4 ms·
A reCAPTCHA is more than a CAPTCHA, it also helps to digitize old books.
- amichail 20y agoIMO, human computation is a field with enormous potential. But strangely, there are few publications on the topic. It's not taken seriously in computer science yet. Or maybe it requires a different sort of thinking than what most computer scientists are accustomed to? Or perhaps most computer scientists feel that human computation is cheating? For them, maybe there is no reward in this sort of research. They are more interested in something technically difficult, even if the results are much worse.
- amichail 20y agoBTW, in terms of cleverness in human computation, http://listengame.org http://listengame.org is a must-see. You can use the approach behind listengame.org to build a service that gives first impressions on all sorts of things. In fact, you can apply the listengame.org approach to online advertising to encourage people to look at online ads very carefully. Moreover, advertisers would get feedback on their ads.
- nickb 20y agoJust a quick warning regarding this service. After playing with this, I've noticed that about 20% of the words that were displayed were completely unreadable. If you're a startup that doesn't want to lose any potential users, I'd advise against using this service. If you're an established company that is getting hammered with spam, well, maybe you can afford to use this.
- vlad 19y agoThe ones that are unreadable are likely the words from the texts. Those words do not count against you. You only have to get the real captcha correctly, which happens to be the clearest word. Try experimenting by entering just one of the words. I had no difficulty with these, while I usually have a LOT of difficulty searching vBulletin forums I am not a member of. I would write "If you're having difficulty, enter one of these words" at the top. The user will likely enter the clearest.
- spiralhead 19y agowho's to say the user will not enter an incorrect word or mis-type? To get reasonably accurate results, you would have to take a handful of samples for each word and take the one with the highest frequency. And even then, users will figure out that they really only need to enter one word and random crap for the other one (it will always look different). Not a bad notion but too much room for error and randomness to be reliable.
- amichail 19y agoHow do you know it will always look different? Maybe the look could be faked reasonably well so that both words would look similar.
- spiralhead 19y agoMaybe is the key word there. It would be challenging since you would have to apply some sort of noise filter to each of the fakes to simulate a scanned image. You would also have to consider fonts