Coming from this comment. It's quite the change and ppy is willing do it, and I feel there should be support for this. If there is, I'd love to see discussion how things will work from criteria perspective, beatmap listings, modding, etc. so we can move toward making it happen.
Some benefits I see for this:
- Don't need to download difficulties you don't like.
- No more spread requirements - mappers don't need to make easy/normal difficulties for songs those difficulties are not fit for. More attention can be put on difficulties mappers actually care about, checks can go faster as a result too. I have no idea what the means for difficulty rank rate compared to now tho.
- A more variety of songs will be mapped. Not requiring to map a bunch of difficulties for one song means that time spent mapping those can be devoted to mapping other songs. This is a win for people who want songs to be mapped/ranked - well in a probabilistic sense it only makes it more likely to happen
- Duplicate skin files, hitsounds etc will with every difficulty. There can be a local collection of those things shared among multiple difficulties, but unless there are skin/hitsound/etc listings on web with added functionality of checking if you have the assets, this is going to add bandwidth.
- Less incentive to map certain difficulties. A quota system would solve that. Something like being obligated to rank N amount of easy difficulties before X amount of Hard, Y amount of Insanes, etc. Possibly scalable with ranking experience/accomplishments.
- Guest difficulties are obsolete. Not all too familiar with the role guest difficulties play in spreads, but I can imagine lesser known mappers use them to promote themselves. Either that or it's a more convenient way to get a map ranked without doing full spread - which is irrelevant in a system like this. Soooooo, if really need be there can be curated map packs I guess; mapsets made by curators.