While I agree with your high level view, @devinprater@dragonscave.space, and also think social media (or media portals in general) should have caption for all nontext content, IPFS operates on blocks of data without knowing what they are. While it can be used as a storage medium for the web, it belongs to a lower layer: what you'd want is a web-like protocol where a11y is strictly enforced, not limited to alt texts but also to forbid all flashy things confusing people with a perfect vision.