Here we go again: TIOBE and Delphi
TIOBE has had another dig at Delphi
Every so often, the TIOBE Index likes to have a raid on Delphi and, as usual, there was a lot of reactions, which is probably what they were aiming for. In other words, they wanted the clicks and they absolutely got them. But I digress. As others in the community have already pointed out, TIOBE isn't reliable in any way, shape or form if not merely because there was a time when they didn't even include the word "Delphi" in their online queries. This is why my analysis of what TIOBE did stops here. Kim Madsen has already published a follow up about what Jim and Jon said: everything there was to say has already been and I couldn't add anything more interesting on this specific topic even if I tried, mostly because I think TIOBE isn't worth my time.
Statistics
This is however a great occasion to talk about statistics: what it is, how it works, etc. The fundamental feature of statistics is that statistical analysis can be done mostly after the fact. It is true that you can also predict, but you still have to wait for the constraints to happen in order to test your prediction. In other words, if you predict that in 2040 there will be mostly flying cars, we have then to wait for 2040 to see if there are mostly flying cars. Statistics at the end of the day is a snapshot of the current state of affairs, even in prediction, because it can only be based on historical data and that points forward to the present moment and stops there, in terms of how far data go.
This, by the by, is exactly the position the companies paying TIOBE for their historical data are in: all they need to know is whether, for example, JavaScript will take over the world even more than it does now. That is the kind of information they're after. The popularity contest, to them, doesn't really matter. And it doesn't matter to TIOBE either. A few % differences does not affect their business model one yota and therefore they are probably not going to fix it anyway. The reason for this is that all they need is to know who's in the first #3. I doubt Delphi, even with the addition of FPC, is in the first 3 even when we take into account the 6 million developers using Delphi (which is a number that has been floated by the way).
领英推荐
It's not always so, however
There are niches where precision matters to a near-manic point. Take for example intelligence prediction: that needs to be based on data and experience, but it's also adjusted much more finely than TIOBE does. Particle physics is another: surely you want to make sure that your Higgs Boson is actually there, right?
TIOBE doesn't matter to you
TIOBE does not, or at least should not, matter to you as a developer. Not even as a Delphi developer, merely as a developer. TIOBE isn't in the same industry as you and they don't make a living the same way you do. Stop worrying because there is nothing interesting to see there for you.