Matt Kettler wrote: > Dude, razor is NOT a bayes subsystem. > Dude, I didn't say it was Bayesian.
Perhaps not, but claiming people should razor-revoke nonspam messages as a general whole implies a gross misunderstanding of razor. That model is only aplicable to bayesian or other token analysis systems.
> Razor is a database of hashes of known spam messages. Period.
Wrong.
Obviously, you've never looked under the hood of razor and have no idea how it really works.
No, I have. I understand how razor works extraordinarily well, and from your message you understand it well too. However, I feel you're broadly over-generalizing the use of razor-revoke.
There are cases where revoking nonspam makes sense, but doing it indiscriminately, as you suggest, is foolhardy and wasteful.
I will not retract my statement that razor is a database of hashes of known spam messages.
The on-line server which you can query stores only message hashes and their associated cf scores.
The fact that the entire message is reported to the server is irrelevant, as this is only used to recalculate hashes in the event of a change of the e4 hash selection criteria.
The fact that the server stores a CF value changes the details of operation, but doesn't change any of my fundamental points.
The fact that razor treats each mime segment as a separate part doesn't change the fact that it's a message hash system. Yes, it does change things slightly but it doesn't change any of my fundamental points... Let me rephrase myself to be more technically correct:
Old text:
Razor recognizes only the exact message you report. If that message never appears again in the world, reporting it is a waste.
More detailed version:
Razor is strictly a mime-part hash system and only recognizes the exact mime-parts of the message you report. If all the mime-parts of the message you send to razor-revoke are unique and do not ever appear in any other messages, then your submission to razor-revoke is a waste of time and bandwidth.
However, the minor differences between the two statements are IMO bordering on pedantic.
For example, if you were to submit this very message to razor-revoke, you'd be wasting time and bandwidth.
This message is single-part text/plain. It's contents are unique and are not likely to be attached to other messages. It's entry into the razor database is pointless and worthless as it would not aid in detecting spam, nor would it have any probability of reducing false positives.
That is a voting system. True, there isn't anything that says a revoke on some mime part will be recorded if there wasn't a previous report.
Agreed, Razor *is* a voting system, but you need to keep in mind that voting only matters in the case of non-unique mime parts.
Simply making the broad statement that we should send non-spam to razor revoke is fundamentally flawed. A large percentage of nonspam messages are simple single-part text/plain or text/html messages with no embedded images or other mime parts. To report these messages is senseless and wasteful.
Now, IF you have a multipart nonspam message, and it has a "common" item as one of it's parts (ie: an embedded image from a common clip-art library) then revoking DOES make sense, particularly if the embedded item is in the database as spam.
However, this is a very different case than broadly stating "revoke your nonspam too".
<snip long message detailing how razor works in great detail and at casual glance is correct, but IMO it's irrelevant because it's not related to the point of *why* you should revoke.>
-------------------------------------------------------
This SF.Net email is sponsored by: Oracle 10g
Get certified on the hottest thing ever to hit the market... Oracle 10g. Take an Oracle 10g class now, and we'll give you the exam FREE.
http://ads.osdn.com/?ad_id=3149&alloc_id=8166&op=click
_______________________________________________
Razor-users mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/razor-users
