Log in

View Full Version : analyzing spam..constructed for a purpose.


BanMe
January 19th, 2011, 13:30
So recently I've seen something here and other places that annoys me..spam marketing,which isn't really different from other spam except the authors look for the places where links are allowed and exploit those spots..our blog.I also know of the details around the subject of filtering, and the more known captcha methods..no ocr can recognize a fuzzed pic. but what otherways could there be..would the staff send me the 3 most recent entries please,or quote them here for side by side analysis...not at home..thanks

disavowed
January 19th, 2011, 18:16
Actually, most CAPTCHAs can and have been broken. See http://bitland.net/captcha.pdf, for example.
And of the ones that really are strong and can't (yet) be broken automatically, the bad guys just use http://mturk.amazon.com/

BanMe
January 19th, 2011, 18:46
two things I didnt know of.. thank you.. :]
Is mollum one of those 'tough' ones?

BanMe
January 20th, 2011, 10:53
Btw all that Paper sums up to is that a technique with no solid colors,maybe even splashes of color overlappng making up the various letters would remove the effectiveness of attacking that surface in this manner,but then your just in a waiting game for fuzzy pattern recognition,and its at it again..

BanMe
January 23rd, 2011, 20:43
Code:
http://www.woodmann.com/forum/attachment.php?attachmentid=2411&d=1295832192

so I tried a different vector of analysis..

this is very bad o0...(seriously!)...think 3 way redirect.. I wont go into the automation details..and I do understand the losses.. but..iono, a container check for resource location..might do it..
2nd test
being that the first one "failed." or "passed"?..
Code:

http://www.woodmann.com/forum/attachment.php?attachmentid=2412&d=1295834519

this one passes..barely.. but i bet with the right 'wrapping' it would work...

no more tests.. or weapons for teh kiddies.. pls delete this is necessary..