We have over thousand instances and a bunch of them are caching the same images as other instances. Wouldn’t it be nice for users to get image from the closest instance? We already have a lot of data overlap and we would need to „just” implement protocol to find the same data and retrieve it to the user.
Just came across this thread. I would definitely love to see ipfs being used for lemmy. It fits perfectly and makes sense as well as save storage space. That study was published in 2022. Has there been any improvements to ipfs latency since then?