I agree what mo said, it gives a risk of mappers pushing single diff tv size for ranking

However, the main point i want to bring is that the proposal itself is meant to encourage mappers to push full sized song as there's isn't much players play full sized low diffs, which this phenomenon isn't clearly happening when it comes to tv sized maps (at least compared to full size). So it's reasonable enough to remove 2 diff rule for full version songs (>4:15) but not in tv sized maps (mostly referred to 2:30 if counting rhythm game size and anime TV size songs, but it this case, it's referred to songs shorter than 3:30). And removing 2 diff rule is still unfair to full size mappers because mappers can map a tv size calm song (e.g. R3 music box) with only 1 normal or easy diff, rendering they can rank single diff mapsets with (more than) half of the drain/play time then full size mapsets.

So besides what mo has proposed (to prevent abusing this rule by ranking low diff approval while the song is upbeat), i would like to propose the guideline in another way to prevent this happening so tv size mappers should map almost the same drain time as single diff full version maps:

Ranking Criteria wrote:

If the drain time of a beatmap is...
...lower than 3:30, the lowest difficulty of each included game mode cannot be harder than a Normal and should include 2 or more difficulties, regardless of how the song feels. Because osu!mania does not have a difficulty-specific Ranking Criteria yet, an osu!mania beatmapset's Normal difficulty is defined as a difficulty below 2.00 stars. For hybrid beatmapsets that include osu! difficulties, the additional modes’ lowest difficulties cannot be harder than a Hard.

While i felt that songs with between 3:30 and 4:15 probably stays on the grey zone, since some "full sized" songs with that size (especially between 3:50 to 4:15, from my experience) and "Hard" isn't exactly lands on either "low diffs" or "higher diffs" but stays on between. So it's hard to desire should we allow a map that sized can be ranked with one diff or not.

Sorry if it's hard to understand, my English is quite bad and i am not good at explaining things

Just make an exception that maps under a certain length need more than 2 diffs, ez.
If single diff set and sr[diff] <= minimum required then add one more diff' with sr[diff'] < sr[diff] (assuming mappers map the highest diff first and then filter out on lower diffs)

Sets with 2 diffs on short and slow songs have been a thing in the past so ya, if that chill slow song is 4 min now you wont end up with 2 diffs.

Reference: https://osu.ppy.sh/help/wiki/Ranking_Criteria#rules

Wording amendment

Old
...lower than 3:30, the lowest difficulty of each included game mode cannot be harder than a Normal.
New
...lower than 3:30, the lowest difficulty of each included game mode must be Normal or easier.

Old
...between 3:30 and 4:15, the lowest difficulty of each included game mode cannot be harder than a Hard.
New
...between 3:30 and 4:15, the lowest difficulty of each included game mode must be Hard or easier.

Old
...between 4:15 and 5:00, the lowest difficulty of each included game mode cannot be harder than an Insane.
New
...between 4:15 and 5:00, the lowest difficulty of each included game mode must be Insane or easier.

--
Reason for change:

It's easier to answer the question what is the lowest difficult I can go. The current one makes you do a double-take

I agree with mo, just maybe dont force Extra if the song suggests expert difficulties

I have no idea what Firis is trying to get across, because you say you propose something and then quote the RC and dont propose things? i dont know

Okoratu wrote:

I agree with mo, just maybe dont force Extra if the song suggests expert difficulties

I have no idea what Firis is trying to get across, because you say you propose something and then quote the RC and dont propose things? i dont know

The quote was his proposal; he proposed that songs higher than 3:30 should be extempt from the 2-diff rule, while TV size songs and any thing less than 3:30 should still have atleast 2 diffs; it seems like his proposal of solving the issue of people ranking a single easy diff of R3 music box 30 seconds, for example.

^ That's what i want to propose

Mohab500 for Firis' explaination wrote:

he proposed that songs higher than 3:30 should be extempt from the 2-diff rule, while TV size songs and any thing less than 3:30 should still have atleast 2 diffs;

I don't think this would fix it, that would make a lot of lazy sets with a single Hard diff, as what mo said.

