Or maybe one could reasonably expect that Google, with all that money and engineering power, would implement something like a dictionary lookup to short-circuit single word translations?

Edit: I did not want to complain about the translation quality nor show my sense of entitlement. You have authoritatively stated and used a some jargon to tell us it doesn't use English when translating between languages, so I simply wanted to understand more about this process because, on the surface, it looks like English was involved in my specific example.

Don't you think that if it would be that easy they would've already done it?

I don't say it's easy, at least from organizational/product-development perspective. But that's basically how every translation website before GT worked: type a word in language A, select language B, get the result(s) from a dictionary, done. In the nineties I had this pocket device: https://i.imgur.com/Yp9yTRL.png which would look up and translate words in 8 languages. Today I have a smartphone which can connect to a datacenter with enormous computing power, but a dictionary lookup is a hard problem?

I think the problem is cultural, not one of engineering.

I can't say I can point to any sort of evidence of this, so take it with a pinch of salt, but my gut feeling is that the people in Google who are responsible for GT are somewhat fanatical about making purely statistical machine translation (and, in particular, deep-learning machine translation, a.k.a. neural machine translation) work. They are most probably perfectly well aware that it is possible to improve their system by using some rule-based fall-backs, or some background knowledge, but they have decided that they dont' want to use those and instead want to do machine translation from scratch and from data only, without any expert input.

So it sucks, like all pure systems suck.

There are plenty of dictionary lookup apps, if that’s what you want. It’s just not what GT is.

As a developer I kind of get it: "we have all this machinery so we won't compromise to use the simple route even if it makes sense short-term; let's improve our product instead". But as a user I'd expect Google to be at least as good as dictionary lookup from two decades ago, and do PL -> RU or PL -> DE instead of PL -> EN -> RU, PL -> EN -> DE when I type in a single word.

Google Translate simply isn’t meant to be used for dictionary lookup, though. If you use it for that then you’re using the wrong tool.

This is like complaining that Word isn’t as good as vi, which is decades old, for text editing. Sure they could bolt a text editor onto Word if they wanted to, but why bother? That’s not what people are looking for when they use the product. And if that’s what you want, there are many other choices that are much better.

or maybe single word lookups are pretty useless for most translations? How would you identify those words which can be directly looked up in a dictionary vs those that require the current context for translation?

> How would you identify those words which can be directly looked up in a dictionary vs those that require the current context

I would start by determining the number of words in the query. If it's equal to 1, the user probably wants to translate a single word, maybe give him the 1:1 dictionary result. If there's more, the other words possibly provide context, so let's use the sophisticated thing.

Polish PM would probably disagree it's better to have imperfect auto-translation than none at all ;)

https://www.bloomberg.com/news/articles/2018-02-02/translati...

Dissatisfaction with the status quo is precisely what keeps automatic translation companies in business, and indeed drives all innovation. The author explicitly gives credit to GT for being good enough for lots of people most of the time, but that does not excuse its fundamental failures.

GT Engineers don't want a pat on the back. They want to do better.