Seriously how.jpg? The CAPTCHA makes it difficult to run something from a server, and I know there's CAPTCHA solvers out there but don't know how legit they are.
Fags aren't really running greasemonkey scripts in background, with CAPTCHA already filled out, waiting to post are they? Even worse would be the F5 fags tiling two browsers, one to refresh and the other to post.
The problem of breaking CAPTCHAs with moderate consistency is not a difficult one- if you have the time to create a tailored solver and some basic computer vision experience you can go a long way.
That said anyone who does have the required skills would likely have far better things to do than spam /g/ with Gentoo posts.
Either they are using a 3rd party human solver service, decaptcher et al, or more likely, manually solving a batch of them before turning on a spambot.
The CAPTCHA makes it difficult to run something from a server
This is not the case at all, any student half way through SICP could write a cURL/wget script that manages CAPTCHA entry and usage in an efficient manner.
I wrote something like this once in .NET but gave up when I realized 1) It's almost impossible to get any decent precision since there are so many constantly fluctuating variables, and 2) others have already done it: http://4chansearch.com/wizardget or http://segnosis.net/postpredictor/
If you simply must write your own, though, this is the rough idea:
Step 1) Notice that when you have javascript off, the CAPTCHA procedure is different. You fill it out in a separate iframe, which is actually a google site, and copy over a long base-64 code into a separate box into the 4chan form. Thing is, these long strings contain no information in them except that they came from 4chan. i.e. they're not actually attached to the post form they were loaded in. By simply loading the google site, filling out the captchas, and saving the base-64 strings in advance, you can submit these in the appropriate form field to post anywhere on 4chan. (Note that they expire in 5 hours, though)
Step 2) Repeatedly load /b/ (I assume) programatically to monitor the last post number used together with the time.
Step 3) Using whatever method you think will be accurate, estimate the precise moment you need to submit the post to get the number you want.
Step 4) Have your program construct a fake post request and submit it at just the right moment. You might use libcurl for that, or some other library, or do it directly with whatever internet capabilities the language your using has.
Step 5) Tweak your estimation algorithm, add more and more variables, get frustrated, realize this will never work as well as you want it to, and give up.
function getpostnumber(html) {
var p1 = /[\'|\#\q][0-9]{9}/;
var p2 = /[0-9]{9}/;
var out = p1.exec(html);
var out = p2.exec(out);
return ''+out; //gaah! does anyone know of any better way to typecast?
}
function countlastdigits(html) {
var str = getpostnumber(html);
var n=0;
var last = str[str.length-1];