Ranking Criteria wrote:

If the drain time of a beatmap is between 3:30 and 4:15, the lowest difficulty of each included game mode cannot be harder than a Hard.

then if a song can actually support a Hard difficulty, the song should support lower diff or higher diff like Normal and Insane, too, so there shouldn't be any extempt of two diff rule for set higher than 3:30 length.

so my propose of this is adding a new rule that clearly states to have at least 2 diffs inside a set with <5:00 length.

Ranking Criteria wrote:

Mapsets must have at least 2 difficulties. With an exception if the drain time of a beatmap is longer than 5:00.

Applying the minimum 2 diff rule for <5:00 sounds like a step in the right direction

um gonna stay out of this but i do have 1 thing to say... u do realize that was the old rule right? how is it the step in the right direction if it really just taking a step back...

anyways the rule just got implemented and there is more songs gonna still have 2 diffs for sure and even if some/few people is gonna abuse it (which they would abuse anything either way), you guys aren't even giving it a chance. lets test the waters out and see if this really is a problem which is why there is a 6 month to evalute to begin with (I think that is still a thing from 1 of oko's posts?)

“ima stay out”
steps in anyway

jokes aside, a step back can still be a sound decision in the right context.

i dont mind waiting to see how the current ruling plays out in 6 months but i would still prefer the 2 diff rule maintained in conjunction with the lowest diff depending on length thing going on rn

But what Mo is proposing is an actual compromise, i dont think it's a step back to add this to avoid single diff easies on 3:30 upbeat anime songs being acceptable

The problem with Mo's proposal is that it can be taken to the other extreme: a beatmap being forced to require 3 or 4 diffs (from Normal to Insane/Extra), more than the not-so-long-ago required 2. But the idea of considering the song in the condition upon which the guideline can be broken could be used in my previous guideline proposal. This would cover Oko's objection with my proposal that a mapper could break the guideline just by saying "I don't want to map a 2nd diff".

So something along these lines could do:

anyone who thought removing the two diff rule was a good idea should be removed from any decision-making body regarding RC

@ZiRoX The proposed guideline shouldn't be broken with just a simple "I don't wanna" since now the mapper should exhaustively explain why an extra difficulty wouldn't improve a mapset.

Your proposal could be interpreted as "my new Hitorigoto map has an Easy and a Normal. That's reasonable enough spread."

-Mo- wrote:

@ZiRoX The proposed guideline shouldn't be broken with just a simple "I don't wanna" since now the mapper should exhaustively explain why an extra difficulty wouldn't improve a mapset.

I was referring to my guideline proposal, not yours. In any case, it would be good if you let Oko know that "I don't want to" is not a proper reasoning:

-Mo- wrote:

Your proposal could be interpreted as "my new Hitorigoto map has an Easy and a Normal. That's reasonable enough spread."

Yeah, my proposal allows for a EN spread on Hitorigoto, but so did the old rule about having a minimum of 2 diffs. Basically, my idea is to rollback to the minimum 2 diffs, but giving the possibility of doing a single diff on songs that do not give much space for significant differences in density between diffs (hi, R3!).

In any case, requiring additional diffs is up to the BN. Personally, if I think a map should have one or two additional higher difficulties and the mapper doesn't want to add them, I just skip the map.

Having a guideline to say "you should map two diffs" but it not being required would make reasoning among the lines of "I dont need two difficulties" exhaustive.

Making it a requirement again is probably forcing difficulties that are either just dupes of existing things (aka this is the easy but harder) or altogether chore-difficulties for the sake of their existence and not for the sake of fitting music

Okoratu wrote:

Having a guideline to say "you should map two diffs" but it not being required would make reasoning among the lines of "I dont need two difficulties" exhaustive.

Making it a requirement again is probably forcing difficulties that are either just dupes of existing things (aka this is the easy but harder) or altogether chore-difficulties for the sake of their existence and not for the sake of fitting music

Having a guideline that says the top diff should correspond to the feel of the song would be equally affected by a reasoning like "I don't think it needs a higher diff", but somehow you're way more open to that idea.

