[mniejiki]

>The way I understand that: If you don't pay the people who do the work, and people who make a collage out of their work get paid, you'll end up with much fewer new translations being undertaken.

That's silly. Those previous translations don't magically go away, they still exist. And new translations exist as well except they're now being done by a machine or by a much smaller number of human translator (see below for why this doesn't do any harm).

Having a million people translate "I like cats" into French doesn't give you a million translations. It gives you one translation repeated a million times. That's 999,999 redundant translations. Nothing novel or amazing about those 999,999 translations.

What is the difference between a machine doing those million translations and humans doing identical translations?

Of course, there are novel translations and those would require human input but that's a minority. If the machine can do 99% of translations but requires humans for the additional 1% why is it better to have humans translate 100% rather than just that 1%. You end up with the exact same quality and quantity of translations in both cases.

>You take money from the people who do the work and give it to those who made a collage from it, and the total amount is much less.

The machine is lowering the cost of translations and thus may increase the amount spent on translations. The people who made translations got paid already.

>And I disagree. If I can buy a bun for 20 cents, or someone could rob the baker and give it to me for 5 cents, I'd rather have the bun for 20 cents, because that means more delicious bun tomorrow, and the day after that.

Except that's not the situation. The situation is that someone took the bun, figured out how to make one from it and then started selling their own buns for 5 cents each. You can have that exact same bun forever from the new seller.

Of course, one day you may want a new bun and, you'd argue, that there are no bakers left. But that's silly and utterly simplistic. Once there's not enough bakers with buns to copy that doesn't mean that new recipes will cease to exist but rather than an an alternative will be found. So the new seller will have simply hired some bakers to create a new recipes. Some bakers will continue to have jobs except that instead of wasting all their time on the repetitive tasks of baking they'll instead spend it on innovative tasks of making new recipes (or improving existing ones). You end up with the same diversity and quality of buns now and in the future but at a lower price.

Of course, many bakers will lose their jobs but that's the consequence of all technology and progress. Light bulbs meant that many candle makers lost their jobs. Frankly I prefer light bulbs to candles.

In their place many new jobs will have opened up via the money which people used to spend on buns or candles. That's in addition to all the jobs needed to keep the new bun or light bulb making infrastructure running.

edit: Actually, the "perfect" machine translation would be superior to the average human translation for that 99% since it'd be based on the best human translation for every single piece of text. Needless to say machines are far from there yet but that also means human translators have little to fear job wise for now.

Amusingly, the next hot thing in web translation probably isn't machines. It's humans. Using mechanical turk. So thanks to the web there may soon be more human translators employed than ever before. That's on top of the machines which, to be frank, rather suck at any translation that really matters.

[neumann_alfred]

"The situation is that someone took the bun, figured out how to make one from it and then started selling their own buns for 5 cents each."

So why are they not simply buying the rights of the translations they are using? Because chopping it into pieces and analyzing it automatically doesn't require it? BS, we're talking about big data, if they wanted they could know exactly how many vowels and consonsants they used of each individual translator, and then start haggling.

And "heh" to "all technology and process costs jobs." How many jobs were lost when we found out washing hands before surgery is a good idea? It's such a mindless thing to say. It's a mantra, it's a goal, but far from the truth.

[mniejiki]

>So why are they not simply buying the rights of the translations they are using?

Why do you think they haven't done that? Or rather they paid the entity that owns the rights who at some point in the past paid those translators. Generally speaking the translators gave up their rights in exchange for money long ago.

Specifically google uses, for example, UN documents which translators were paid to translate and which are publicly usable.

If I write some code for a company, get paid for that code and then the company sells that code why should I get paid again?

Or do you think google is the first automated translation service? This has been done successfully for 30+ years commercially and academically. Nothing different just because it's on the web and not on some guys PC.

>How many jobs were lost when we found out washing hands before surgery is a good idea?

Actually, I'm sure there were all sort of quack solutions to infections and whole industries making them which went out of business as a result of doctor's realizing washing hands did the trick.

Edit: Not to mention the loss of jobs for morticians (in the short term), makers of amputation implements (and other tools to deal with all those infections), nurses specializing in dealing with infections, traditional healers (as doctors were now an even better choice) and so on. Lowering infections would have impacted every single person and industry that was built around the previous high-infection status quo.

[mniejiki]

Also, what you seem to not get is that this isn't a zero sum game.

Google translate doesn't cater to the same people as traditional translators. Google translate basically sucks, it can't compete with a half competent translator. It can however compete with a random friend who knows a hundred words in a language, it can compete with spending 6 hours digging through dictionaries, it can compete with asking people on forums for translations and so on. And it has done more than that.

It has increased the market for translations, things that no one would have ever wanted to translate before can now be translated. Things that would have forever been locked away in one language can now be read, barely, by everyone. This is something that human translators without the net could never achieve. The costs and latency were just too high to use them.

That is what technology does in a nutshell, it takes something that used to belong to the elite and brings it to the masses.

[neumann_alfred]

"That is what technology does in a nutshell, it takes something that used to belong to the elite and brings it to the masses."

You can't put technology in a nutshell. Also, you just said stuff nobody wanted to translate can now be crappily translated -- am I understand that this was previously an elite privilege? Why is the one-sentence explanation of what all technology is about, always, different in each post? And what does any of this have to do with me correcting the horrible misrepresentation of the article? That, which I consider my main point stands, the rest I happily concedem because I don't care enough, and you do have a point. But what you said would also be true also for a collaborative, public domain effort, so Google and their middleman dreams can gtfo either way as far as I'm concerned.