What would be really interesting is if they implemented a system like the old ESP game. For those unfamiliar, it was a collaborative game with strangers over the internet. You and your stranger partner are both shown the same digital image, and you have a textbox to enter tags for that image. There was a limited time and the goal was to get to a point where both of you had entered the same tag. Over time, as the database got trained, common tags such as girl, man, bird, etc were blacklisted, and so you and your partner had to come up with more informative tags to play the game.
Unfortunately, Google bought out the ESP game, and then after shut them down.
Thanks for the idea! We've gotten a bunch of feedback asking about more specific categorizations once the basic set of votes are saturated. This could be a pretty fun way of getting at something like that. We'll keep it in mind going forward!
Done with Meteor! I recognize the uids in the urls, also "Meteor.release" at the console.
Btw, loading a single gif url, like http://gifgif.media.mit.edu/gif/13AbwdfEHTvih2 is too slow. dev console shows "Cannot read property 'num_votes' from undefined". You need to prepare the template code for receiving an empty collection while the subs fill the data, i normally shortcut it as Collection.findOne({id: foo}) && Collection.findOne({id:foo}).num_votes. to prevent your error.
Also, some waitOn and loading templates on iron-router will make the overall UX much better with few changes. specially on the results page that is empty for a long time until content kicks in.
Regarding the site concept, its fun. I spent too much time there already. I will show it to my eve game mates. At eve chats the text line to "emotional" gif ratio is about 1:1. Good job.
Neat concept, I rather enjoy the website design. However when I'm looking for reaction gif, I use the album I've bookmarked on IMGUR which is already hand curated by Reddit.
This could make for a twist on CAPTCHA. If you took pairs of gifs with huge vote disparities you would effectively have a right answer and a wrong answer.
You'd have extremely low entropy though. If I write an auto-signup script that just selects one or the other randomly, I'm still getting 50% of my botted accounts through.
Sure you can add more, but how many can you add before users get upset about doing the same thing over and over again? Three? Four? At four times I'm still looking at a 6% chance of getting an account created if I just select randomly. That's plenty.
Suggestions to use things that humans can easily distinguish but computers can't as captchas come up often. However, being only human readable isn't the only requirement for a captcha - another requirement is that it can't be computer guessable. If a computer can guess correctly with a high (>1%) chance of success, then the captcha is pointless.
Oh I'm not advocating that this actually being used as a captcha. Though, as a thought experiment, I think it would be more effective if you had, say, eight gifs. Instead of an either-or choice you could ask the user to choose all that match a given condition.
I had a teacher who employed "multiple multiple choice" questions on his exams. The question would have some number of choices (e.g. a,b,c,d,e) and you had to choose the correct answers. These questions were tricky because the answer choices could all be wrong (meaning you select none of the answer choices), could all be correct, and any combination of the answers could be correct. It made guessing go from a 1 in 5 situation to 1 in 32. I was not a fan.
Doh' - a check showed that we were "load balancing" across all of 1 port. Pulled from the wrong git branch on one of our updates apparently -_-. Running through many more node instances now - hope it's up for you. Thanks for pointing it out.
Unfortunately, Google bought out the ESP game, and then after shut them down.