Europe is to some extent a special case - even outside of software, English is the pan-European lingua franca; lots of college-educated people know a lot of it even before getting into tech. In places like China, or Central Asia, or the Middle East, this is not as much the case, and so picking up English skills add a further barrier to entry to the field.

(Europe is only a special case "to some extent" because English also competes for lingua franca status with Hindi in India; in Japan it's taught (badly) as a result of connections with the US; and in countries as varied as Israel and Singapore it has a foothold.)

If you're starting from first principles, and are willing to go with that common-second-language threshold instead of insisting on people's first languages, English is still definitely a clear first choice - it barely edges out Chinese for total L1 + L2 speakers, and those speakers are concentrated in the centers of software development.

After that you have Chinese and Hindi (both fairly restricted in geographical reach, but involving lots of people), then Spanish (most of Latin America) and French (spoken in a dizzying array of cultures as L2, because of colonial history in Africa and the Middle East).

And then you finally get to Arabic :-P

As discussed elsewhere in the thread, the few dozen keywords are a different matter (and are just memorized to mean "make the computer do ..."), but having useful error messages from your compiler helps get stuff sorted out more quickly.

There are, to this day, people who get in trouble when their computers reply to them in English, including coding folks. I bet they were (and are) thankful for initiatives like yours, but they won't have the vocabulary to tell you.

> having useful error messages from your compiler helps get stuff sorted out more quickly.

The definition of useful varies though. Many people starting out don't have the ability to understand some error messages, so being able to search for the error message is most useful. If everyone in the world has the same error message, finding an explanation for the error is a lot easier.

> finding an explanation for the error is a lot easier.

Two options:

1. LC_ALL=C $command to get the universal version of the error to search/report

2. Add a numerical error code that remains stable across locales

LC_ALL=C changes all the locale-related behaviors. I would try LC_MESSAGES first.

The tech community greatly benefits from settling on a lingua franca. Imagine if everyone posted on StackExchange using their native tongue.

> The tech community greatly benefits from settling on a lingua franca

Maybe we see that in a few years, with everything moving to Chinese.

> Imagine if everyone posted on StackExchange using their native tongue.

Imagine users going to forums in their own language. People actually do that.

People have been talking about "everything will be moving to Chinese in the next few years" for the last ~20 years.

I'm well aware. But China is also the only country (or even bloc) in the world that's aiming for a fully internal supply chain for high tech. Not sure you can transform basic materials into a laptop (to pick one simple example) at scale without China anymore, but China is on a pretty successful track to do that without anybody else.

Once they get there, I'm not sure they have a lot of motivation left to do high-tech on foreigners' terms and in foreign languages.

While it doesn't follow that "we" need to learn Chinese then, whatever they'll use instead of StackExchange will simply outnumber the English version.

For hardware, perhaps, but the Chinese have no realistic way to replace the rest of the world for software. If they want to sell their hardware to the rest of the world, English is the only realistic language to use for that. I see no evidence of mass learning of Chinese anywhere.

The replacement for StackExchange already exists, on WeChat. More for hardware than for software.

Probably the Chinese will embrace and extend the English language, the way the US did.

The US did not embrace english, the US were a set of english colonists who managed to absorb lots of foreigners with the clear contract that not adapting to their culture/language would result in bad outcomes.

> having useful error messages from your compiler helps get stuff sorted out more quickly.

We're still struggling with this in english. Think about how often the default reaction still is to google a compiler error, unless it's one of a handful of common ones you see often. Compiler errors are often opaque, even to the best programmers that speak english natively. A significant base of software emits error numbers with thin descriptions at best, which then have to be looked up in a manual anyway.

Localized error messages are detrimental to solving the problem as googling the error message in a different language will prevent you from finding a solution in the main language.

In compilers it's not uncommon to output an error code in addition to the human readable message. That allows you to google independent of language.

Of course if you turn on localized error messages you might actually prefer results in your own language.

This sounds like it's coming from someone who has never had to help relatives fix their broken Windows PCs localized in a language spoken by only a few million people.

I'd love to code in an all Swedish environment. But when the I18n is incomplete or poorly done, all English wins out. Better if 100% of the compiler messages are in English than a 50/50 mix of Swedish and English messages because then you have to context switch between the languages all the time.