal use or maybe bouncing e-mails if you want risk pissing people
off occasionally.  It is the reporting without human supervision aspect 
that bothers me (I agree with Rick Moen's sentiment on this).  

    KMP> I'm sorry, but I'm not going to spend five minutes (or two
    KMP> minutes, or even thirty seconds) manually tracking down the
    KMP> 40-odd spams I get daily.  That's more than three hours daily
    KMP> at the first estimate, a half hour at the low end.  [...]

I think the numbers are in your favour in this case.  _You_ don't have to
report each and every one of these things!  If it is spam, it is going
to a gazillion other people and if a tiny proportion of the recipients
report it "we" are set assuming reporting does any good at all.  So if
the premise is that you are required to spend the time, that premise
is flawed.

cheers,

BM



More information about the linux-elitists mailing list