A too-twitter-didn't-read-summary is that IF is interesting even if you're not a wordy-mcwordsmith writerly kind of person, simply because it toggles the unspoken default of essentially all other games that you have to make graphics for them before you can even get off the ground.
I think it's why I've gravitated more often to simple geometric graphics in a lot of my experimentation because (although it still entails choosing viewport size and colors and shapes and etc.) it feels at least like less of an upfront price to pay than firing up the gimp and drawing a whole palette of sprites or whatever. Or, god forbit, firing up blender and doing 3d modelling. That's pretty much what's kept me out of being involved with 3d gaming (which, let's face it, is virtually synonymous with modern, big-budget commercial video games) --- the idea that asset creation alone is a huge fraction of the effort of creating a game, when I'm more interested in the rules part.
(On the other hand, I don't want to give the impression that I think game art, for example, is not important, or not valuable. I'm just not as trained in making it, and it seems hard to make any that is not obviously totally shitty without significant practice and/or training.)
Then I reflect on the phrasing of the proposition in my head that:
It's easier to express pure gameplay ideas in IF because it doesn't have graphics
and there's something about it that itches me, something that reminds me of saying people like languages that don't have a type system because then the type system doesn't get in their way.
I'd rather say: sometimes it's easy to sketch what you want to say in scheme because its type system (it has one!) is highly constrained, and has good defaults. Because the language has a single type that to our squishy brains feels nicely "homogenous" or "regular" or something and its libraries are full of good tools for manipulating that single type. All of the effort of the toolbuilders has been centered on that one type, so the tools are very mature. Scheme is not typeless, it is unityped, and that type is chosen to be a good default type. So good, in fact, that it almost tempts you to forget it's there.
And I'd rather say: It's easier to express pure gameplay ideas in IF because its visual presentation (it has one!) is highly constrained, and has good defaults. All you get is a stream of characters. But it's a system of communication with a rich history, because a community of people have standardized on all speaking that language, and have used it for untold cumulative lifetimes, and so the tools are very mature. IF is not graphicsless, but it has only one kind of graphics, but it's a phenomenally expressive kind of graphics for what it is.
Then I wonder what I really mean by "good defaults". It's something about a effort-to-result-curve that is not super steep near the origin: if I want to reliably inform the player that there's a loaf of bread here, typing " b r e a d " in the emacs buffer already containing the code of the game is a lot easier than drawing a convincing picture of bread, saving it somewhere, and plumbing the image loading process through some code and deciding where it goes on the screen, etc. etc.