I get this same thing from the Delphi users forum, where the diehards strain to convince themselves that Delphi has "three million users" and "is about as popular as Python". They insist that there's no reason Stack Overflow questions should be correlated with real-world usage. One suggested, seriously, that perhaps Delphi is just so easy to use that users don't need to ask questions. This excuse has been adopted to explain a lack of commercially published books ("maybe everything to say about Delphi has already been said") and there was even an attempt to explain away job posting data. In that case it was theorized that Delphi jobs are filled as soon as they're posted while Java jobs go unfilled for months so when you see thousands of Java jobs on Dice but no Delphi jobs that actually means there's a huge Delphi jobs market and no one's actually using Java. :-(
Never underestimate how far users of "sunset technologies" will go to convince themselves that their skill set is outdated. Heck, Embarcadero's C++ product manager insists that "customers tell him" that Delphi (Pascal!) is "five times more productive than Python" and he actually believes this too. The Delphi product manager, meanwhile, told me he sincerely believes that Delphi has had more of an impact on the business world than Python ever has. Meanwhile, the former VP of Developer Relations once polled the first 500 people to upgrade to their newest release of Delphi, crunched the data, and concluded that users really love Delphi (of course, the many who felt the upgrade was light on features and high on price and chose not to upgrade weren't reflected in the data).
Did you actually read the article? Of course generating more questions is likely indicative of more usage. But the graph shows it from being in last place (of the listed languages) in 2012 with ~4% of the question views to being first place right now over all other languages with ~10% of question views. Python has grown a lot over the past 5 and especially 10 years in wide usage but I've seen no other evidence that the chart in the article should be believed to have a direct correlation with usage. There are many possible explanations of a more indirect correlation that could be skewing the numbers in unexpected ways but none are discussed in the article.
I'm the biggest python fan you'll meet (ok I'm sure that's not strictly true) but I followed the headline hoping to see more concrete data to support the claim of "Incredible Growth" and came away disappointed.
I did, but I also read the comment: "While this is an interesting statistic there is no additional analysis that it correlates to python's usage growth" -- which is what I responded to.
>There are many possible explanations of a more indirect correlation that could be skewing the numbers in unexpected ways but none are discussed in the article.
Like what though? Aside from some kind of bias on SO I can't think of any alternative explanation that actual growth.
Now, what that growth was based on, that can have many explanations, sure.
I completely agree that there is growth but the question is how much. To me it feels like the 2012 python statistic is unexpectedly small. It's not like python wasn't very popular in 2012.
To put it another way: do you believe that the first graph in the article directly represents programming language usage percentiles, in the field, over five years? I'm guessing not. I certainly don't.
Here is a graph of programming language popularity on github from 2012-2014:
While this metric is arguably just as fuzzy as any other, the lines in the this graph definitely seem to agree with my gut feeling for language percentiles during that period. I'm not trying to say that my "gut feeling" has any significance. I'm just re-emphasizing that the article didn't give me any useful insight into the significance or meaning of the Stack Overflow statistics. I couldn't find an equivalent chart for the years since 2014 but would love to see one. Either way, it's not like python was ever far behind. It was basically in third place in both the beginning and end of that time period. But in the beginning of the SO chart it's in sixth place. And the SO chart doesn't even include ruby, which is ahead of python in 2012 in the loggly chart. I guess people really don't go to SO for ruby questions? It's certainly possible that the quality of answers on SO for different languages has a lot of variance, which is one possible explanation for the difference. Maybe python questions in 2012 on SO weren't that good. In my experience I'm far more likely to check the python docs before I go looking on SO. Maybe there were changes in google rankings for certain things. Maybe there were certain types of articles that took up the lion's share of growth, like "how to port to python 3", which could have brought existing python devs back into heavy view rotation. Python has also gotten a lot more complicated with things like asyncio, etc., which could again draw more article views but not be directly related to usage growth.
You believe that Python would ever generate more questions over time on SO without getting more users/usage?
I guess the mysterious dumbing down of a stable user base is an alternative explanation.