«

VanguardLH wrote:
> "Mike Easter"

>> Conceivability is relative.

> And I can conceive that any e-mail that *I* receive that has nothing
> but numbers is always spam (and has been, so far).

The original premise was a 4 number string but your solution was instead
to filter anything without a vowel in it.

Those are quite different programmatically and problematically. They
both have severe false positive potential problems. I don't know
exactly which is worse, the original idea of the OP or your solution. I
think it might be your solution.

I'm pretty sure that a regex for a subject of a 4 digit decimal number
would have less bad effects than a regex for a subject without a vowel;
I also doubt that OE could make a message rule for either. Clearly it
has no regex power, and I doubt if the option for any of the one
character strings of a vowel would work; but you are welcome to try.

I trust your integrity to report back accurately if you should bother to
try the experiment.

--
Mike Easter

Mike Easter wrote:
> Blinky the Shark wrote:
>> Mike Easter wrote:
>
>> I'm very careful with my own filtering.
>
>>> My current filters are working excellently. I have whitelists.
>
>> Then no-vowels eater wouldn't be a problem.
>
> Do you have a 'no-vowels in the subject' filter? I'm presuming "No"
> -- ergo see below.

No, because I have not seen a single Subject header that's all
non-alphabetic. I don't make filters unless there's a need FOR ME, see.
Similarly, while I have a fly swatter in my apartment that can be used
on insects, I don't swing it at thin air -- I save it for when there's
an insect to swat. Now apply that to spam: if that kind of header is
bothering *others*, I recommend that *they* use the filter.

>>> When I accidentally get a spam into the Inbox, I don't start making
>>> up bad rules because the spam had numbers in the subject.
>
>> Neither do I.
>
> If you don't have a no-vowels rule, then you must not think it is a
> good rule, in which case I'll presume that we agree on the reciprocal,
> that it is a bad rule, absent any neutral position for the purposes of
> voting.

Wrong premise followed by wrong logic. Oh for two. *I* don't have one
because *I* don't see that kind of header. Why would I make a rule to
cover something I don't see. Dood, you're a lot brighter and more
helpful than you're sounding here. If A has a roach infestation, he
should exterminate them, even if B isn't similarly infected.


--
Blinky RLU 297263
Killing all posts from Google Groups
The Usenet Improvement Project - http://improve-usenet.org

VanguardLH wrote:
> "Mike Easter" <> wrote in message
> news:...
>> VanguardLH wrote:
>>> "Rules should not be made which can 'conceivably' cause a false
>>> positive"
>>>
>>> Yeah, name just one rule that would never conceivably cause a false
>>> positive.
>>
>> Conceivability is relative. You can't conceive of a goodmail with
>> numbers in the subject. I can.
>>
>> Is that conceivable or not?
>
> And I can conceive that any e-mail that *I* receive that has nothing
> but numbers is always spam (and has been, so far).

And *I* can conceive that because of my other filters I don't personally
*get* any all-non-alpha spam, so while I don't personally need the
no-vowels filter, I can conceive of others wanting to use it. Mike,
surprisingly for him, can't conceive of that, either.

Here's what remains. For those who need it: use it. For those who
don't need it, don't use it. This, as I think *you* know, isn't exactly
rocket surgery.


--
Blinky RLU 297263
Killing all posts from Google Groups
The Usenet Improvement Project - http://improve-usenet.org

Blinky the Shark wrote:
> Mike Easter wrote:

>> Do you have a 'no-vowels in the subject' filter? I'm presuming "No"
>> -- ergo see below.
>
> No, because I have not seen a single Subject header that's all
> non-alphabetic. I don't make filters unless there's a need FOR ME,
> see.

Let us try to stay with the main and original theme. The OP got /a/
spam with a 4 decimal digit number in the subject, and which was
'riddled' with all kind of other unknown features in the header and the
body. Instead of evaluating the entire spam, his concept was to make a
'rule' (presumably with his OE) somehow that would sort such a subject.

I contend that that is the wrong strategy. Misguided and based on
inadequate information. That a better strategy is to consider and comb
the entire spam headers and body with a powerful tool like SpamPal; and
further I gave a specific example of how to use a better filter for a
similar numerical subject spam.

You are defending the idea of making such a rule when I have already
demonstrated with the complete and precise illustrative example that
there is a better rule strategy..

There is no competing example of the OP's spam. The original 'example'
was a weakly descriptive allusion to a 4 digit subject. That does not
qualify as a proper example for the purposes of discussing how to
configure a spamfilter for the example of "How do you filter a spam like
this?".

Traipsing into the entomological resistance endeavors of fly-swatting
and roach cohabitating doesn't help here.

