Disagree strongly on spread. Few points:
* I'm in a belief every player has a comfort zone when it comes to difficulty of map. Maps outside this comfort zone are either too easy and boring to play or simply too hard to be enjoyable. As time goes on and your skill increases, this comfort zone goes towards upper spectrum of maps. This leads to easier difficulties becoming too easy and uninteresting to you. I'm talking about easy/normal/hard becoming irrelevant to you after certain skill level. This could be verified by poll, statistics etc.
* As time has went on, average comfort zone has ended up at "Insane" level of maps (based on graphs ephemeral posted). Maps below insane have playcount of half or third above it, making difficulties of that level a minority in popularity.
* While I do agree newcomers need content to play, I don't think there needs to be constant supply of entry level content. This is because the lowest difficulties serve as tutorial to game and generally you progress past this skill level extremely fast. Like I stated earlier, majority of players are Insane tier players but low skilltier content is forced to be produced. Why? This content becomes redundant for player as time goes on and there is already a ton of entry-level content.
* There has been an argument about a need for modern new songs, like trash metal, to be mapped because newcomers are looking for recently released songs to play. Well first of all, newer songs aren't necessarily better or more desirable to look for. I would argue everyone listens to older songs too and you don't only listen the newest rad top50-played-in-radio-100-times-a-day song. I do agree it's nice to find new songs being mapped too but they aren't the ONLY songs mattering. And there is absolutely nothing uncommon about finding songs you like from older production. Because of these reasons, I would argue osu! already has enough content for everyone to enjoy and discover during their entry skill level days. I see no merit in FORCING mappers to cater to newer players and invest time into lower difficulties.
* There has been a point missing in this thread, and that is: Each difficulty requires time and effort to make and ensure quality, this time is taken from mapper(s), modders and BN's. More difficulties = more effort required to make it ranked. I am in understanding that the more complex the difficulty is, the more time it requires to create and review quality. Now because of the spread rule, let's say you are mapping some aggressive metal song that screams for brutal 7* difficulty to make justice for the song. To get this ranked you would need 2*, 3*, 4*, 5*, 6*, 7* difficulties as bare minmum. Song is pretty average metal song, 4 minutes long and has a variable timing and could benefit a lot from special attention to hitsounding. To get this ranked you need to invest major time to timing (variable / complex) and hitsounding design. You also need to map the same 4 minutes for each difficulty 6 times, this is 24 minutes in total. Each difficulty also needs to meet strict quality criteria and you know 7* alone is gonna take at least 3 pages of mods and you wonder how to express this minute long guitar solo in 2* map. I would see this situation being quite common when trying to provide ranked content for current top players AND future highly skilled players. Because of current, and proposed spread rule, you need to make some extraordinary effort to get high* maps ranked alone. It isn't simply linearly increasing workload, its multiplicative.
* Related to above, there are extremely few ranked/approved metal songs. Same goes for rock (excluding jrock), pop (excluding jpop/kpop). Currently ranked sets are saturated with tv sizes and dnb. Why? Obvious music taste differences aside, songs of those genre tend to be shorter and therefore have less to map. Current spread rule promotes mapping shorter songs because it takes less effort to do so, 6 difficulties on 2minute song is just 12 minute of mapping. This compared to other extremely unlucky 4 min song requiring TWICE the time to be mapped. Also keep in mind, it also means more of mapping to be checked by multiple modders and bn's.
* The issue above could be alleviated in many ways, personally I would see midde ground solution working best. I would add seperate category for full size songs where you could make do with more lax difficulty spread and moving approval time limit further.
Mapset with 7* difficulty (proposed criteria)
2 min: 6 difficulty spread =12 minutes to map
4 min: 3 difficuly spread = 12 minutes to map
6 min: 2 difficulty spread = 12 minutes to map
Mapset with 7* difficulty (old criteria)
2 min: 6 difficulty spread = 12 minutes to map
4 min: 6 difficulty spread = 24 minutes to map
6min: 1 difficulty = 6 minutes to map
* As you can see, currently workload is extremely inbalanced. Half of work needed for approval, twice for 4 minute song.