TIOBE Index prepares for ranking engine shutdown, plans to include new sources

Tiobe index february 2022

The TIOBE Programming Community Index for February 2022 is out, and while nothing much has changed on the surface, a closer look reveals a tech-induced boost amongst the top-rated languages.

Compared to last month, numbers for Python, C, and Java have gone up by more than one percent, catapulting Python from 13.58% to 15.33%, while C and Java went from 12.44% to 14.08%, and from 10.66% to 12.13% respectively. A bit further along the list, PHP made its way back into the Top 10, pushing Swift out and settling back in at #8, and Objective C crawled all the way from 24 to 18.

Although some fluctuations are normal, jumps like that have to leave the casual observer suspicious, given that the month to month differences seldomly cross the 1% threshold, and an upward push amongst all of the Top 3 seems even rarer.

The reason for the sudden change is however relatively clear cut and has to do with the upcoming retirement of Amazon’s web traffic analysis service Alexa.com. Until recently, Alexa was used to select the 25 highest ranked search engines for the calculation of the TIOBE PCI, which controversially boils down to counting the hits when querying +”<language> programming”.

With May and the associated shutdown coming up, the company had to therefore start looking into alternatives. The February index is the result of those investigations, being the first edition to base calculations on Similarweb’s rankings as an Alexa replacement. 

While this doesn’t sound too exciting, the swap has the potential to improve the index a good bit as it goes hand in hand with a technology shift in the company’s backend. According to TIOBE Software CEO Paul Jansen, changing to Similarweb meant switching over to GUI-less browser HtmlUnit for web crawling as well – a tool that will allow the company to consider popular developer resources like Stack Overflow and GitHub for its index in the near future. 

Currently, the list of search engines consulted in the calculation process largely comprises various localised Google offerings, Amazon, Wikipedia, and Chinese webportal QQ (previous sources Sohu, Ebay, Indeed and Flipkart weren’t considered in the February list since they have to be integrated into the new process first). Adding more relevant sources could help the index’s credibility, as common points of criticism include results being somewhat removed from what developers are actually interested in at a given time, and comparatively easy to game, which might get harder with the new input.

It will be interesting to see if upcoming further modifications will lead to some bigger ranking changes in the next few months and maybe align the index a bit closer to the results of complementary offerings such as the RedMonk programming language ranking (which only takes GitHub and Stack Overflow rankings into account) or PYPL (which checks the popularity of tutorials for a number of languages). Lists like these should always be taken with a grain of salt, as all calculations have their advantages and drawbacks. However if compared and used sensibly, they can surely help to spot trends or serve as pointers when making learning or technical decisions.