Let me see if I understand this: The argument is that a hypothetical language descending from Haskell will be (a) superior and (b) more popular than Lisp as we know it today and to any Lisp that might evolve from it.
I think there's too much hand-waving going on to agree or disagree with the premise. Language popularity is notoriously fickle and driven by considerations that don't even earn a footnote in the OP.
For example, one of the most popular languages in the world today is Javascript, a language that owes a little of its heritage to Lisp and none at all to Haskell. Why is it eating Lisp's lunch and Haskell's lunch today? Because it ships as the default runtime for the most popular application in the world, the web browser.
Another language that has the same breadth of reach is PHP. Again, its popularity is due to the fact that it is a default runtime for most of the world's hosted web server platforms.
If we are to guess which language is going to gorge itself on everyone else's lunches in the future, I would suggest we spend our time thinking about which platform is going to be insanely popular and what new language might be its default runtime.
I'm not providing any fresh insight, of course:
Let's start by acknowledging one external factor that does affect the popularity of a programming language. To become popular, a programming language has to be the scripting language of a popular system. Fortran and Cobol were the scripting languages of early IBM mainframes. C was the scripting language of Unix, and so, later, was Perl. Tcl is the scripting language of Tk. Java and Javascript are intended to be the scripting languages of web browsers.
But there is one point worth underlining a bit, which is that you can't start with (traditional) Lisp and then build in something that can only be built based on total, guaranteed restriction, because Lisp always lets you out. Haskell's STM is probably the best example; so far I don't think anyone has built a successful imperative STM because if you let people out even a little bit the STM abstraction tends to break. You have to be restricted for that to work in practice, even if in theory you could load that all onto the programmer; the constraints are too hard for a programmer to just implement, let alone a team.
You can write macros in Lisp to implement these restrictions, of course, but they will be hard to make water-tight, and then you will be left unable to use anybody else's libraries because they will not honor these restrictions.
If there is anything Lisp can't do, any language in moderately common use that it is not strictly superior to, it is Haskell. You can not just turn Lisp into Haskell with a few well-chosen macros, or even a lot of macros, without making it Not Lisp Anymore.
It may have taken a while, but we finally have a modestly practical language that is not merely Lisp warmed over or Lisp horrifically subsetted and bastardized; it's genuinely something different.
It could happen sooner: If there are enough posts like this, someone may be developing a programmable platform of some sort and think "I know! I'll borrow some ideas from Haskell for the DSL!!"
I think there's too much hand-waving going on to agree or disagree with the premise. Language popularity is notoriously fickle and driven by considerations that don't even earn a footnote in the OP.
For example, one of the most popular languages in the world today is Javascript, a language that owes a little of its heritage to Lisp and none at all to Haskell. Why is it eating Lisp's lunch and Haskell's lunch today? Because it ships as the default runtime for the most popular application in the world, the web browser.
Another language that has the same breadth of reach is PHP. Again, its popularity is due to the fact that it is a default runtime for most of the world's hosted web server platforms.
If we are to guess which language is going to gorge itself on everyone else's lunches in the future, I would suggest we spend our time thinking about which platform is going to be insanely popular and what new language might be its default runtime.
I'm not providing any fresh insight, of course:
Let's start by acknowledging one external factor that does affect the popularity of a programming language. To become popular, a programming language has to be the scripting language of a popular system. Fortran and Cobol were the scripting languages of early IBM mainframes. C was the scripting language of Unix, and so, later, was Perl. Tcl is the scripting language of Tk. Java and Javascript are intended to be the scripting languages of web browsers.
http://www.paulgraham.com/popular.html