Iconic Conflict
We’ve seen a number of mini conflicts on Tiny, but this appears to be one of the more intense click-offs.
Update: I don’t like having swastikas or other such symbols on Tiny, and there are concerns that it is against school policy to publish this sort of material. We’ve seen a lot of offensive material on here, but when I come across an ideological and religious battle like this it is a reminder that these symbols, despite their small size, can have a very real effect. Perhaps an open, anonymous system like Tiny is not feasible. I can’t police it well myself, nor can I automatically filter the content (although I suppose that’s a topic for research). I’ve considered community moderation systems, but that’s a heavy layer to add on to such a lightweight system. I’m not sure how to proceed. Thoughts or advice appreciated.
rss
October 15th, 2006 at 2:58 pm
In light of this (and of the words formed on the site, too), you sure you want to be associated with this (and how would MIT feel, should the PR people have to explain supporting Nazi symbols)?
October 15th, 2006 at 3:15 pm
Good question.
I’ve thought about it a lot, and I’ve decided to adopt a wait-and-see approach. When I first saw swastikas on Tiny I was dismayed. But I’m also very curious about the apparently common urge that compels people to scrawl offensive or controversial imagery in a public space. I’m not going to be dogmatic about it; if MIT administration or student groups ask me to take Tiny down I will probably comply. I’m not looking to stir up controversy, but I like letting people express themselves in an extremely small, anonymous way.
October 17th, 2006 at 11:18 am
Hey Brent, I kind of took your conflict as a challenge to see what I could do in Python to detect if two black and white images are the same. The result is on my blog at: http://www.whatspop.com/blog/2006/10/detecting-duplicate-images-using.cfm
I presume you can load the image as a blob from the database (or whatever representation) or scoop it up as a file. Perhaps this can work out for you. Do let me know…I would love to help out with such a cool project.
October 17th, 2006 at 11:55 am
Hi Kunal, thanks for your comment. Detecting duplicates isn’t really the issue here. It’s easy to check for uniqueness with a simple database query. The harder task is finding variations of the offensive symbols, or accurately predicting which images people will find offensive. Of course the larger issue is whether such censorship, human or automated, is actually a good idea. We decided to take down all the icons rather then selectively ban specific images.
October 17th, 2006 at 1:03 pm
I just realized the whole interface uses JavaScript…yeah - a database query would absolutely work. As for variations, I can’t think of anything more effective than manually flagging bases, and then running a cronjob to check for those matches. Given the canvas size and color options, I think you’ll be able to find offensive images (at least to you) fairly quickly. Perhaps unleash the site into a Digg like gallery, where people can up/down and mark images as offensive. That would can help solve the issue of finding generally offensive images, rather than the ones you find. With all that said, I look forward to seeing the gallery.
December 24th, 2006 at 7:57 am
The swastika is not considered offensive by many religions, it is used in both right and lefthanded versions by the Hindu. The natzi’s highjacked it.
It is time that people stop attributing power to these symbols then they would not stir up such controversy. One man’s peace sign is another’s insult and as such I believe they should be left alone.
April 27th, 2007 at 10:13 pm
The swastika is a Hindu symbol… ironically I believe it represents peace although don’t quote me on that.
If you want to filter out the database for such things for removal then you have an array of different methods at hand.
Firstly you could use the MD5 hash method, where you create hash strings of files from the images to be filtered and then check the hash of every file in the database to see if it is the same if so remove it. Also useful for checking if a new submission is a duplicate.
Secondly I believe you store the image data as a string of “0″ and “1″s in the database. Decide which string is for the offending icons and remove them with a simple database query.
April 29th, 2007 at 12:04 am
Hi Simon, we knew how to filter it, just not whether we should.