Russian words are usually longer, than the English ones. Does number of words differ a lot_

That may be part of it. I just did a little word counting but I'm still on a ratio of about 1 to 2.

hard to compare for me as i haven't seen any article, which was the same on both english and russian news site.
e.g. comparing tech news:
http://news.google.com/news?ned=ru_ru&topic=t
http://news.google.com/news?topic=t
all articles I see are different (you can compare it easily without russian knowledge by time released and pictures)

Russian is not longer then english, the reason may be different – in russian language there are less information and relevant articles, so for having the simillar page length, google shows longer part of the article

I think the new English<-->Russian translation is using our internal statistical machine translation technology.

If fact, Google have enough information to translate almost all the languages (or a big part), isn't it?

TOMHTML, that depends what you mean by "enough information". Sure, they have lots of samples of lots of languages. But do they have aligned translations – i.e. where they know that the source and target texts match? (Whilst Google might make it look easy, I don't think it's quite that simple to train a Statistical MT system to work for new language pairs.)

Russian <-> English translator is amazing. It thinks that Venezuela is Japan, for example. It sometimes translates one word wildly differently in different contexts. Most of the time it's not easy to understand the resulting text no matter the direction of translation.

It really looks like statistical translator. It does not do the usual mistakes of all russian/english translator engines while doing lots of very weird and non-obvious ones.

Generally, it's far from usable but very interesting.

Google is relying on major corpora and algorithms in a pure SMT approach that is antithetical to what progressive researchers are thinking about. For a look at the real future of SMT, check out the corporate site of Meaningful Machines. As Google moves up the n-gram model, they fail to recognize syntax and grammar in every ladder step. As a result, this will only ever be good enough and never great.

There was a rather interesting article about Google's situation in Russia in the international herald tribune on Monday.

http://www.nytimes.com/2006/12/18/technology/18google.html?em&ex=1166590800&en=8a2866fd625fbe47&ei=5087%0A

> it’s interesting to note that result snippets are over 3 times as long as the English ones presented at Google News USA
^ I don't see any difference now – maybe Google experimenting with something.

Google English<->Russian automatic translation is not impressive by now. My native language is Russian and know English (not very well). For me, it's more simple to understand original English text (if you know English at least a little), than understand _any_ automatic translator.

Interesting: I found very, very helpful Japanese/Chinese->English Google translator. I absolutely doesn't know any Japanese/Chinese hieroglyph, and surprisingly, for me, it's more easy to read the result of Japanese->English Google translation, than any English->Russian translation. Maybe I just don't see those horrid things Google translator done in English :-)

About nytime.com article:
I agree with author with almost all he wrote, but I think one fact is missed from entire picture: Google came to Russia too late, when all "early adopters" was binded to Yandex/Rambler/etc... When "ordinal users" came to Internet – they asked "early adopters" – and directed to local sites. In any way – now more people in Russia turns to Google.

The Russian<=>English translator has no ability to rearrage words that is essential for quality translation. Right now it just follows original order of words. The bottom line: it's far from usable.

Serge: At least I was able to understand a rather colloquial text. When Google did not know the word, it simply transscribed it, so I was able to figure out the meaning.

Bottom line for me: It was usable for me and I would like to see it getting better.

I never blame those who create new things and offer them for people. Google's translator is free, and we are free to use or not to use it, as well as other online translators. But I have to say that the new English-Russian translator is far below the level of the modern translation systems (such as www.freetranslator.com, www.online-translator.com, www.systransoft.com etc.). Serge and Mathias Schindler have right noticed that it cannot translate some words, and cannot rearrange words properly. For me, the most problem which strikes the eye is that Google has no ability to concord the words. Perhaps, for languages with little word inflexion this is not critical, but in Russian it's a real obstacle for understanding the text.