And then he somehow tries to make it "better" by ripping on himself, too, saying he's a part of the problem. Um, no, being self-deprecating in the same way that you're insulting everyone else does not magically make it ok for you to insult everyone else.
I've been using Linux (and a couple UNIXes on and off) for a little over 10 years. So I can get around a UNIX-like system pretty well. A lot of things are easy, and a lot of things aren't. Saying that it's somehow someone's fault is ridiculous. Claiming that all software developers are collectively lazy or don't care about user experience just doesn't hold up.
The funny thing is that he works in a position that naturally involves some difficult stuff. Let's say my favorite language to write software in is called XYZ. Say it's super easy, intuitive, concise, performant, and the method for compiling/deploying/distributing the end result of your hard work is trivial. In all ways, this system is just beautiful to work with.
Great, but I'll bet you the guy who wrote all the development tools and runtime for XYZ had to do a lot of difficult work to make that possible. Dahl is building a runtime for web applications. Unless he's writing it in some high-level language, it's not going to be easy. Supporting every platform he wants to support isn't going to be easy. User interfaces should be as simple as they can be, but often that requires a lot of complexity under the hood.
Go down even farther. Let's think about our basic building blocks. Transistors. Hgih and low, ones and zeroes. It's a very simple interface. You construct logical operations by using NAND, NOR, NOT, etc. gates, which are built from transistors. Also simple. But the next step for our modern computer is... well... the microprocessor. And while it's made up of these incredibly simple building blocks, the combination of them is extraordinarily complex. So the interface into that mess is also not the most friendly thing to work with: a machine instruction set. So we build things on top of that to make it successively easier: assembly language, C, Ruby.
And the tools that come along with this are only as good as the technologies they're built on. Tradeoffs must be made to be portable. Yes, all this is a huge mess that "we" have collectively invented over the past 30-50 years or so, but it's simply not possible to go back to the 1970s, know exactly where we're going to be in the 2010s, and design the perfect system, even with foreknowledge. The current state of computing is a product of the evolution of our technology. Often that means doing the best you can today, and hoping for something better tomorrow.