How do they find the 'new user
registration' page? (I'm especially
surprised because some forums don't
have a dedicated URL for this eg,
www.forum.com/register.html , but
instead use query strings or even
other methods invisible to the URL
They find new sites by:
- Crawling and looking for signatures of known software. Usually this is a snippet of text like a copyright or a meta tag but it could be any consistent identifier. This usually applies to blog and forum software.
- Manual inclusion. Human beings, whose labor is cheap in many parts of the world, look for known software or forms that are easily exploitable and add them to a database. This usually applies to custom registration and contact forms.
- They buy lists. Just like email addresses are sold by spammers, known vulnerable or preferred target site lists are sold as well.
How do they know what to enter into
each 'new user registration' field?
They know what to enter into each field by using the field names as a guide. 99.99% of the time the email address field is named "email" or something containing the word "email". You don't have to be a rocket scientist to know that field probably is for an email address. For things like names, login ID, addresses etc. it works on the same principle.
How do they determine what's a page
they can spam / enter data into and
what is not?
They don't care. The automated tools can try so many forms in such a short period of time at virtually no costs so trying every form possible is a no-brainer to do. When human labor is involved they can be "script kiddies" and try the obvious stuff to see if they get any kind of response that indicates the form is potentially vulnerable. Basically, any form is a potential target to them as is any page that accepts user input.
How do forum spambots work?
Do they even 'view' this page at all? ..If not, then I'd assume they're
communicating with the server directly
- how is - this possible? How do they do it?
Where do spambots come from? Is someone sitting behind the computer snickering as they watch their bot destroy site after site? Or are they snickering as they simply 'release' it onto the internet somehow? Are spambots 'run' by an infected computer somewhere? Do they replicate themselves?
It's all automated. Tools like xrumer are built, and sold, and contain the ability to exploit software with known vulnerabilities. Anyone can buy it and after setting it up it's more or less fire and forget. It goes to every forum in its list and tries to spam it to the best of its ability. Just due to brute force it is successful and worth it for the spammers. That's why they never stop. They barely have to lift a finger for it to work.
Can forum spambots break CAPTCHAs? Can
they solve logic questions (how?)?
Yes, but not always. Depends on how well it is implemented. But many captchas, including those offered by big companies, have been beaten and are effectively useless. That's why multiple forms of protection are required to stop them. Even then, humans can usually beat any system.
What techniques are still valid to
From a previous answer: You could do several things (and should be doing more then one) including:
1) Putting a fake field that only bots will see. Then if that field is submitted with the rest of the form you can ignore it (and ban them if desired). You can also trap bad bots who follow a hidden link.
2) Use a CAPATCHA like reCAPTCHA
3) Use a field that requires the user to answer a question like what is 5 + 3. Any human can answer it but a bot won't know what to do since it is auto-populating fields based on field names. So that field will be either incorrect or missing in which case the submission will be rejected.
4) Use a token and put it into a session and also add it to the form. If the token is not submitted with the form or doesn't match then it is automated and can be ignored.
5) Look for repeated submissions from the same IP address. If your form shouldn't get too many requests but suddenly is it probably is being hit by a bot and you should consider temporarily blocking the IP address.
6) Use Askimet. It is great at identifying spam.