This looks like a potentially great project, but this is probably the most badly designed landing page of all time.
The call to action is a really tiny gray icon on the bottom. It made my head hurt just trying to figure out how to sign up for their beta list.
Really? I had the opposite reaction. I found the simplicity drew me to the text. At the end I clicked the footer because it seemed shiny/pretty, and it took me right to the email form.
I feel like a lot of websites, especially the ones which want me to sign up or understand their service, throw way to much crap at me for it to make sense. Two sentence mission statements, ominous screenshots, and cheeky feature lists only go so far --- in this case I appreciate the straightforward mission statement.
I agree, it's a little obscure. But to be fair, the page is so simple, and there's so few things to explore, it's hard to not get curious about what's in the corners. On the other hand, a poorly calibrated monitor could potentially make the lightish-gray-on-not-quite-as-lightish-gray stuff nigh invisible.
Sure, you can see it if you look for it, but the only purpose of this landing page is to collect emails and twitter followers, and it could do that much better if it had a big visible email input field right below the main paragraphs.
I can't talk for others, but I love that page. The "Call to action" button took me literally no time to find and I was happy to click on it. A bit half-page button would have just frustrated me and I wouldn't have subscribed.
While I agree it's not the greatest landing page ever, I don't think it's THAT bad. I spent more time typing this then it took me to read the page and give them my email for an invite.
I like the simplicity of it. Simple logo, text that isn't under contrasted. Simple description of what they're working on, and simple ways to sign up for notifications.
For those who don't know, the founder is Luis von Ahn[0], a CMU CS professor. I know someone majoring in computer engineering there who's taken one of his courses, and apparently he's a very engaging instructor as well.
I saw him present his talk about "harnessing the power of people doing meaningless things on the internet" at an AI conference back in 2003, where he introduced what became the Google Image Labeler.
Easily the best talk of the conference - he's a great (and engaging) speaker.
I think this google talk is roughly the same presentation:
I've done research with Luis, and some of my friends are working on Duolingo. He's certainly an interesting person and a great lecturer. You really come to appreciate his teaching skills after taking his course (most take it as freshmen) and realize all the other professors are nowhere near that competent in teaching.
dissenting opinion: It's easy to teach an interesting course when you get to handpick all of the material. 251 is called "Great Theoretical ideas of Computer Science" because it covers many of them. Alternatively, it's hard to come in and blow your students away with matrix algebra day after day.
His other popular class is called "Science of the Web," which is similarly open-ended.
He's undoubtedly a great teacher, but the other professors are also very competent; albeit in less sexy topics.
small counterpoint to your counterpoint: I had him for 15-381 (Intro AI for non-CMU people), it was his first time teaching the class, and his lectures were substantially better than those of the other professor he co-taught with and had been teaching the class for many semesters. He was either my favorite or second favorite CS prof at CMU.
I worked with him in my undergrad research. Luis is one of those people who communicates with such ease and simplicity that not only is it obvious he's extremely smart, he actually makes you feel smarter as well.
I've seen Luis von Ahn (founder of Duolingo/ReCaptcha) speak twice about his new project (once at a CMU Project Olympus update and once at TEDxCMU a few weeks ago).
A few things that may be of interest to the HN crowd:
* This project is currently academic in nature, funded by grants he has received. However, he does see an opportunity to monetize the product if they choose to by offering translation services to companies or organizations in the future.
* The product is currently in testing. According to their metrics, the crowd-sourced translation is as accurate a professional translator. At TEDxCMU, he showed a professional translation side-by-side to a Duolingo created translation - the two were nearly identical. Likewise, according to their metrics, the education received is as good or better than the leading language education solutions (ie, Rosetta Stone).
* He showed some amazing projections on how quickly they can translate a set of text from one language to another. I forget the exact projections so don't hold me to this, but with 1,000 users it would take, say, 3 months to translate English Wikipedia into Spanish. With 1M users, it would take less than a week.
It's hard to imaging how the task of translation can be combined with language learning since one requires great competency and the other requires little or no competency.
If you can't wait for Duolingo, you can start learning French, Spanish, German, Italian or Dutch vocabulary at http://inglua.com/en (a YC reject from before it became fashionable).
I just tried this for my native tongue, Dutch, and I am getting stuff wrong because it doesn't have the correct translation for it.
Stuff like "the friend" is translated to "de vriendin". Or "near the sea" can't be translated as "bij de zee" but has to be translated to "vlakbij de zee". A lot of them are context dependent as well, and don't make sense the way they are presented.
"Forward" is one I kept running into, it can be translated in two different ways depending on the context, "naar voren" or "vooruit" (BTW, define: vooruit in Google doesn't give you anything, search in de Dikke Van Dale a Dutch Dictionary). It made it immensely frustrating to go through Level 1 and I just gave up after a while.
Thank you for your feedback. I try to add as many synonyms as possible (5 for Dutch in the case of "near the sea"), but some always slip through.
Also, any words that are context dependent should have an example that clarifies the meaning: e.g. for "the friend" the example given is "After I told my friend, she started crying." so you know it concerns a woman, not a man.
In general, inglua works much better if you use it as intended, that is, to learn languages you barely speak at all (and that's obviously the use case I aim for).
Reminds me of http://www.solvemedia.com
A company from my hometown. Advertising for movies and media, within the CAPTCHA. Their conversion rates are quite impressive.
This was a SolveMedia CAPTCHA? The ones I've seen that they having been promoting force you to watch 10-sec movie clips, among other things, and then enter the tagline of the film.
They've been going at now for at least two or three years now, finally claim to be in the black. Some deals pending with pretty big publishers, should be announced soon.
It seems that killing two birds with one stone has worked well for the founder in the past. ReCAPTCHA does it itself, so it will be interesting to follow this project and see where it ends up.
I'm not so sure about that. Just as reCAPTCHA seems like it's just aimed at blocking robots but is really digitizing thousands (millions?) of books, this project claims to do education and translation.
Google Translate is nigh-unusable for anything besides French, Spanish, and sometimes German. Russian is bad enough -- Chinese, Japanese, and Korean are incomprehensible.
I've seen Google Translate do excellent translations for Russian, Hebrew, and Arabic. Certainly not fully gramatically correct, but accurate enough to convey the meaning of virtually every sentence and paragraph -- which is the point of machine translation.
For western languages, yes - for others, not so much. It's still pretty bad for chinese/japanese/korean - even if it's better than anything else out there.
I made a quick site a few months ago that will let you search for stuff in other languages on Twitter - it translates the results back to English for you.
Unfortunately, I couldn't figure out the magic incantation to get the courtesy limit raised at Google, nor any way to pay for more capacity so it will probably get rate limited for the day pretty quickly.
I don't know how the system would work from a technical perspective, but I can think of a few use cases from the past couple of months that fit very well. For example, all the reports coming out of the Japanese media/government. Now that I think of it, anything with time-critical data would benefit from this.