--
Mike Easter

"Blinky the Shark" wrote ...
>
> And *I* can conceive that because of my other filters I don't
> personally
> *get* any all-non-alpha spam, so while I don't personally need the
> no-vowels filter, I can conceive of others wanting to use it. Mike,
> surprisingly for him, can't conceive of that, either.
>
> Here's what remains. For those who need it: use it. For those who
> don't need it, don't use it. This, as I think *you* know, isn't
> exactly
> rocket surgery.


The no-vowel rule is nothing but a reduced rule used to detect a blank
Subject. Many users want to filter out any mails with a blank
Subject. Could a non-spamming sender accidentally forget to include a
non-blank string in the Subject? Well, obviously yes. Have I ever
gotten any? No, usually because the sender's e-mail client warns the
sender about a blank Subject. I do, however, get spams with no
Subject hoping you'll go read the body of their crapmail.

Outlook and Outlook Express don't have rules for checking that header
has a null string. Some e-mail clients do. Outlook and OE don't.
However, you can test for the absence of any character, so the rule
checking for a blank Subject header looks like:

Apply this rule after the message arrives
Where the Subject line does not contain 'a' .. or 'z' or '0' .. or '9'
(pick a desired action)
and Stop processing more rules

Because neither e-mail client has the range specifier shown above, it
is just shorthand used here for including every character from 'a' to
'z' and every number from '0' to '9'. The non-alphanumeric characters
aren't included but (just as has been argued above) no one expects
legit e-mails to only have periods, commas, pound signs, exclamation
marks, or solely non-alphanumeric characters in the Subject header.
How to test on a blank Subject has been requested MANY times in the
Outlook [Express] newsgroups.

If you don't correspond with anyone that writes in Russian or Hebrew
which can have words with no vowels (there may be other languages,
too), the "Blank Subject" rule can be shortened from 'a' to 'z' to
just looking for the vowels to indicate that actual *words* exist in
the Subject header. So the rule gets shortened to:

Apply this rule after the message arrives
Where the Subject line does not contain 'a' or 'e' or 'i' or 'o' or
'u' or 'y' or '0' .. or '9'
(pick a desired action)
and Stop processing more rules

If you don't ever expect anyone to send you a non-spam e-mail with
just numbers in the Subject (Mike is the exception) then you can
exclude checking for all the numeric digits, so the "Blank Subject"
rule simply becomes:

Apply this rule after the message arrives
Where the Subject line does not contain 'a' or 'e' or 'i' or 'o' or
'u' or 'y'
(pick a desired action)
and Stop processing more rules

Whether you omit 'y' is your choice. There is, for example, the word
"syzygy" but do you really expect the word to appear without other
words in the Subject, too? Obviously the shorter you make the rule
the less time must be spent wading through all the OR conditions.
Unlike Mike, I have not experienced a lack of reliability in the
execution of the rules simply because the "words" in the OR conditions
within the same rule are shorter than 3 characters in length.

Rules should be defined based on your own history of e-mails, not on
someone else's unless they fit for you needs, too. Mike happens to
send e-mails with non-spamming friends where the Subject has only
numbers. Okay, so use the vowels+numbers rule. If you want to check
only that the Subject is blank using the longer rule then use the
'a..z'+'0..9' rule. If you are so weird as to actually expect to
receive e-mails where the Subject has only non-alphanumeric
characters, like "&&.^!$,$_+%" then put every character in the really
long conditional rule. Personally I find the vowels-only rule
sufficient.

Mike Easter wrote:
> Blinky the Shark wrote:
>> Mike Easter wrote:
>
>>> Do you have a 'no-vowels in the subject' filter? I'm presuming "No"
>>> -- ergo see below.
>>
>> No, because I have not seen a single Subject header that's all
>> non-alphabetic. I don't make filters unless there's a need FOR ME,
>> see.
>
> Let us try to stay with the main and original theme. The OP got /a/
> spam with a 4 decimal digit number in the subject, and which was
> 'riddled' with all kind of other unknown features in the header and the
> body. Instead of evaluating the entire spam, his concept was to make a
> 'rule' (presumably with his OE) somehow that would sort such a subject.
>
> I contend that that is the wrong strategy. Misguided and based on
> inadequate information. That a better strategy is to consider and comb
> the entire spam headers and body with a powerful tool like SpamPal; and
> further I gave a specific example of how to use a better filter for a
> similar numerical subject spam.
>
> You are defending the idea of making such a rule when I have already
> demonstrated with the complete and precise illustrative example that
> there is a better rule strategy..

