Any word on Chatmon?

[Nishida]

As this thread got locked: http://www.bluegartr.com/forum/showthread.php?t=85739

Wondering what the status was on the plugin update? Getting spammed w/ RMT tells w/ chatmon on..the only thing it successfully ignores is tells from legit people..oddly enough ; ;

[Cairthenn]

As this thread got locked: http://www.bluegartr.com/forum/showthread.php?t=85739

Wondering what the status was on the plugin update? Getting spammed w/ RMT tells w/ chatmon on..the only thing it successfully ignores is tells from legit people..oddly enough ; ;

I find it really strange that you're getting "spammed." You shouldn't get more than a few a day now anyways as I'm fairly sure SE limited the /sea capabilities of trial accounts.

[Wrish]

I find it really strange that you're getting "spammed." You shouldn't get more than a few a day now anyways as I'm fairly sure SE limited the /sea capabilities of trial accounts.

Where did you learn of this?

I'm averaging one tell every 5 hours on all my /anon characters, without any filtering whatsoever. Seems bearable. Use IM or a cell phone for urgent LS notifications. Heck, include IM handle in search comment.

[Nishida]

I find it really strange that you're getting "spammed." You shouldn't get more than a few a day now anyways as I'm fairly sure SE limited the /sea capabilities of trial accounts.

Not spammed as in 20+ a day, but I'm still getting at least 7-10 a day. The idea of a spam filter is that it blocks all spam not just some. Hence me asking if it's been updated yet.

Aikar's post said "over the next few days" and that was weeks ago so I'm just wondering what was up.

[HKR]

Not spammed as in 20+ a day, but I'm still getting at least 7-10 a day. The idea of a spam filter is that it blocks all spam not just some. Hence me asking if it's been updated yet.

Aikar's post said "over the next few days" and that was weeks ago so I'm just wondering what was up.

I'm getting about 20+ a day :/

[Callisto]

I get at least 20 a day and I have both S-E filter and Chatmon up, lol.

[calinz]

Aikar is probably busy irl. I mean if he still played, he would surely be pissed about these himself and long since updated it.

[jurjur]

why not just ignore the rmt tells?
7-10 a day is nothing

[Cairthenn]

Where did you learn of this?

I'm averaging one tell every 5 hours on all my /anon characters, without any filtering whatsoever. Seems bearable. Use IM or a cell phone for urgent LS notifications. Heck, include IM handle in search comment.

My friend made a trial account for some odd reason, tried to /sea all to find me and it literally did not work at all (for me, or any other name).

[Spekkio]

it would be very easy to use a cheap, bought account or lvl a single account a few lvls in order to feed the bots names, not to mention making a database of names to spam and simply using that. nerfing /sea is not the solution.

[Xaru]

it would be very easy to use a cheap, bought account or lvl a single account a few lvls in order to feed the bots names, not to mention making a database of names to spam and simply using that. nerfing /sea is not the solution.

Pretty sure FFXIAH.com has all the names they need anyway

[nitsuj]

Aikar is probably busy irl. I mean if he still played, he would surely be pissed about these himself and long since updated it.

this.

[Nishida]

this.

Ya I figured as much..just figured something would have surfaced by now, but I guess I don't realize HOW busy he is. No worries =P

[Subadai]

Hmm, I only get 2 or 3 a day max... In the end, the only filter that is 100% is the human filter (simply ignore them).

[Deimos]

I had a brief conversation about this with Aikar, but the problem is that ChatMon, in its current form, can't solve this problem any more.

The only reason it used to work is because spammers weren't actively testing against ChatMon. If Aikar's suspicions are correct, and spammers are downloading ChatMon to test against, there's nothing it can possibly do. All they have to do is run it on one computer, and try to send tells to that computer from another and see what goes through. Once something goes through, they can safely spam it and it'll get through to every user of ChatMon. When a pattern to block that tell is added to the ChatMon rules, they just start over and find a new one.

The only way to make it work would be a total rewrite, using more sophisticated techniques similar to those used for detecting spam email. Simple pattern-matching won't do the trick any more if they're testing against it. Something like a bayesian classifier would probably work, but would require large corpora of spam and non-spam tells, which I doubt many people have available, or would even want to share.

[NynJa]

Not spammed as in 20+ a day, but I'm still getting at least 7-10 a day. The idea of a spam filter is that it blocks all spam not just some. Hence me asking if it's been updated yet.

Aikar's post said "over the next few days" and that was weeks ago so I'm just wondering what was up.

Do you know of a concievable way to install a filter based on how these messages get through these days?

And I hate to break it to you, no filter is infalliable.

[nitsuj]

I had a brief conversation about this with Aikar, but the problem is that ChatMon, in its current form, can't solve this problem any more.

The only reason it used to work is because spammers weren't actively testing against ChatMon. If Aikar's suspicions are correct, and spammers are downloading ChatMon to test against, there's nothing it can possibly do. All they have to do is run it on one computer, and try to send tells to that computer from another and see what goes through. Once something goes through, they can safely spam it and it'll get through to every user of ChatMon. When a pattern to block that tell is added to the ChatMon rules, they just start over and find a new one.

The only way to make it work would be a total rewrite, using more sophisticated techniques similar to those used for detecting spam email. Simple pattern-matching won't do the trick any more if they're testing against it. Something like a bayesian classifier would probably work, but would require large corpora of spam and non-spam tells, which I doubt many people have available, or would even want to share.

This is why I'm not sure I'll be doing any sort of chatmon for windower v4 yet. Most likely will make the "tell alert" and "check alert" functionality built into a callback, but the spam filter may be better left to SE at this point.

[jurjur]

is there a way for chatmon to stop filtering and only do "incoming tell" and all that garbage?

[Pikarya]

Stolen from FFXIAH. But wow... lol

http://i48.tinypic.com/2wdvbeg.png

[aucks]

Would it be impossible to generate a Whitelist function for Chatmon that only allows certain names through while not seeking for a party? Like check to see if it's a GM, then scan the friends list, or linkshell, or txt file of allowed names, and if the name is found, let it through, if not block? Then, if you set yourself to seeking status, allow all tells through?
I'm no programmer, but to me that seems like the only way to block the vast majority of them...