Thursday, March 5, 2009

Porn or Not Dot Com

Here’s an idea I’ve had for many years but have not been brave enough to launch. (a) Computers cannot perfectly tell whether an image is pornographic or not; (b) sites such as image search engines need to block pornography; (c) many people like looking at porn. Everything aligns perfectly: Why not let people who like looking at porn tell us which images are pornographic? As a reward, the more accurate they are, the better pornography they see.

The site would be simple. The user sees an image and they have to say if it is pornographic. If it happens to be pornography they rate how “good” it is. By giving them some images for which we know the correct answer (porn or not), we can measure how accurate they are. The more accurate they are, the more high quality porn we give them.

39 comments:

  1. Million dollar idea Luis.

    ReplyDelete
  2. I'm kinda surprised Jimmy Wales hasn't already done this.

    ReplyDelete
  3. We would need someone to categorize porn as ''good'' or ''bad'' (to be used as reward) in the first place.

    ReplyDelete
  4. No. The users themselves would do that!

    ReplyDelete
  5. So we're looking for two orthogonal axes? Goodness and Pornness? Or is it one spectrum from Good not porn to Good Porn?

    I like the brute-force empiricism applied to these sticky (pun!) problems.

    ReplyDelete
  6. I was thinking of 11 choices:

    NOT PORN or PORN: (1, 2, 3, 4, 5, 6, 7, 8, 9, 10)

    ReplyDelete
  7. The only difficulty I see is keeping the kiddies out. This is what tends to generate the most flack consumer watch groups ... solving this problem would be another million dollar idea.

    ReplyDelete
  8. Wait, sorry to kill this parade...

    When someone wants to watch porn, they don't want to spend time sifting through non-porn. There are hundred of youtube-like websites now that take care of any "porn problems." What does the user benefit from this? "Better porn?" I really don't think that's an issue...I think porn was proven to be "NP-complete" a while ago.

    ReplyDelete
  9. Yes, better porn with no ads. Plus, once we get the user to rate a few images, we can recommend better porn specifically for them (like Amazon recommends books that you may like). So if they rate a certain kind of porn higher, we give them more of that.

    ReplyDelete
  10. Luis can you educate us on the types of porn?? Any special preferences or something so the results may be biased towards a certain category??

    ReplyDelete
  11. Better idea. Structure it like http://thefairest.info. Just let people pick which image they like better. Give it a domain name that suggests that the images will be porn, and the non-porn should naturally filter to the bottom as people indicate they don't like it. The only problem you might have is with the really bizarre fetishist stuff that a random porn viewer might find to be less arousing than a random picture from the internet.

    ReplyDelete
  12. Oh snap! I've owned pornornot.com for years. Somebody finally did this :) Good find. Do you know if they are actually using the results to block pornography in other sites?

    ReplyDelete
  13. Also forgot to mention: A higher precision technique is to just replace a random image in a random image search with the image you'd like to classify. If that image gets a hugely higher clickthrough rate than any other, it's porn.

    ReplyDelete
  14. A couple of problems with this:

    * People who like looking at porn typically have specific themes, bodytypes, etc that they are looking for and probably don't want to sift through a bunch of non-porn or non-exciting images before they see something they like.

    * People who like looking at porn might have a different idea of what is offensive or pornographic than the people who want their content filtered. For example, just because a nipple is showing doesn't make an image pornographic or offensive, in my view. Some people get offended when they see that.

    You will have to find a way to mitigate those factors. Maybe add some known porn to the stream of photos to increase the signal/noise ratio for porn surfers.

    ReplyDelete
  15. I still don't get the ideas behind this. How to implement those results back to the porn filter?

    Any specific example?

    ReplyDelete
  16. Hmm sounds like hotornot.com. Nice domain name.

    ReplyDelete
  17. http://crowdsifter.com/

    ReplyDelete
  18. That's just paid humans doing it.

    ReplyDelete
  19. Why would anyone bother to use that web site. Isn't porn free and abundant?

    ReplyDelete
  20. Cuz after you rate a few thing, it will recommend stuff that you like.

    ReplyDelete
  21. This has already been done hundreds of times, please see the video lecture on human computation:

    http://video.google.com/videoplay?docid=-8246463980976635143

    ReplyDelete
  22. Uh, not sure if you meant this as a joke, but the guy speaking in that lecture is *me* :)

    ReplyDelete
  23. earthpr0n: http://www.saveearthgame.org/

    ReplyDelete
  24. This comment has been removed by the author.

    ReplyDelete
  25. So long as it's a-s-d-w navigable.

    ReplyDelete
  26. http://www.PornMuvee.com

    ReplyDelete
  27. it's amazing that so many people know so little

    ReplyDelete
  28. http://flowingwoo.com/
    http://handjerker.com/
    http://webgoggles.com/
    http://pornborg.org/
    http://amianal.com/
    http://amiasian.com/
    http://aminaked.com/

    ReplyDelete
  29. Poker - Free bankrolls $50 - INTERNATIONAL
    no deposit poker bonus Just play and win more cash :)

    ReplyDelete
  30. Online poker - Free $50 for start playing
    No deposit capital

    ReplyDelete
  31. Maybe you claim CS expertise, but it seems to older than you are. Hackers and crackers already have a field day with smut sites as a screen for checking out users. Asking users to rate controversial published images will not affect how those images are used. I find it amusing that a prof. is so naiive in his own field.

    ReplyDelete
  32. I recently came across your blog and have been reading along.
    I thought I would leave my first comment. I don't know what to say except that I have enjoyed reading.
    Nice blog. I will keep visiting this blog very often.

    project free tv

    ReplyDelete
  33. The BASIC definition needed to bring UTILITY to this idea proposed by Luis is the ETHICAL and MORAL definition of PORN. These is a fuzzy logic problem because of the thresholds.

    My definition (and the original one) of porn is "something that disturbs and perverts the rational behavior and the right thinking". It embraces things that are not about sex but can acctually be found mind-linked in psycos, perverts, criminal, sick, deeply depressed or "people without any clear purpose".

    Relating just to sex. Porn or not and the scale proposed by Luis simply needs the DEFINITION about what would be GOOD OR BEST AND BAD OR WORST. The system then will develop toward those definitions.

    Luis, It seemed to me you were sensing the colective stupidity and colective inteligence with this waived idea.

    Your approach will be very more sinergic putting into it more epistemiology, linguistic and philosophy. Don't you think?

    ReplyDelete
  34. Ah, you are the cunt that came up with CAPTCHAs.

    Hope you die nice and painfully when your time comes.

    Burn you prick

    ReplyDelete