I don't know about the actual usage statistics for Eclipse and IntelliJ, but I very much agree with all the points he raises in the article:
1. Eclipse 4
Yes, this was the beginning of the end. 4.0 and 4.1 series of Eclipse were buggy as hell, unusable for me even for just core Java and PDE work. I only switched from 3.x to 4.x when Eclipse 4.2 was out, and even then it was bumpy. CSS/theming engine was a failure and massive waste of time, at least for IDE (dunno about RCP). See point 4.
Other e4 technologies were also of questionable value, IMO. Dependency injection? I've never used them in the IDEs I've development. Did it bring significant improvements to technical quality of the IDE?
There there was p2:
2. Equinox P2
Ok, this wasn't as bad as CSS/theming, it works fairly well nowadays (I think? Others might disagree). But it was still another massive investment and breaking change that cost adopters time and effort. Was it worth it in the end? Maybe, I'm not in a position to judge. But other more important aspects were forsaken in the wake of this.
4. Ugly as sin
True. Since Eclipse 4, I've always configured Eclipse to look like the classic theme. Including manually have to tinker with .e4css/
files to restore traditional style tabs. Nowadays, with Eclipse Neon, I
just have the theming engine completely disabled. Good riddance.
3. Not a product
True as well, in the sense that there is no proper product director/manager and product vision. This was mentioned before in this thread, see Doug's comments on "product focused organization", etc.