Hacker News Reader

Comment by icey | Hacker News Reader

Transifex is probably a competitor in this space (http://www.transifex.net/).

One of the problems that I've had with lowest bidder translations is that the cost of a poor translation can be quite high. Working on a large real estate management application a few years ago, we hired a firm based on cost alone for a bunch of translations; and they did some weird things like using the word for "zipper" when we wanted "zip code". The translations were generally correct, but little things like that really made us look pretty bad in the eyes of our international customers.

So, my point is that I would go out and look for a company that had a good translation track record before trusting the public face of my company to someone on the basis of cost.

View on HN

I'm hoping that some crowd-sourcing features will handle quality better. For example:

1) Similar to 99Designs, etc, everyone could vote on quality, so that could give you some confidence in the text.

2) Translator users could accumulate karma and reviews.

3) Translators' previous work would be public for inspection, and you could see comments and votes on previous work.

4) The requester could seek higher quality by offering a higher price.

I think double-checking results would bring the most benefit, especially if done by a native speaker of the target language. They don't have to understand the source language, just point out that a sentence makes no sense or uses awkward grammar.

They don't have to understand the source language

Find the mistake in the following sentence which cost my company $X (can't tell you, but it is eyepopping):

"The subject registration screen should list the lecture subject name, drawn from the database table for the lectures subject name."

I can find at least one grammar mistake.

There should be an apostrophe in lecture's.

Did that cost you money?

Your correction was precisely what cost $X.

Can you explain why this cost money without revealing too much?

There is a long standing tradition at a major customer of my company, unrelated to this system, that any worker anywhere on the line can hit a Big Red Button which will stop the line if they detect a quality issue. Hitting the Big Red Button has immediate costs: it delays production ($$$), it incurs overtime expenses to get the lost production back ($$$), it throws your supply lines into chaos ($$), it causes you to miss targets with your customers and pay out contractual damages ($$$$$$), etc, etc.

Hitting the Big Red Button on a software project is cheaper, but not free. (For example, the Big Red Button plan for customer acceptance testing might sound like "Nobody goes home if there is an unknown risk to delivery", and a translation error anywhere means there is an unknown risk to delivery until you have verified that whatever process introduced the error was not similarly out of control elsewhere.)

Yes, double-checking like the way 'recaptha' sign up forms work! Two images re-confirmed by millions across the web.

Great idea.

I wonder if there are other ways to confirm data in other instances... (Where recaptha-like processing is n/a).

Reply on news.ycombinator.com