Filed under: chain-gang,international,Not so popular,null,wikipedia
Google’s APIs Product Manager Adam Feldman announced on Thursday they will cancel the Google translate API by December, without replacing it, and that all use of it will be throttled until then. Any reusers or libraries relying on the translate API to programmatically provide a better multilingual experience will have to switch over to another translation service. (Some simple services will still be available to users, such as google.com/translate, but APIs will not be available to developers of other sites, libraries, or services.)
Ouch. This is a sudden shift, both from their strong earlier support for this API (I was personally encouraged to use it for applications by colleagues at Google), and from their standing policy of supporting deprecated services for up to 3 years. What could have spooked them? Why the rush? As of today, the Translate API page reads:
The Google Translate API has been officially deprecated as of May 26, 2011. Due to the substantial economic burden caused by extensive abuse, the number of requests you may make per day will be limited and the API will be shut off completely on December 1, 2011.
Most disappointing to me is the way this announcement was released: buried in a blog post full of minor “Spring Cleaning” updates to a dozen other APIs. Most of the other deprecated APIs were replaced by reasonable equivalents or alternatives, and were being maintained indefinitely with limits on the rate of requests per user. None of them is being cancelled within six months, and none of them are half as widely used!
I hope that this obfuscation was an unintentional oversight. There have been 170 irate replies to that post so far, almost all about the Translate API cancellation. But it has been three days already without any significant update from Feldman or any mention of the change on the Google Translate blog. Google’s response to a ZDNet inquiry was that they have no further information to provide on why they made this decision.
I am honestly unsure why they are doing this. Google used to have guaranteed relevance to many undiscovered future innovations in translation, thanks to their excellent API work and their monopoly on reuse today. They had a few years’ lead time on Microsoft (for instance), they could see how almost everyone in the world was using their translation tools. Now they are giving up that guarantee and goodwill, but for what?
What was the extensive abuse? Spammers and copycats? What constitutes a substantial economic burden? I can think of no abuse that could not be effectively stopped by throttling or charging for the service. Many of the current users asked to be able to pay for their API access, but have received no response.
The open question for groups like OLPC and Wikimedia, which had considered ways to use the Translate API to provide meaningful translation of interfaces and streams, is what toolchain to use instead. And whether to invest in making one of the current (incomplete) free toolchains much, much better.
Now 165 replies. ZDNet’s Ed Burnette calls this ‘pulling the rug out from under web developers‘.
Some people have suggested other closed alternatives, like MyGengo, whose founder says “since our startup relies on translation, you can be guaranteed we’ll never deprecate our API :)” — but that’s hardly compelling. As soon as they change their business model or get acquired by a larger organization, they may well drop it. And their API is no good if they go bankrupt or close shop for any reason.
What we need is an open API to a library of freely licensed and publicly maintained code — which should include wrappers that interfacing with a spectrum of options from automatic machine translation, to human-assisted translation that updates a free public translation memory, to private human translation.
Tools like those from MyGengo or Babelfish or MS or (for the next 6 months) Google, which are public in a sense but could be killed without warning, could be available as options provided through these public wrapper. But updating a service that uses one of those tools to use a different one should take minimal effort.
Comment by metasj 05.29.11 @ 4:52 pmThanks for the post didn’t realise this was in the pipeline.
I’m using plugins based on googles API’s for easy translation of my Quebec sites. I will take a look at mygengo and see if that will work.
thanks again for the heads up could have been painful if i didn’t know
Comment by Dan Jensen 05.29.11 @ 10:28 pmIn the News: AT…
In the News: AT…
Trackback by Brandchannel 05.31.11 @ 8:59 amThanks. Cloud computing ….
When the clouds get dark it’s better have our things at hand, on the ground, as our feet …:
Comment by Pegro 06.01.11 @ 7:10 am==== Why we will have web servers at home ====
http://sinwindows.wordpress.com/2010/12/22/por-que-tendremos-servidores-en-casa/
What this hammers home is that web APIs are just another form of proprietary software.
Comment by David Gerard 06.01.11 @ 2:40 pm