[Scspamcop] Re: Why no Chinese or Japanese blocking?

[Peter Pearson]

On Sat, 11 Apr 2009 00:32:08 +0000 (UTC), Michael R N Dolbear wrote:
> Peter Pearson <ppearson at nowhere.invalid> wrote 
>
>> It's great being able to steer Russian-language messages
>> directly into my Held Mail folder.  Why isn't there a 
>> similar option for Chinese, Japanese, and Korean?
[snip]
> A very recent enhancement (27Sept08) to the SpamCop Mail service.
>
> It blocks email which have "koi8-r" in the headers.
>
> I used to have lots of these but when the flow of spam resumed after
> the 12Nov to 22Jan drop I saw hardly any (45 in Sept, 111 in Oct, 10 in
> Feb).
>
> If you would care to make an exactly specified proposal and check for
> false positives using Search and/or Filter and reports some numbers
> then JT might consider it.

I'm not sure exactly what search and/or filter technique
you're suggesting, but I the string "gb2312" does not appear
(independent of case) in the subject line of any of the 32,820
subject lines in my non-spam email archives.  Is that the
sort of thing I need to say?

If I learned more about email standards, I could probably
distinguish between a GB2312 at the beginning that means
"This subject line will be displayed in Chinese characters"
and a GB2312 that happens to be part of the subject, as in
"Let's discuss filtering on GB2312."

Filtering out emails whose subject-line character sets are
for languages that I can't read seems extremely safe.

> Do you have the China country blocklist set ?

Yes.  (Also SpamCop Blacklist, Spamhaus Blacklist, Nigeria,
Composite Blocking List, and Spamhaus XBL.  Unchecked are
South Korea, Argentina, Brazil [which I'm thinking of
checking], and Spamhaus PBL.)

> What SpamAssassin (SA) setting ?

I set SpamAssassin Limit to 6.

> I have just checked and with 3000+ spams a month and about 1% leakers I
> have had no Chinese, Japanese, and Korean leakers in at least the last
> 6 months . Apparently for my input SpamAssassin (SA) stops them all.

Interesting.  My total spam load is about half that, but
just since the beginning of April, six spam messages with
Chinese subject lines (GB2312) have slipped through.  I
concede that's not enough to warrant a lot of whining, but I
thought it might be pretty easy to detect these.

-- 
To email me, substitute nowhere->spamcop, invalid->net.