From the FAQ:

  Do you really expect to be able to detect and 
  filter anything that's conceivably stupid?

    No, of course not. You'd need real AI for that, 
    and beyond a certain point it's simply subjective; 
    after all, a sufficiently advanced AI would probably 
    filter out the whole of human discourse, which 
    isn't the idea.

I have to admit that I laughed pretty hard when I read that...

I think this is barfing on a lack of punctuation.

"this is the most shittest game online ever full of little nooby kids" (taken from a youtube comment) is "not likely stupid". Add a period at the end, though, and it works.

Maybe the parsing system isn't designed to handle incomplete sentences or sentence fragments? That could be difficult but key; many of the bottom of the barrel posts on the internet have no punctuation at all.

Update: Starting to really take an interest in this. While there are false negatives, I can't seem to find any false positives.

Unrelatedly, one of the challenges in analyzing modern text with traditional NLP tools is that the tools usually expect standard English, whereas the text is rather colloquial, and the punctuation is for timing purposes, when present at all.

This is sort of like a Youtube Snob for the rest of the internet. http://www.chrisfinke.com/addons/youtube-comment-snob/

"How is babby formed?" -> Text is not likely to be stupid.

Incorrect.

This looks stupid to you because you see it in the context of a stupid meme. When seen this way:

    How is <unknown word> formed?

It doesn't look stupid at all.

A stupidity filter that doesn't block stupid memes is not a stupidity filter.

That question could be about the stages of fetus development, which is not an stupid thing to ponder.

They need to do way instain filter>

Try submitting YouTube comments to the demo. This is shockingly accurate.

I think you just discovered its training dataset

Exactly right: http://stupidfilter.org/random.php

Not surprisingly, 9 of the 10 YouTube comments I tested were classified as "likely to be stupid".

I can't get the thing to detect any stupidity whatsoever. I tried: -dude that was pretty funny lol -dude thats a funny one! lol -ur a ghey lord! All passed as not being likely to be stupid.

lmao!!!

>lmao!!!

Text is likely to be stupid.

This is, apparently, not likely to be stupid:

Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

I hope this doesn't end up equating having English as a second language with supidity.

If something like this was actually used, I predict we would just see a shift from "I'm in ur base, killin ur doods" to "I am in your base, and I am in the process of killing your dudes."

I posted "This is stupid" into the demo ( http://stupidfilter.org/demo.php ) and it said "Text is not likely to be stupid."

I put in "Goldman Sachs is crooked" as a query...and it returned:

"Text in not likely to be stupid"

Clearly this is a very perceptive algorithm!

Anyone up for StupidGolf? :-)

My entry: "haha... lol" (11 chars)

I win.

"!" (1 char) is likely to be stupid.

I feel like the bigger challenge is finding the smallest piece of text that isn't considered stupid. I hypothesized that this would be difficult, but it turns out 'a' isn't likely to be stupid.

I tried single letters but didn't think of that.

"o hai, sup?"

also 11

first! is classified as "not likely to be stupid", which shows how much work it needs.

It probably just indicates that they have not pointed it at Slashdot comments as a training set yet.