Yeah that was a mistake. We were trying to cut out redundant questions, but it would have been so easy just to split this into three answers. We might be able to split this out based on secondary information, but we didn't finish that analysis in time for this post. Maybe we can get it into a future post if we trust our result.
If the survey was from a random sample of all programmers, the usage level would be much lower.
How do you know what usage level is a "problem" without any point of comparison?
I'm interested in JavaScript, for example, but I don't use it more often than a handful of times per year, if that. I wouldn't be surprised if a third of people answering a JS survey were in the same boat. And I'm not worried about the future of JS.
> If the survey was from a random sample of all programmers, the usage level would be much lower.
Is there any language which more than a small minority of all programmers in the world use?
This may or may not counter the point. Your language specifically... per various comments I've read here (esp vs Go language)... targets programmers demanding high-efficiency or low-level usage that are basically using C, C++, Objective-C, Java, or C#. Vast majority are with most in efficiency camp using first two. So, if data self-selected for those, then a third saying No could be a serious failure if the goal was replacing those with significant uptake. At the least, it would be meaningful metric on if Rust was succeeding.
Unfortunately, we need the people responding to represent those groups to be sure. I suggest that whoever has the responses filter out everything but those languages above. Then look at the No's or any problems reported. Then, filter again to get just C, C++, and Objective-C as they're the low-level ones. Then look at the negative responses again. That could get information on uptake in most important targets plus generate ideas for action to take for key demographic.
Second, think about C++. A year after it came out, were 90% (say) of C users who had heard of C++ using it? I highly doubt it.
There simply isn't enough data to compare to. So you can look at that number and support any narrative you want by making up what you think a "healthy" number "should" be.
Or it could just reflect strong interest for a relatively young language. I'd love to write some Rust in my day job, but it's rough getting support for using new toolchains and package managers (i.e., cargo). Relatively modest changes (C++11, CMake) are ambitious.
Anyway, if a suitable greenfield project or even a job offer surfaced, I'd be extremely interested.