Anyone know how to code. Potentially, couldn't a filter be implemented to remove any reference to a term (such as spam from anon-talk), placed on the forum through analyzing ASCII art for a letter set?
In short, what I'm thinking is a basic filter to be placed over new comments on a board that would preview a comment for differences in contrast over the text in order to determine if a letter such as an 'A' or a 'N' is being used. This could potentially provide a filter system which is stricter on defending against a list of banned words. This would both prevent spam on a board (say r9k), or even prevent illicit advertising from other sites.
Does what I'm saying make any sense just by chance? I have some experience coding (halfway through a comsci bachelor and all!), but am not quite experienced enough to determine how practical this would be to implement. I know this sort of system can be used for scanning an object into a program like photo shop, so why not a forum?
anyways, also forwarding to administration as I think something like this might do some good to the board.
-thanks gentlemen!
huh, alright. how taxing would it be on 4chans resources to implement this on a post to post basis?
Name:
Anonymous2010-03-15 1:57
I tried to point this out in a /jp/ thread regarding AnT. Simple regex pattern matching will not solve any but the most simple complications. For example, what do you do if spam is separated by whole words (of a sentence) in between the different letters of the spam? Or if the spam is broken because of a strange word such as spWORDm and a parenthesis says to replace "WORD" with "a." Also, prolific use of wild card symbols and the * regex modifier can lead to incredible slow down, if not an outright freeze.
Now, if you're talking about using the same/a similar methodology that captcha cracking bots use - analyzing the form of the captcha - in this case analyzing the rasterization of the words, looking for patterns of spam - that is a good idea but it also falls prey to the said problems. It is probably easier, too, to use a regex against all typable Unicode characters that look like the letters you want to emulate. You could do simple things with these ideas, yes, but that will just hasten and bolster the use of spam that we can not yet manage, if we are to assume places like AnT are not going to roll over.
In the same vein, imagine if we used such a captcha method to block all images that have a website address in the image, such as the spam we encounter occasionally does. However, this method would also block legitimate website addresses that identify the artist, for example.
moot only employs half-part-time PHP coders who create poorly formed regexps to ban you for a month (hope you're assigned an IP from a pool), with a reporting system that automatically denies your request after 2 hours, so that humans aren't needed.
I rage against the machine by deleting threads I don't like by abusing the "illegal content" reporting feature.
Name:
Anonymous2010-03-15 4:52
I rage against the machine by deleting threads I don't like by abusing the "illegal content" reporting feature.
>>6
Are you implying that those mods that do appear once every few days to delete a few posts here and there actively check their report databases to find people abusing the report system?
>>8 Your proxies can get banned for that*
Hi, I have read your posting over several times now, and I cannot for the life of me find that footnote which you are referencing. Would you please tell me where I can find it?
Name:
Anonymous2010-03-15 10:03
>>9
I think he's flagrantly abusing the regex wild card.