I'm way more open to that idea because it makes more sense than what you propose in context of what i want to avoid, yeah

a single diff on a really slow song makes more sense than forcing two in

If you have an Insane as the easiest required difficulty and map a normal as a second optional difficulty you cant really go with a 1 diff normal for this sort of thing. this guideline would require you to add an insane to fill that in so you can't really bs around with it that much

People are usually way more in agreement of whether or not a song supports an Insane / Hard than whether or not representing it with a single easy difficulty is fine

-Mo- wrote:

Guideline wrote:

The highest difficulty of a mapset should correspond to the general feel of the song. An upbeat anime opening should have an Insane or Expert for the highest difficulty, while a calm piano piece can have a Normal as the highest. This is to ensure that the most popular difficulty of a mapset will properly represent what the song offers.

i'd clarify that this cuts off at hard diffs instead of insane+. hards usually cover rhythm closely enough without butchering it through simplification like e/n

with insane+ as the limit, i can imagine lots of cases where hard diffs fit "the feel of the song" fine, but an insane diff is possible through buffing spacing/activerhythms, so people will think it's needed (example: https://osu.ppy.sh/b/1710521 guideline makes me think this needs an insane when it doesn't)

with hard as the limit though, it would be more reasonable to figure out what fits. if normal is the highest diff and it's simplifying every rhythm (as most normals do) the guideline here would tell them to make a hard diff. if it's a calm piano song or an r3 music box, rhythms on a normal wouldnt be simplified, so uhhhhhh you get my point

there's some dense songs that hards would be simplifying, though they're rarer and harder to nail down, so i'm not sure it would be appropriate for the rc

The highest difficulty of a beatmapset should correspond to the general feel of the song. Easy/Normal difficulties can be used as the highest difficulties of a beatmapset if their rhythms are not oversimplified. A Hard difficulty or beyond should be the highest difficulty otherwise.

that should solve the concern of unnecessary 1 diff baby sets getting ranked because 2 diff rule is gone

RC wrote:

The highest difficulty of a beatmapset should correspond to the general feel of the song. Easy/Normal difficulties can be used as the highest difficulties of a beatmapset if their rhythms are not oversimplified. A Hard difficulty or beyond should be the highest difficulty otherwise.

I still feel like limiting it to a difficulty name such as 'hard' 'insane' etc.. is too contreversial/limiting.

What about something like:

RC wrote:

The highest difficulty of a beatmapset should correspond to the general feel of the song. It should not have oversimplified rhythms and object placement should correspond to the sounds in the music.

Ofcourse it's pretty dumb right now since I don't know how to really word it, but you get the point; someone with experience in RC stuff can rewrite it to get the point across easily.

i think it isnt too limiting, the scenarios where you'd run into this is when you're making easy / normal only on medium sized songs - provided people were hammered in the face for nominating these in the past i don't think this limits you at all - conversely with Mohab's wording someone could claim the songs supports an Ultra and demand an Ultra

and i dont think that's gonna end well

Okoratu wrote:

i think it isnt too limiting, the scenarios where you'd run into this is when you're making easy / normal only on medium sized songs - provided people were hammered in the face for nominating these in the past i don't think this limits you at all - conversely with Mohab's wording someone could claim the songs supports an Ultra and demand an Ultra

and i dont think that's gonna end well

it's just I am hesitant about this because I believe this is just too contreversial and subjective to an extent; I think we can mostly agree on what is 'oversimplified rhythm' and to a lesser extent 'simplified object placement' rather than, say, 'this song needs a hard', or 'this is fine as a normal'. But really, it's just what I think about the matter.

If the normal significantly simplifies the song as most normals on upbeat songs do it should have a hard

what's controversial about that

As I've already stated, I firmly believe only marathons should be able to have 1 diff. All others lengths should require at least two diffs. If that requires 2 diffs that are very very similar, who cares? If they are very very similar it should be easy to make and satisfy the 2 diff requirement.

tatatat wrote:

As I've already stated, I firmly believe only marathons should be able to have 1 diff. All others lengths should require at least two diffs. If that requires 2 diffs that are very very similar, who cares? If they are very very similar it should be easy to make and satisfy the 2 diff requirement.

