Wiki Spam

[Lalolyen]

Actually if the admin of the wiki would contact me.. I can give ya some code to stop auto-signups on the wiki and can give ya some code to stop spam submissions.

Everyone has seen the "type what you see in the image" to create an account but the anti-spam script is our own that I created a few years back. Simply put, its the same type of preventative script that VB and phpBB uses to keep people from flooding a forum, based upon time.

[Theeper]

CAPTCHA's won't stop the bots. there is plenty of CAPTCHA detecting OCR software out there that works great. The easiest thing to is to put a simple hidden form variable or add some custom field to the reg page. Then, the spammers have to manually modify their bots. They won't want to do it just for one site.

[Lalolyen]

Mine will I haven't seen a CAPTCHA bot yet that can answer questions thats presented in an image =).

Such as in text "What color" In image "is the sky"? And...
in text "What are the first two letters in" in image "Rhumspheld"?

If bots can break that, its time for me to retire.

You would really have to have one hell of a reason system built into that bot.

[Theeper]

You should start your retirement process now then :p~

Since your questions couldn't be completely random, it's just security through obscurity. All I need to do is manually refresh your page a few times and write down all the answers one time.

Since a bot can easily read your image text, it's simple to plug in the answers and have the bot compare text strings and generate the answer.

The main principal behind CAPTCHA is that the text is completely random, so it would be very difficult for someone to generate a hash table or maintain a list of answers. Setting a finite amount of possible solutions actually makes it easier for bots to crack your system.

Granted, noone would spend much effort to crack that type of CAPTCHA on a small site like this, but if it were a piece of widely distributed software or some big site, it would be cracked within minutes.

[sfisque]

except that lalo's solution works fine for "untargetted" attacking. if someone is taking the time to sniff the site for vulnerabilities in order to hand craft their bot's attack, we can assume it is a targetted attack from a (most likely small) set of addresses which can be black listed.

security through obscurity works fine, if it is maintained (read: the dictionary of questions is modified over time to prevent staleness), for a "one off" which our wiki would be (there wont be 1000's of shrink wrapped releases of our wiki).

== sfisque

[Lalolyen]

Originally Posted by sfisque

except that lalo's solution works fine for "untargetted" attacking. if someone is taking the time to sniff the site for vulnerabilities in order to hand craft their bot's attack, we can assume it is a targetted attack from a (most likely small) set of addresses which can be black listed.

security through obscurity works fine, if it is maintained (read: the dictionary of questions is modified over time to prevent staleness), for a "one off" which our wiki would be (there wont be 1000's of shrink wrapped releases of our wiki).

== sfisque

Yeah but come on lol, the spammer is posting advertisements. Thats not targeted in the least, thats the same as email spam; someone has a list of URLs set up in a bot to go out and advertise their websites for the search engines to pick up on. I've seen the same thing on MediaWiki quite a bit.

Image verification, even the simplest verifications will shut down the spam we have been seeing.

If someone is outright attacking the website, then the admins need to get a hold of the FBI, they DO investigate those whom intrude in networks and information systems (including gaining unauthorized access to spam the wiki).

[Theeper]

CAPTCHA will not stop spam. Bots like XRumer have proven that.

Unless our wiki software was custom written, it is just like the thousands of other sites that use it .. which is why spam bots hit us.

Regardless, why go through the trouble of making images and creating a DB of questions/answers ? Just add 1 custom field into the reg page, like a checkbox. Then, any bot would have to be programmed to submit the extra data.

I have run a few sites with phpBB on them for several years and the only thing that has ever eliminated spam on them is by slightly altering the submitted fields. Every CAPTCHA mod I tried was cracked before I even installed it. Granted, phpBB is much more widespread than this wiki software, but the principal is the same.

There is a new trick spammers use .. they set up a free porn site, harvest your CAPTCHA image, post it to the porn site, make surfers solve it to view nekked pics, then store the answer so it can solve it on your site. Even the most advanced and unreadable image can be cracked this way.