It’s 2023 (we made it!) and after becoming a member of Stack Overflow in September 2022, considered one of my first duties as a senior analysis analyst was to tug collectively statistics for our year-end wrap-up, and to which the pure follow-up query was requested of me, “is that this what we anticipated to see?”
I didn’t know, so I dug into two of Stack Overflow’s distinctive information sources: the annual Developer Survey outcomes and stackoverflow.com’s web site information. For analysis, marrying qualitative and quantitative sources is vital with a purpose to validate assumptions and discover the story within the grey space between express and implicit habits.
The 2022 Developer Survey collected responses from Stack Overflow customers all over the world to search out out what programming languages and software program growth instruments are the preferred. And since we’ve been doing this survey for 10+ years, we are able to see traits in rising (or declining) recognition. We will then use our web site information to validate the survey sentiment by taking a look at what customers ask about most.
On this article, we’ll check out what the latest previous tells us about what builders might be loving and/or questioning in 2023.
First, I’ll have a look at what proxies we might use to quantify programming language recognition. Then, I’ll evaluate this to traits for questions posted about programming languages, utilizing a easy regression evaluation with a purpose to elucidate and clarify attainable relationships between acknowledged recognition and questions requested on Stack Overflow.
Hottest and in keeping with whom
No supply of knowledge is best at tapping into developer sentiment than our personal Developer Survey. What languages did the developer neighborhood inform us they liked in 2022?
Within the survey outcomes for Most Beloved, we categorize every part so it’s simpler to match like-to-like (i.e. languages vs. frameworks vs. libraries, and so forth.). I’m going to take a cue from the survey and deal with programming languages for this query; drawing comparisons inside sorts is sensible and avoids introducing one other layer of complexity.
Rust, Elixir, Clojure, Typescript, and Julia are on the high of the record of Most Beloved Programming Languages. Nevertheless, in wanting on the final three years, we see a little bit of motion.
Most Beloved Rank in Developer Survey
In 2022, we added a drill-down to particularly present recognition amongst these studying to code. As a result of Stack Overflow is a studying useful resource, I’d count on that recognition amongst these particularly studying can be a superb indicator of present and future programming language recognition.
There’s an attention-grabbing sample in evaluating Most Beloved and Studying to Code Reputation: folks studying to code aren’t utilizing essentially the most liked languages. The distinction between these two measures of recognition might be essential in distinguishing each as attainable explanatory variables for traits in query posts. Lower than 1% of these studying responded they had been utilizing both Clojure or Elixir:
- 1.2% are utilizing Julia
- 7.1% are utilizing Rust
- and 15.1% are utilizing Typescript.
How else would possibly we arrange expectations for traits amongst the various programming languages being requested about on Stack Overflow? I discovered two good sources which can be worthy proxies for recognition: Google and GitHub.
For internet searches, I’m utilizing the already established PYPL index, which is an aggregated supply for Google Developments information particularly for programming language tutorial search historical past. From this dataset, we’ll deal with annual traits in programming languages share of search.
GitHub publishes statistics on public repositories for anybody to make use of as a useful public dataset inside Google BigQuery, and though we lose the knowledge from non-public repositories, we are able to assume the general public accounts converse extra on to recognition as they’re tied to studying initiatives, portfolios, and open-source collaboration, that are largely self-directed somewhat than mandated by current enterprise guidelines. From this dataset, we’ll deal with the annual development in public repo pull requests by language.
Wanting on the fundamental relationship between Most Beloved % and annual rank in questions requested, we see a slight relationship over time, however not a powerful one. The easy regression right here exhibits 2022 has the strongest correlation within the final three years and that solely 7% of the variation in rating for 2022 questions requested will be defined by 2022 Dev Survey outcomes for many liked programming languages.
This graph exhibits that being liked (through the Developer Survey) just isn’t associated to producing extra questions on Stack Overflow. And this is sensible: posting questions most definitely speaks to friction with coding, a friction which will result in loving a programming language much less.
After we add in our further proxy variables for language recognition, utilization share amongst these studying to code within the 2022 Developer Survey, the development in PYPL from 2021 to 2022, and the development in Github pull requests from 2021 to 2022, we get higher regression outcomes. Utilizing simply Studying to Code Reputation will get us a greater regression that explains 67% of the variation in rating for 2022 questions. A logical conclusion right here is that Stack Overflow questions are extra inclined to the preferences of these utilizing the location as a studying instrument somewhat than these of extra superior builders.
Including within the different recognition proxies and liked share beneficial properties us further regression energy (75% variation defined!) and we’ve landed on our ultimate reply: traits within the variety of questions posted a couple of programming language on Stack Overflow will be defined by what extra builders studying to code are utilizing (most importantly of all elements) together with Google search traits, GitHub public pull requests, and the Developer Survey Most Beloved share (much less considerably of all elements). Our newest Developer Survey confirmed us that ~32% of programmers have been professionally coding for 4 years or much less, a big quantity of people who find themselves most definitely concerned in studying programming languages. That’s, beginner-friendly languages get essentially the most questions and recognition, however the Most Beloved languages make veteran builders joyful.
A peek into the final three years
Let’s have a look at the highest tags from questions requested in 2022 and the way they line up with what we might count on from the regression mannequin above.
We counted the variety of questions related to every distinctive tag; every can have as much as 5 tags, so questions will get counted greater than as soon as. Python and JavaScript are solidly positioned of their respective high spots, Reactjs and Java present competing query counts beginning in direction of the tip of 2021, and finally Reactjs takes the lead with constantly extra questions tagged in 2022. HTML and C# swap spots month-to-month in 2021, although C# strikes forward in 2022 with constantly extra questions. Within the decrease ranks, Pandas sees three years of development, R will increase rank in 2021 and holds, nodejs breaks into the highest 10 in 2021 and holds, whereas each PHP and C++ decline.
In comparison with our studying to code recognition metric, Python, JavaScript, and Java are inline with expectations being on the high of each lists. Based on the identical metric, we might count on extra questions on SQL and PHP. This exhibits that there’s extra to the development than simply measurable recognition. The elements that lead up to looking for tutorial on Google or GitHub pull requests from public repositories, for instance, might be influenced by content material creation within the tutorial and on-line studying areas, which in flip might be influenced by nuances in tech evolution (e.g. Python-3.x as a short-lived top-ranked tag earlier than it turned the usual model). The subsequent Developer Survey would be the canary within the coalmine illuminating any modifications in expectations for the forms of questions being requested on Stack Overflow.
