Forum Moderators: open

Message Too Old, No Replies

Duolingo to replace recaptcha

recaptcha's Inventor has a new brain twister idea

         

Leosghost

11:34 am on Jun 20, 2012 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member Top Contributors Of The Month



Duolingo hopes to convince millions of people to work for free and thus translate all web content in a matter of years.


At present, it only caters for English speakers looking to learn French, German or Spanish, and Spanish speakers who want to learn English. They start with very simple sentences and work up towards more complex ones, increasing their value as a translator as they progress.


To weed out bad translations, the site asks users to rate each others' answers and chooses only the top-ranked solutions.


[bbc.co.uk...]

IMO ..yet another race to mediocrity via crowd sourcing from the uninformed..those who are beginning to learn a language will vote for what they think is the most accurate translation..the most "popular" will become the "one" ..whether it is correct or not..the "blind leading the blind"..translation depends on context..and far more text is required to give "context" than can be given in a "problem solver box" ( beginners will go "en masse" for what they think is correct, and the actually accurate translations will become the "wrong" )..amazing how intelligence and knowledge has now become something to be "downgraded" in favour of what the ignorant majority think, know or "like"..

bhonda

12:52 pm on Jun 20, 2012 (gmt 0)

10+ Year Member



For some reason this reminds me of the old 'thousand monkeys with a thousand typewriters typing for a thousand years, ending up with the complete works of Shakespeare'.

lucy24

8:26 pm on Jun 20, 2012 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member Top Contributors Of The Month



It made me think of Google In Your Language. I don't know what "Google has high confidence in this translation" means, but I'm pretty darn sure it does not mean it has been looked at by someone in a position to know. I mean, when even I can tell it's gibberish...

A similar venture has been proposed for converting raw OCR into clean text. Same issue there: the majority of readers will change the text to say what they think it should say, not what it actually does say. When you have people who think that a long s is, literally,* an f, and have no concept of semantic change, this can cause problems.


* In the most narrow and stringent meaning of the word ;)

Leosghost

8:33 pm on Jun 20, 2012 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member Top Contributors Of The Month



When you have people who think that a long s is, literally,* an f

Unfortunately in recaptcha, an f is what you have to type for a long "s" :(