Due to the high number of regional languages, most startups simply start with the English version of their product (if the product is not primarily aimed at their home country). So Google in France would have been in English first and French second.

AFAIK, and IANAL, in France, you'd still need to support 2 versions from day one:

https://en.wikipedia.org/wiki/Toubon_Law

I just internationalised a UK startup to enter other EU markets, since doing so it has been a non-trivial amount of extra work just to keep it up. Not immense, but definitely a drag. There's all sorts of little extra cognitive overhead once a product supports two languages. Button widths, text lengths, layouts, email layouts, normalised date formats, even icons (for example we'd used a £ icon to indicate paid). We use SendWithUs templates, and now we have to upload all the translations too as part of a deploy. And that's not even counting the extra time to do the translation, upload it, etc.

Silly little things that a lot of programmers use without thinking, like basic date formatting like "ddd HH:mm", you simply can't do that any more. You need to agree shared formats and they don't always fit as you'd like them to, or quite convey the info you want them to. For example recently I'd have loved to say "Today 9:15", "Tomorrow 12:15", "Thu 14:15" or "Next Week", just for one specific place. But then you need to create a function to perform l10n on that and suddenly you're adding a big bit of extra functionality and you end up falling back on the shared .ToL10nDateTime() even though it's objectively worse at conveying the info you need to convey.

We've got a bunch of work in Alpha at the moment that's English only because we were under such time pressure to deliver it for one client. The time to insert all the place holders and do the client-side i18n meant we've parked it all to get the functionality out of the door for that client. Even though it will take more time in the long run.

As I understand it, a hypothetical French startup equivalent would be breaking the law by doing that.

Would it? 18 years ago? I doubt it. Google itself is really not as good in other languages as in English, even today. Could French google in 1999 have become world class in a non-native language?

That would never have worked.if you want to enter the French market you need a French product.

Have you been to much of Europe? English is spoken very well in much of Europe and, well, there's the UK and Ireland...

All but a handful of EU countries. Lived in DK, ES and BG. I can tell you that very few teenagers speak English well enough to learn to program so they can start a Facebook at 18 year.

When I was a European teenager in the 1980s, I met many peers who were great programmers - given their age (around 15, 16). They all spoke English very badly. They just weren't interested in English, only in programming languages. These were self-taught kids who were also good in math and science. Apparently you don't need to know much English to get good at programming.

I think the argument was that a non-English Facebook likely wouldn't have taken off the way it has.

Have you been to France ?

No it is not except for Scandinavia and just around Denmark. In reality the further south you go the more you really need to speak the local.

I have found it a lot easier to get by (talking to locals) in Tokyo using English than in for instance Madrid.

I can't comment on Tokyo but just to offer another point anecdote, I've found it pretty easy to talk to people recently in Portugal, Spain, Germany and Belgium.

Most communication starts with my apologetic expression for the cultural faux pas which is about to commited.

I then move on to a token effort at mangling the local language which finally ends with me being asked to repeat in English by a patient native.

That's just not accurate at all.

It's even less accurate again if you're moving in business/tech circles and not travelling around as a tourist.

No, if you go as a tourist to tourist locations you have better chances. But as a local when you take care of daily things you'll need to speak the local, and that's a given. And really as far as recruiting talent between EU countries goes, the language barrier is real between countries that do not speak officially same language.

I would not consider the language barrier as the largest obstacle. But the European scientific infrastructure can't even compete with Californias institutions and it gets much worse if you count in the whole US scientific infrastructure. Most of the world leading scientific institutions like Standford, Berkley, CalTech etc. are located in an area where scientific knowledge can get appropriate funding for a product. You simply cannot innovate or create something new without extensive research from universities or research labs. This is one of the problems.

The other problem is, in Germany at least, that the general population is not that well educated in computer science, since the educational curriculum is organized in a strict federal way. The German constitution even prohibits the state from cooperating with the federal education departments. CS should now be a major subject on the same level with math and physics, yet this kind of system makes it very hard to implement a curriculum for all of our pupils. This causes, that even a lot of young people are not that interested in CS and lack of basic CS knowledge.

This is a bit tangential and only based on recent events, but the US immigration uncertainty is probably going to be a factor in sustaining the ecosystem..

It seems to be much harder for European services to be "interesting" in the USA. There have been local counterparts to e.g. facebook before Facebook was a thing here. Still, we all moved on to FB later. Why? Not sure... Suddenly you find more and more of your friends in Facebook than in StudiVZ, more business partners in LinkedIn than Xing and so on. Many other companies just moved to the US, for example Unity.

I don't really think it's necessarily a language thing.

Unity's an interesting case, because they officially moved, but it was only the marketing and business side that moved, while programmers all stayed in Copenhagen. So this gave the impression of being an SF-based company, but SF was purely a biz office. More recently, the programming is actually being slowly moved out of Copenhagen too, for cost reasons, but to eastern Europe, not the US.

Right, seems many companies like that "SF/SV/whatever"-company label. That's sad. Goes so far that we have really small, local startups doing pretty impressive stuff... With a more or less empty "office" in SF and then in the hypish articles it's already an SF-based company.

Just because it sounds more impressive than "based in Rübenwald, Austria" ;). Patriotism isn't very strong either in German-speaking countries since WWII.

Would it really be hard to transform a search engine tech developed for french or italian to english ?