GOOD BYE BLOCKBUSTER (HUMAN TRANSLATION), IT WAS NICE TO KNOW YOU?

I read in the paper yesterday that after 25 years, Blockbuster filed for bankruptcy to reduce its debt and restructure its business. I remember when Blockbuster moved into my neighborhood in San Francisco, about 25 years ago. It was on Geary and 14th Avenue. I used to love to walk there in the afternoon from my apartment near Golden Gate Park, look at the new films being offered and people looking for cheap and (mostly) harmless entertainment and then walk with a tape or two back home. I used to time it so that I would be coming back at dusk. I loved walking the streets of San Francisco when the sun was setting and the colors were fading.

Blockbuster followed me wherever we moved. It was not too far from our  house in Santa Rosa. And it is only a 10-minute walk from our house in suburban Chesapeake here in Virginia. I will be sorry to see them go if my local Blockbuster store closes too. But the fact is, I don’t walk to that store anymore. Like so many former Blockbuster customers, I now have a blu-ray player connected to a wireless network. If I want to watch a movie, I go online to Netflix, put a few film titles into my queue of films to watch and I can then watch them instantly on my teevee if there is nothing worth watching on the movie channels which is usually the case. How did Blockbuster miss the fact that Internet would kill its walk-in business? Wasn’t it obvious already quite a few years ago that renting a domain online was much cheaper than paying rent for stores in thousands of buildings? At the peak of its successful business, Blockbuster had about 9,000 stores worldwide. It still has about 3,000 stores with an average size of 5,000 square feet in the US. There is a blockbuster.com domain, but my blu-ray player came with an offer of a wireless connection to Netflix, so that is what I signed up for. Somebody high up in the Blockbuster management must have been asleep for a long, long time.

The most popular articles on my blog are posts dealing with machine translation. When I look at the search terms people used to find my blog, I see that literally every day people end up on my blog after running a search with a phrase such as “When will software replace human translators?” (Note that this query, from yesterday, is not about “whether” but “when”. Translators are doomed. It is only a matter of time, see also the last comment on one of my posts about machine translation). Human translators are afraid that computers and software will obliterate their jobs, as they made redundant so many other professions, from bank tellers to travel agents and all kinds of managers. They are afraid that they too will be busted by Internet, which is what is happening to Blockbuster now. Machine translation (MT) is indeed ubiquitous on the Internet. In most countries where English is a foreign language, MT is just a tab in your browser. It is free and everybody is using it. Including my brother in Europe who uses it among other things to read my blog because he does not know English. But he is complaining bitterly that he can’t figure out from the MT function what it is that I am saying. So we spend an hour or so talking on Skype, which is another example of how Internet and software ruined a very profitable business for long distance phone companies. A few years ago, that hour would have cost me 20 dollars. But talking on Skype is free, just like machine translation.

I have good news and bad news for translators who are worried that Internet and machine translation will kill their business model as well. Let’s have the bad news first. It will probably kill the kind of translation that is not really very important. The kind that people don’t really want to pay money for. Translating my blog, for instance. Really stupid business owners abroad will continue to use MT to translate the content of their websites into English. I have seen websites of “pensions” in Europe in English that were so hilarious that they must have been done with machine translation. It is not a good idea to use MT for this purpose. Potential customers will see right away that nobody speaks English at this establishment.

But MT is not a threat to more complicated, specialized kind of translation. The kind of translation that people are willing to pay money for in order to make money. Patent translation, for example. Both the Japan Patent Office (JPO) and the World Intellectual Property Office (WIPO) websites have an MT function. The JPO had it for many years, WIPO added it just recently. You can translate Chinese to English or French to Japanese with the Google Translate function from the interface on the WIPO website with a couple of clicks. But the translation will not make a whole lot of sense (see my post here). The machine translation function has been available on the JPO website for 10 years now. I think that it only brought more business to people like me because patents in foreign languages that would not be otherwise found were discovered with the MT function. And then they had to be translated. By a human. Sometime, by this human.

Regardless of what vendors of “machine translation solutions” are claiming, machine translation has not made a lot of progress in the last few decades.  Just ask my brother, he’ll tell you all about it. Machine translation software does not really translate. It is just a piece of software that replaces words in one language by words in another language according to an algorithm without any understanding of the meaning of these words. You can program an ATM machine to do what human tellers used to do a few years ago. There are really only a few dozen possible variants for simple bank transactions. But you cannot translate anything without understanding the meaning first.

So if you are wondering whether your job will go the way of Blockbuster one day thanks to Internet and machine translation, I would not worry too much if what you translate is important enough for people to pay good money for it now. They do it because they need a real translation to do whatever it is they are doing, not machine translation. Without a real translation, they would be losing money.

But if somebody pays you to translate something inconsequential, a blog, for instance, your business model  may be in trouble. Although I doubt that there are many people willing to pay for something like that. Generally speaking, if what you are translating is of little or no real value, it can be done by a machine because the quality of the final product does not really matter in such a case. But if your translation is a valuable product, machine translation is likely to result in more rather than less demand for your services.