And I and others have maintained that a simple client rule, applied to
the general odds of receiving useful mail with no alpha characters in
the Subject header is good enough, even if your niche requirements for
personally receiving numeric-only Subject headers. Your niche needs are
not mainstream. Do your thing. Others who don't speak only in numerals
can do their thing. Just as well for them as yours is for you.

> There is no competing example of the OP's spam. The original 'example'
> was a weakly descriptive allusion to a 4 digit subject. That does not
> qualify as a proper example for the purposes of discussing how to
> configure a spamfilter for the example of "How do you filter a spam like
> this?".
>
> Traipsing into the entomological resistance endeavors of fly-swatting
> and roach cohabitating doesn't help here.

When one is up against a blind man, sometimes one tries to use analogies
that his limited senses might understand.

I'm out of here on this topic.

Have fun slootwegging on your own.

--
Blinky RLU 297263
Killing all posts from Google Groups
The Usenet Improvement Project - http://improve-usenet.org

On Sun, 30 Sep 2007 22:14:13 -0500, "VanguardLH"
<> wrote:

>"Mike Easter" <> wrote in message
>news:...
>> VanguardLH wrote:
>>> "Rules should not be made which can 'conceivably' cause a false
>>> positive"
>>>
>>> Yeah, name just one rule that would never conceivably cause a false
>>> positive.
>>
>> Conceivability is relative. You can't conceive of a goodmail with
>> numbers in the subject. I can.
>>
>> Is that conceivable or not?
>
>
>And I can conceive that any e-mail that *I* receive that has nothing
>but numbers is always spam (and has been, so far).

I get support emails that just contain the users number so a "numbers
only" filter would not be safe for me. That's the whole point about
email spam. The final filter must be done by individuals, not
generalized.

Steve

--
Stephen Wolstenholme Neural Planner Software Ltd

EasyNN-plus. The easy way to build neural networks.

http://www.easynn.com

Steve Wolstenholme wrote:
> On Sun, 30 Sep 2007 22:14:13 -0500, "VanguardLH"
><> wrote:
>
>>"Mike Easter" <> wrote in message
>>news:...
>>> VanguardLH wrote:
>>>> "Rules should not be made which can 'conceivably' cause a false
>>>> positive"
>>>>
>>>> Yeah, name just one rule that would never conceivably cause a false
>>>> positive.
>>>
>>> Conceivability is relative. You can't conceive of a goodmail with
>>> numbers in the subject. I can.
>>>
>>> Is that conceivable or not?
>>
>>
>>And I can conceive that any e-mail that *I* receive that has nothing
>>but numbers is always spam (and has been, so far).
>
> I get support emails that just contain the users number so a "numbers
> only" filter would not be safe for me. That's the whole point about
> email spam. The final filter must be done by individuals, not
> generalized.

Are those emails so varied that they can't be easily whitelisted by
*some* characteristic?


--
Blinky RLU 297263
Killing all posts from Google Groups
The Usenet Improvement Project - http://improve-usenet.org

On 1 Oct 2007 08:54:59 GMT, Blinky the Shark <>
wrote:

>Steve Wolstenholme wrote:
>> I get support emails that just contain the users number so a "numbers
>> only" filter would not be safe for me. That's the whole point about
>> email spam. The final filter must be done by individuals, not
>> generalized.
>
>Are those emails so varied that they can't be easily whitelisted by
>*some* characteristic?

Not really. I have thousands of users and everyone of them has a
different 9 digit number. Any user could send a support message
because it's built in to the application. The latest version adds a
specific keyword that can be used but I need to allow for older
versions.

Steve

--
Stephen Wolstenholme Neural Planner Software Ltd

EasyNN-plus. The easy way to build neural networks.

http://www.easynn.com

Steve Wolstenholme wrote:
> On 1 Oct 2007 08:54:59 GMT, Blinky the Shark <>
> wrote:
>
>>Steve Wolstenholme wrote:
>>> I get support emails that just contain the users number so a
>>> "numbers only" filter would not be safe for me. That's the whole
>>> point about email spam. The final filter must be done by
>>> individuals, not generalized.
>>
>>Are those emails so varied that they can't be easily whitelisted by
>>*some* characteristic?
>
> Not really. I have thousands of users and everyone of them has a
> different 9 digit number. Any user could send a support message
> because it's built in to the application. The latest version adds a
> specific keyword that can be used but I need to allow for older
> versions.

Well, since the two patterns are that distinct, you could whitelist on a
nine-digit number only (old situation) and on a nine-digit number and
that keyword (new version).


--
Blinky RLU 297263
Killing all posts from Google Groups
The Usenet Improvement Project - http://improve-usenet.org