What I think would be more interesting would be a way for viewers to stimulate the fish.
Say you have a link that charges the viewer a small fee to feed the fish on a certain side of the tank. Scale the price with how much food the fish has been given (all the way down to $0.00 and up exponentially to not overfeed the fish). Basically, let the internet handle feeding your fish.
I don't think this is a possibility. While you could get some sort of feedback loop going (fish gets hit it gets shocked, fish hits it gets food or something), there would be no way for the fish to understand the massive amount of information they'd need to grok from the game, so they'd just be getting all this random stimuli with nothing to attach it to.
I can't see anyway to bridge the gap of making the fish understand the stimuli in relation to what's going on in the game. I would imagine the same is true for rats.
The fish doesn't need to grok the entire game, it just has to be better than the other fish. (Cue that joke about two people being chased by a bear and one pausing to put on his sneakers. "What are you doing? You can't outrun a bear!" "I don't have to, I just have to outrun you!")
For example, if fish A realizes that swimming around corner X gets it food, because X happens to correspond to "move forward and punch", that's enough to win the game if fish B doesn't figure out the same.
If you like to watch computer-controlled 2d fighting characters with actual decent AI (most of the time) squaring off against each other, check out http://www.saltybet.com/ .
There are over 5000 characters in the database, and it runs 24 hours a day. The matches can get really amusing sometimes.
From a technical standpoint it doesn't seem too hard to set everything up (over time), but the concept and then the monetization of it is pretty ingenious.
I've seen Salty himself in the chat every now and then. I wonder how much time he spends working on the site vs how much he makes off of it every month.
I am not sure if cats are a good idea for this. But what about rats? Could we teach them to do image recognition? Perhaps it could answer questions like "is there a cat in this picture?". Since rats are quite small this approach could scale well.
Watching them for a bit, I would say they're ~=> my mom playing Street Fighter - the occasional times she would play video games with me, the results of her play often looked like this.
Soooooo much more engaging than the Fish Plays Pokemon the creator links to. I explained it to a coworker (...in marketing) and couldn't help but giggle at every other sentence in my explanation. "So the fish swim around in that virtual grid, see, and that triggers button presses in the fighting game." Sweet action, as the kids say.
The position of two fish in a tank are tracked. The tank is divided into a grid and the position of the fish in the grid determines the input into the game.
The description is further down the page. The top-middle sector of the tank activates the jump button for example. If both fish are in that sector, then the two characters will do nothing but jump on the spot until either fish moves elsewhere.
Humans are pattern recognition animals, we like to look at anything random and extract meaningful patterns from it.
I see this game as a clear example of that, where (random) fish movement is fed to a computer game and translated into game play (pattern). I fail though to conclude anything particularly interesting if one fish won the game :)
What I'd really want to see is two monkeys playing the game against each other for the treat. If given some time, I believe that they can get better than most humans.
I had a dream one time when I was a kid that I trained a puppy to play Street Fighter 2. I'd never imagine it would ever be linked to anything this close to reality.
I can confirm that FishPlaysPokemon uses OpenCV. There's a wealth of knowledge and tutorials, so if you just want to isolate certain color ranges (what they're doing), I guarantee there are tons of resources. There's also Python bindings to make things even more friendly. After that, it's some simple math to figure out which quadrant it's in and then trigger the correct button press.
Actually they just have to detect more than $x pixels of $color in each region then do a button press on that. I don't think in the abstract that it's all that complex, given that the fish are of starkly different colors that do not occur in the tank or elsewhere in the video feed.
Just gotta say this is a strange outcome of our technological developments.. I mean think of the energy that was spent making this work - it could've been done doing something.. I dunno fucking useful?!
It's almost as bad as watching two "grown-ups" play Street Fighter lol.
1) the creators had a chance to learn/practice their favorite libraries and languages -- computer vision is hugely useful and applicable to nearly all domains.
2) the creators got spectators to think about the technologies involved, possibly motivating some to learn about them, and undoubtedly teaching a few people about things they didn't know the concepts behind. (computer vision. hell, check the comments on HN, a few of us didn't know how it worked and were thus elucidated.)
