Hacker News Reader

Comment by fabiospampinato | Hacker News Reader

I don't know why people put any stock in this index, the methodology seems flawed. They say that Ada's rating is double that of TypeScript, surely that doesn't reflect reality in the slightest?

View on HN

Tons of folks in industry use Ada, they just don't talk about it much (or are forbidden). Ada is the darling of USDOD, so there's untold LoC in missiles, fighter jets, and spy satellites.

Yup i did some Ada work in a lockheed internship in college. Was for various things on a plane being sold to another country. Not classified but plenty of things around me were (of course nobody gave me any classified information but everybody walked around with their clearance level on their employee badge and the broad strokes of what other programs were isn’t classified, for example lots of F-35 work was being done in the building)

Isn't the F-35 also famous for the fact that they used C++ (instead of Ada) for the software?

Based on what I understand of the rating methodology, though, I'm still very surprised that Ada beats TypeScript. My understanding is that it's basically a measure of search results, and I would expect TypeScript to have vastly more of those than Ada, given that most Ada applications aren't going to be publicly discussed.

I suspect that it has to do with names—Ada is highly overloaded in ways that are adjacent to programming (ADA, Ada Lovelace) and is a short enough word that Google may ignore it. A naive search for `ada programming` turns up 896 million results, compared to 28 million for `typescript programming`

Exactly, Ada is searched for 2x as much as TS? No way.

Although my intuition would seem to agree with you, Ada is still heavily used in industries where programming errors are fatal like avionics, medical devices, military, space, automotive. None of these industries typically like to operate in public view so I believe there is a reasonable argument that Ada may even be underrepresented in most indexes.

TIOBE has always been flawed; it doesn't claim to be perfect. It's an interesting metric, and deeply unpopular languages don't make the cut. Once-popular Pascal has been plummeting in recent years. That said, their chart has a particularly unsettling feature: the SQL line is dead flat from early 2004 to 2018.

Yeah...I think you'd expect it to grow a good bit during that time frame, unless it's heavily normalized somehow.

I don't know what is the methodology behind the TIOBE rating but what is clear is that a lot of stuff happens below the radar of the average silicon valley dev.

Most software is not user facing.

The TIOBE index is effectively just based on the number of estimated hits for "$LANGUAGE programming" on Google. That's it. It's not based on actual usage, or interest, and certainly not importance; it's just counting search results.

(They actually search on a few sites other than Google, but they're bizarrely chosen -- they include search hits on Walmart and Etsy, for example. And the way they weight search results effectively means that the wildly inflated Google search result counts dominate the ranking.)

Reply on news.ycombinator.com