Debunked: Certain languages lead to buggier code

https://www.theregister.co.uk/2019/01/30/programming_bugs/

Long and short of it, code in certain languages has fewer bugs, but that’s probably down to the specific group of people who normally use that specific language being trained in particular mindset.

They mention Python, which gets used all over and was one of the items pushed with the Pi in schools IIRC. It was probably chosen for that as something a child could handle and produce “something that works” as opposed to a neat tidy and efficient programme that doesn’t go A-over-T on 13th January when it worked perfectly on the 1st through to the 12th or take 100MB of memory to blink an LED.

Like many say about Windows, if you make a system so simple even an idiot can use it then don’t be surprised when lots of the users are.

Their example of Haskell is probably of one where it’s fairly opaque to the untrained potential user so the ones who do use it are those trained into the mindset I spoke of earlier.

Which just proves that people with proper training are less likely to make the mistakes routinely made by the untrained people.

Most amusing is the idea (wrt the Haskell case) that someone with more academic training would put less bugs into any code they write.

I think it is very hard to come up with a study design that would allow to meaningfully extract the influence of using specific languages on the amount of bugs. Same for any other metric really – time-to-completion, maintainability, performance, complexity…IT is notoriously bad at even defining meaningful metrics for such quality indicators, which would be the first step.

And that is even ignoring the problem that one bug is not like the other bug. Data loss or UI glitch?

Most amusing is the idea (wrt the Haskell case) that someone with more academic training would put less bugs into any code they write.

I don’t see their correlation of academic box ticking to any proficiency in coding either. However, they are more likely to have been trained to the mindset required.

I think it is very hard to come up with a study design that would allow to meaningfully extract the influence of using specific languages on the amount of bugs.

I think that if the same methodical person wrote code in each of the languages under consideration they would probably produce fewer bugs in each case than a less methodical (sloppy) person.
Only if you have equal numbers of methodical and sloppy programmers using each language could you start to get anything remotely meaningful in the way of statistics. Unfortunately the sample size of people using all the languages would probably be too small to be statistically valid.

IT is notoriously bad at even defining meaningful metrics for such quality indicators,

IT is notorious for having people who speak in a foreign tongue as far as non-IT people are concerned.
Late completion is often the result of the non-IT person realising that what is being produced doesn’t match their “vision”¹ and changing the specification several times.

¹ The fact that most people doing the “vision” speak in management pseudo-language bollox that even they don’t understand (but they heard someone else say it) so the “vision” never gets through first (or even fifth) time may have a bearing on the “failure”