So there's been some cases already where a mapper pretends to use some of their old creations for a compilation, or simply use an already ranked map of theirs, change few notes, and get it re-ranked under a different set.

I (and some others) believe it's not a decent practice to do this, as it bloats the ranked section with redundant content. But there is no rule about this as we speak, so I would like to propose some kind of rule that would get added under the General Rules in here: https://osu.ppy.sh/help/wiki/Ranking_Criteria.

The rule would look something like:

1. Content that is already ranked cannot be reutilized for rank unless there is at least a 80% difference between the two creations. This is to avoid unnecessary redundant content in the ranked section, in addition to keeping it as fresh as possible.

80% is initially an arbitrary number, I'm simply using what's normally used in scientific works (they allow up to 20% of same content due to article/other work citations). Any ideas are welcome to discuss.

I support the addition of this RC rule as it would provide a reasonable ground to avoid any form of self-plagiarism. We do not need the same content twice in the ranked section so a 80% difference between two works seems suitable.

In fact I am a bit surprised that we do not have a rule regarding this for so long. As we do not allow beatmap stealing in general, mapper copying their own creations should also be governed as well.

I think the idea is good, but "80% difference" could mean a lot of things.

Does it count as "different" if I only swap two patterns in time that map to the same rhythm?
Does it count as "different" if I only Ctrl+H +J a few things?

Nitpicky, I know, but I think it should be made at least reasonably clear since it's labeled as a Rule. I think a good rule of thumb (hah) is that all Rules should leave very little up to interpretation

Creating some kind of program to more consistently determine percentage changes might actually be feasible, although this'd probably require tons of testing and time to get working. I'm implementing a snapshot feature into the aimod I'm working on anyway, so along with that I could add a percentage change algorithm for each mode, which we could then use to compare two different maps in this way and draw a line. Just keep in mind that this could take absolutely forever since I still have a lot to work on, so getting some faster alternative in the meantime would probably be a good idea.

Anyway 80% change seems pretty unrealistic (at least for taiko and mania) considering that they are mapping the same song, after all. Scientific works don't necessarily need to write to the beat and follow the intensity/measures/whatever so that would explain why they work with a large difference requirement. For standard this might work though if you consider positioning and all that, but we'll have to see I guess, really depends on how it's calculated/judged (like how much you weigh each part etc, could potentially get really complicated, hence why a program is probably necessary here at some point unless we're only interested in dealing with the most obvious stuff).

Even though the idea seems more than clever and brilliant, the actual wording makes it automatically impossible for anybody to rank anything in standard. I mean, "content that is already ranked" doesn't specify the fact that the already ranked map cannot be ranked one more time, people will say the sentence is so vague they won't be able to utilize 1-2 jumps or symmetrical patterns or even wave sliders.

Also, as long as such a rule would make people more creative, I beg for this proposal to never become a true rule as the standard community has been becoming way too creative with their 'avangarde' and 'technical' maps. I'd rather see another Gunther compilation with same patterns rather than one's 'creative' death streams on this artist's songs and I don't want to see Andrea become a 8* mapper, and with your rule that would be the case.

Imagine the situation where 5 different mappers want to map the same anime song, and it is not an unrealisitc idea, and we all know they will use very similar jumps and distances and even settings. Again, we all know the maps would be extremely similar in design as anime maps happen to utilize those 80% of same ideas. As long as it enforces creativity it doesn't actually support the idea of mappers becoming more of the quality content creators, they will just find more lazy ways to be edgy.

However, if this rule is osu!taiko only then I can't say 'no' as I am not proficient in this game mode.

clayton wrote:

I think the idea is good, but "80% difference" could mean a lot of things.

Does it count as "different" if I only swap two patterns in time that map to the same rhythm?
Does it count as "different" if I only Ctrl+H +J a few things?

Nitpicky, I know, but I think it should be made at least reasonably clear since it's labeled as a Rule. I think a good rule of thumb (hah) is that all Rules should leave very little up to interpretation

As stated on the original post, the whole percentage thing remains arbitrary, as there would be little to no objective criteria to establish a % of "acceptable copying". I do not know how that would work in standard, to be completely honest. But that's why this thread exists.

Naxess wrote:

Anyway 80% change seems pretty unrealistic (at least for taiko and mania) considering that they are mapping the same song, after all. Scientific works don't necessarily need to write to the beat and follow the intensity/measures/whatever so that would explain why they work with a large difference requirement. For standard this might work though if you consider positioning and all that, but we'll have to see I guess, really depends on how it's calculated/judged (like how much you weigh each part etc, could potentially get really complicated, hence why a program is probably necessary here at some point unless we're only interested in dealing with the most obvious stuff).

