Edwin Abbott's 1884 novella, Flatland, recounts the misadventures of a square that lives in a two-dimensional world called "Flatland". In this story, the square has a dream where he visits a one-dimensional world (Lineland) and unsuccessfully tries to educate the populace about Flatland's existence. Shortly thereafter, a sphere visits Flatland to introduce our protagonist to his own home, Spaceland. The sphere looks like a circle to the square, because the square can only see the part of the sphere that intersects Flatland's plane. The square can't fathom Spaceland until the sphere actually brings him into the third dimension.
Having his view of reality sufficiently rocked, the square postulates the existence of still higher dimensional lands. The sphere denies this possibility and returns the square to Flatland on bad terms.
The irony here is that, in spite of being aware of the square's previous insistence that Flatland is the only reality there is, the square's corresponding limitation of thought, and the square's subsequent enlightenment; the sphere arrogantly asserted that his own land represented the limits of dimensionality and that there can be no more dimensions.
I begin with this summary not as an impromptu book report, but because this is a perfect analogy for why I've decided to learn Haskell.
It isn't hard to imagine the inveterate C programmer being hesitant to embrace higher-level languages like Python and Perl. "Imagine the whole world that opens up to you when you don't have to worry about memory management and resolving pointers," requests the proselytizing Python programmer.
But the C programmer got on fine without knowing Python all this time. Besides the (standard) Python interpreter is written in C.
"Why? So I can change types on a whim?" retorts the C programmer. "...so I can pass functions as arguments to other functions? Big deal! I can do that with function pointers in C."
While it is technically true that, in C, functions can be passed as arguments to other functions, and can be returned by functions, this is only a small part of the functional programming techniques that Python allows. But it is the only part that most only trained in C can understand. This is like when the square thought that he saw the sphere because he saw a circle where the sphere intersected Flatland. There's a bigger picture (lambda functions, currying, closures) that only appears when the constraints imposed by a programming paradigm are pushed back.
Shifting tides in industry eventually compel the C programmer to give Python a shot. Once she is fluent in Python and its idioms (becomes a "pythonista", as they say) things begin to change. The OOP of Python allows the programmer to gain a new perspective on programming that had been, up to this point, unavailable to her simply because the concept doesn't exist in C.
Resolved to follow the road to higher abstraction, the (former) C programmer asks the Python programmer if there are other languages that will provoke a similar shift in thought relative to even Python. "Haskell, perhaps?" she offers.
The Python programmer scoffs, "Nah, Python is as good as it gets. Besides, purely functional programmers are a bunch of weirdos."
Haskell is a relatively obscure programming language (20th place or lower on most popularity indices), but accounts for a disproportionate number of the "this-language-will-change-the-way-you-look-at-programming" claims. By the accounts of Haskell aficionados, finally understanding Haskell is a transformative experience.
Up until recently, I found myself mirroring the thoughts and behavior of the Python programer in this story; I keep hearing about how great Haskell is, and how it'll facilitate an enormous shift in how I'll view computer programming, but I largely dismissed these claims as the rantings of eccentrics that are more concerned with mathematical elegance than practical concerns.
Is it any wonder that I didn't believe the Haskell programmers? I can't see the benefits of Haskell within the constraints of how I view computer programming—constraints subtly imposed by my language of choice. (Lisp programmer Paul Graham describes this as the "Blub Paradox" in this essay Beating The Averages)
But just like the sphere and the arrogant Python programmer should, I have to remain open to the idea that there's a whole world out there that I'm not availing myself of just because I’m too obstinate to try it.
Any language that is Turing-complete will let you do anything. It's trivially true that there is nothing that you can do in one language that you can't do in another. What sets languages apart (from most trivial to least) is the elegance of the syntax, it's community and third party packages, the ease in which you can perform certain tasks, and whether you’ll achieve enlightenment as a result of using it. I don't know if Haskell will do this for me but I think I ought to give it a shot.
If the ideas of relativity when applied to the domain of programming languages interests you, you might also be interested in the Sapir–Whorf hypothesis, which applies similar ideas mentioned in this post to the realm of natural languages.