TIOBE Is (Unintentionally) Misleading; in Truth, Interest in Java Is Surging

There's a lot to dislike about java, but the platform independence story of the JVM really is the best of anything out there that I've experienced. That doesn't necessarily mean Java anymore, Clojure, Scala, etc. all benefit from the work done to make the JVM truly platform independent. But when I compare it to my experiences with Node.js, ruby, or haskell, (dependencies not working because library maintainers forgetting (or don't care) that not every one uses their preferred development operating system), I am really grateful for all the boring, finicky work that contributors have put into making a truly cross platform development target.

This article might have some truthyness but the arguments would never lead me to that conclusion.

1. "wouldnt it be reasonable for searches for a mature language to decrease since devs are already very familiar with that language" - No, only if the language is also dead. Any actively developed language is going to always be expanding its libraries, frameworks, feature set, etc. and as a developer you constantly have to keep up.

2. Also saying total "java programming" searches quadrupled sounds good, but all other searches had a 5 times increase using his own data.

3. When I use his same method for "python programming" (http://www.google.com/trends/explore#q=python%20programming&...), I get an increase to 86 (2013) from 54 (January 2007) for python. This would mean compared to java, python has had an 8 times increase!

4. Basically unless you know exactly how google is normalizing the data, you are not going to have anything worth reporting on using actual numbers anyway.

Surely the same logic can be applied to the other languages?

Yes... the absolute number of Google searches for "Java Programming" is growing.. but the relative number of "Java Programming" to "* Programming" is shrinking.

Measurements of this kind without anything to compare against are just misleading IMO.

A flaw that makes the Tiobe index easy to game is it uses the maximum results from querying several search engines to calculate the ranking, not the sum.

That's how in Oct 2013 Groovy rose to #18 in the index from outside the top 50 only 5 months earlier (May 2013). 3 months later (Jan 2014), Groovy had dropped back out of the top 50 (#32 in Nov, #46 in Dec). According to Tiobe's Janssen [1]: "It turned out that the data that is produced by one of the Chinese sites that we track is interpreted incorrectly by our algorithms. After we had fixed this bug, Groovy lost much of its ratings."

Someone is also using a Chinese proxy to exploit Maven, boosting the apparent download figures for Groovy to target rankings like Redmonk [2]. Click on "country" and see where 85% of those downloads come from.

[1] http://www.infoworld.com/t/application-development/c-pulls-a...

[2] https://bintray.com/groovy/maven/groovy/view/statistics

Ohloh shows Java holding steady for the last several years at about 10% of all commit activity in the open source world.

http://www.ohloh.net/languages/java

I think there are a number of better ways to determine which programming languages are popular than what Tiobe uses. If they were only counting pages generated within a specific timeframe, perhaps a month or year, then their results might have some validity. Every month they are including all of the web pages that exist. I believe this skews their results.

I'm also not convinced that language popularity is an important metric. I think knowing which languages are strongly sought after by employers is more important and relevant. I think it is also important to know which languages are suitable for particular tasks, for people looking to create new code bases.

It would also be interesting to know how much code exists in various languages, and how many people are currently employed maintaining code bases in various languages.

The author supports his thesis by taking statistics which are adjusted for overall volume of google traffic, then de-adjusting them. That right there invalidates the results.

What do you mean unintentionally? TIOBE have always been bald-faced liars who know their methodology is broken, their data useless and who only publish clickbait to get ad impressions.