The Human-speed Web
For most Google services, the policy seems to be: “if it’s not instantaneous, it’s broken.” For search, this is great.
For translation, it depends on what you want. If you want an instantaneous approximation of understanding, your option is machine translation, and that’s what Google does at translate.google.com.
But we’re a long way from artificial intelligence, right? And machine translation isn’t going to be putting translators out of business in the foreseeable future.*
So where does translation fit into the age of the internet? The possibility that we believe has a lot of merit, is to use computers to help human translators do their job more efficiently.
Guess what this means from the point of view of someone who’s requesting a translation from a skilled translator?
You gotta wait.
That sound you just heard was the eyeballs of the blipvert generation exploding, right? And this is the sort of internal kvetching I’ve been doing ever since we began this project: “Good grief,” I’d mutter to myself, whilst gnawing at my fingernails, “how am I going to explain to people that I’m working on a website that involves waiting?”
But you know what? I have come to believe that the idea of waiting for a quality service by a human is not really not so weird, even out here on the series of tubes. There are successful services out on the web that are not instantaneous.
Yahoo Answers is a good example: visitors ask references questions, and wait for responses prepared by others.
And I suppose one could argue that journalism in general operates at “human speed,” at least things that are written. The news doesn’t wait to happen, but we’re all willing to wait for helpful interpretation.
So, I dunno, maybe I’m nuts, and the whole world really is utterly impatient and unwilling to accept anything that isn’t instantaneous. But I don’t think so, and we’re betting on it. And we are going to make the processes of requesting and doing translations more efficient, and yes, faster.
* It’s my own personal opinion that once really good MT exists, ie, content that’s indistinguishable from the work of a human, we’ll more or less be living in the age of artificial intelligence. And at that point, all bets are off anyway: whether “real” machine translation is feasible will be a moot point, because there will be machine authors!
2 comments.
Technorati tags: translation
I think there’s a difference here between journalism in the old sense of the word (you know, digging around for facts and actually doing research about them) and in the more PR-related sense of the word (which is more like cutting and pasting press releases. And sometimes correcting the typos in them.)
The former really takes a human being. The latter? news.google.com could become prfeed.google.com very easily.
I think I agree with you that when content and form have actual importance, we need a human in between. Otherwise, “quick, dirty & fast” is good enough.
Hi Dena,
I think I disagree… Let’s imagine something like prfeed.google.com… we can assume it would be filled with press releases. But if press releases are really good for anything (and I have to admit that a lot that I’ve seen aren’t good for much), then it’s about making a good impression on potential customers.
Machine translation is good for understanding, sometimes, but it’s generally lousy at making good impressions. The general response to actually reading machine translation seems to be bemusement. The text itself becomes the topic, and when you’re trying to promote something, you want the thing itself to be the topic…
Or maybe I’m misunderstanding what you mean by PR in this context?
Thanks for your thoughts!