Over the past few years of writing, I've noticed I've become more comfortable experimenting with point of view, technique, style, etc, and that I've been willing to tackle more difficult projects. Throughout it all, however, I haven't lost my focus to start and end with story. To me, experimentation isn't - and should never be - an end in of itself. It should be a byproduct of the requirements of the story ... used when a "weird" perspective is the best, strongest way of telling the story.
With "Who Wants To Be A Hero?" I've dabbled in multiple kinds of points of view - a combination of camera-POV third person (no internal narration), deep third person, and first person - and what you might call metafiction. It walks a delicate line between making the parallels to reality competition television evident and breaking the fourth wall ... and I wanted, first and foremost, to create an internally consistent story. There's also some metahumor, jokes and call-outs the characters wouldn't recognize as humorous, but the reader hopefully finds so. I've made fun of creationism, the American Revolution and various aspects of technology. But these weren't elements I consciously started out with: they grew from my figuring out how to handle the storyline.
Another story finished recently is "Of Two Minds," in which there is no narrative / description, simply dialogue - two characters trapped in the same head, arguing with each other. The action and setting are conveyed by the characters' comments within the dialogue. But the starting point for this was not a conscious challenge ... rather, these two people popped into my head (would that make three people in the same head?) in response to a free-write prompt, and what I "heard" very clearly was their sniping, rapid-fire argument.
Occasionally, I do choose between projects due to technical merit. "Scylla and Charybdis" edged to the top of my list of next-projects (back when I wrote it, in ye olde days) because it was science fiction, and I thought attempting to write it with thought for rotation and revolution, light-distances and technological incorporation - even if none of that appeared obviously in the text - would be good for my skills as a writer.
My point here, I think, is there's room for experimentation and works that require walking a difficult technical line ... but for me, it has to be a necessity, an outgrowth of the story. If you could take the same story and tell it just as effectively without the experimental technique, it shouldn't be there. Obviously, whether or not an individual story requires its whacky elements is a judgment wherein reader and writer may not see eye to eye. But to simply try it on at random, without fusion with the content and feel of the story, seems disgenuous and show-offy. Experimentation is not an end in of itself.
Your mileage may vary, of course ... but this is the reason why I, personally, bristle when a reviewer comments on my use of experimental elements.