Brotli is awesome for html/css etc web assets. On a few projects I’ve been able to use it to cut significant load time off projects.
Since Google provides libbrotli, I see no reason why it should be implemented from scratch in Crystal but linking in the C code would make a nice simple addition. It would be nice to get it for free in stdlib HTTP simply for how many projects would benefit but I’m not sure what the core teams philosophy is on that kind of thing these days.
As an application of Crystal toward a real world (useful) task, it would be at least an intellectually curious exercise to see how idiomatic Crystal code would compare to the reference C code (in size and readability, performance, memory, etc). It would be a nice person|team project just to learn|apply Crystal to a nontrivial problem if nothing else, for peole with the time to devote to it.
But every example of Crystal being used to do real stuff in a “better” way can only enhance its status, and answer the perennial question people ask (of any language really) - What’s it good for?