I agree with this, the removal of the 2 diff rule allows 30 second mapsets with only an Easy or Normal to be rankable (same with 3:30 single hard diff or 4:15 single insane diff) and 1 diff will only appeal to a certain group of players. The main issue about it is that longer songs won't have low diffs anymore so new players can't play them, this is especially problematic on maps starting at insane or having only 1 insane diff at 4:15 length.

I tend to agree but can see where people are coming from saying that they have one diff that supports the song fine (e.g. a hard) and forcing people to make more difficulties than that may be detrimental to the overall quality of the mapset.

The proposed guideline would limit the easiest hardest difficulty to be a hard in case a Normal / Easy do not fit the bill when it comes to actually following the music without skipping collossal parts of it

As such the following scenarios are possible:

1. Lowest difficulty Normal required, song fits extra: highest difficulty should be a Hard
2. Lowest difficulty Hard required, song fits extra: highest difficulty should be a Hard
3. Lowest difficulty Insane required, song fits extra: highest difficulty should be a Hard (so it's optional)
4. Lowest difficulty Normal required, song only fits Normal, just normal is fine
5. Lowest difficulty Hard required, song only fits normal: just normal is fine
6. Lowest difficulty Insane required, song only fits normal: just normal is fine

this change would require two difficulties on mapsets that need a normal at least and can support more than a normal

in any case you would have to provide at least a Normal or Hard in depending on song length -> If you want to make a solo normal on a 30 second song you need a very calm 30 second song and for any other length respectively

-------

The main point for the two difficulty requirement removal was that it generates very weird scenarios:

1. Normal lowest diff: forced to make an Easy or a Hard depending on what you like more
2. Hard lowest diff: forced to make a Normal (aka make the proposed change useless) or map an Insane you never intended to make
3. Insane lowest diff: forced to make a Hard, Normal, Easy or Extra <-> if you never intended to make any of these yeah

logically this sort of levels out the "effort" required - putting that guideline Mo proposed in place is supposed to counteract trollsets where the creator just speedmaps an Easy difficulty within an hour and claims it's rankable even though the difficulty itself maps like 20% of the music and addresses a target audience that moves past them within days.

Adding the two difficulty thing as a guideline or rule would require explanation of:

1. what necessitates this? Making lower difficulties optional was done because we deemed the ranked section to contain enough for getting anyone started so forcing either of them in whatsoever is somewhat counter intuitive
2. If this was a guideline would "I think mapping anything harder or easier than what i have done would misinterpret the song" be reason enough to not do higher or lower difficulties than what you have done? How do you want to define this with the previous points in mind?

------------------
As a closing sentence towards anyone bothered to even read what i have to say: please note that i'm a proponent of getting rid of the marathon category as a whole and require everything to have a full spread - the above is by no means my opinion but the result of argumentation within this thread that i don't have counter arguments for that don't conflict with the basic idea of this proposal (which was, making lower difficulties optional based on song length - which was green lit by the lead dev and owner of this platform) - asserting that i'm more righteous than anyone else on the matter is pretty arrogant so I won't

For what it's worth I can see the reasoning for gamemodes other than osu claiming certain parts of it are too small to universely be able ot fall under this rule (i.e. osu!mania keymodes come to mind or osu!catch)

handling this on a per mode basis is possible but I'd just disallow hybrid mapsets if you want to not have this be in the general ranking criteria because you can't claim that normal difficulties convert to normal difficulties in taiko / catch / mania the same way you can right now

the recently discussed guideline in this thread has been applied in https://github.com/ppy/osu-wiki/pull/1651 and will be amended to the ranking criteria in about 5 days

if anyone has concerns about the change, now is the time to show them!

pishifat wrote:

the recently discussed guideline in this thread has been applied in https://github.com/ppy/osu-wiki/pull/1651 and will be amended to the ranking criteria in about 5 days

if anyone has concerns about the change, now is the time to show them!

Shouldn't there be something about marathon mapsets? What if the mapper just wants to make a 5+ minute Easy or Normal mapset?

Songs longer than 5 minutes do not need a reasonable spread which kind of excludes this
the above pull request is merged