Again, the 80% was merely a starting point. It's very much arbitrary, subject to change. But we would appreciate indeed that kind of AiMod implementation.

Krfawy wrote:

Even though the idea seems more than clever and brilliant, the actual wording makes it automatically impossible for anybody to rank anything in standard. I mean, "content that is already ranked" doesn't specify the fact that the already ranked map cannot be ranked one more time, people will say the sentence is so vague they won't be able to utilize 1-2 jumps or symmetrical patterns or even wave sliders.

Um... yes it does? The sentence is not vague whatsoever. Content that is ranked cannot be reutilized. It does specify that an already ranked map cannot be re-ranked, since it's being reutilized.

Krfawy wrote:

Also, as long as such a rule would make people more creative, I beg for this proposal to never become a true rule as the standard community has been becoming way too creative with their 'avangarde' and 'technical' maps. I'd rather see another Gunther compilation with same patterns rather than one's 'creative' death streams on this artist's songs and I don't want to see Andrea become a 8* mapper, and with your rule that would be the case.

Imagine the situation where 5 different mappers want to map the same anime song, and it is not an unrealisitc idea, and we all know they will use very similar jumps and distances and even settings. Again, we all know the maps would be extremely similar in design as anime maps happen to utilize those 80% of same ideas. As long as it enforces creativity it doesn't actually support the idea of mappers becoming more of the quality content creators, they will just find more lazy ways to be edgy.

However, if this rule is osu!taiko only then I can't say 'no' as I am not proficient in this game mode.

The "anime maps" thing is perfectly valid, but with "already ranked content" it's specifically referring to maps created by the same people or stolen from other people (but there's a rule about beatmap stealing already). This would solve the self-plagiarism issue.

This rule would obviously not be mode specific. It would be for all modes.

It would be safer to reword the "content" to "beatmaps" or "maps" as the very actual word might mean "map" as well as "pattern", "design" or even "song" as these are also what we count as the game content that gets ranked. As I said, now people may start thinking they can't even reuse the same slider in another map or a pattern, and let's be honest, most people know how the note-code works and what they also know is how to borrow one's shapes from a map with or without the owner's knowledge. In some lucky cases they won't even need to rescale the objects. And what will they do in the case the slider lasts 10 seconds out of 30-40 seconds of the whole song? In such a case someone can say "oh look this object is literally copied from the X map by Y and the rules say you can't reuse the ranked content!" and if the song is 30-40-second short the rule might cause the riot.

You might say that I am being problematic and imagining the least likely scenarios but the QAT know the standard people will be more than toxic about not using the word "beatmap" and that they can be, putting it lightly, upset about it. We all know there are discord servers where people literally use templates or copy the very patterns and some maps just end up being a 70-100% mish-mash of them being compiled and I wonder how then anyone is going to justify the "redundancy rule" as these patterns are most of the times taken from the ranked maps. In such a case this is a literal copy-pasting and what's even harder, you won't be able to point just one map and say "oh it's taken from the Haitai" if we mix this, Monstrata's triangle maps and UC's high-speed waves.

In these days I am really not sure if this is the most appropriate proposal, really. It would've been better if the rule existed back in the 2009 but now it's... well, pointless from my perspective as people are technically doing all the same unless they create a rocket and a butterfly. I mean, really, if someone tried to make a compilation of Camellia and mapped it from the very beginning you wouldn't've noticed it was any different from compiling any maps from the ranked section.

can you give some examples of the types of maps this is targeting? i have a hard time picturing what maps exactly this rule is stopping

-
i say this because there is already precedence for stopping maps that are too similar to other ranked versions - for example monstrata's original taki and mitsuha things, i assume you can use the same logic to stop harumachi in this case since they are very similar; it doesn't need to be a separate, extremely vague rule that will definitely be misinterpreted by anyone who doesn't know the exact situation that the rule is being made for.

@Krfawy Think Raiden is more referring to that mappers shouldn't plagiarize themselves and be able to rank the exact same beatmap again (or with very minor changes like changing only a small portion of the objects). See it kind of like the current beatmap submission rule which is in effect:

Beatmap Submission Rules wrote:

Do not plagiarise or attempt to steal the work of others.

Except this rule is instead "Do not plagiarize any ranked content, whether made by you or not" up to a similar threshold, which as we're seeing is not so low that people would accidentally plagiarize each other over stuff like haitai. Taking two generic pp maps in standard of the same song and same difficulty and comparing them would still yield a really huge difference so I wouldn't worry about that. Regarding sliders, people copy sliders from each other all the time and they're not getting killed for that even if the beatmap submission rule is already in effect, and considering that this rule is supposed to be enforcing things to a similar leniency that shouldn't be a problem.

I do agree it should probably be "Beatmaps" rather than "Content", as content could refer to basically any portion of a beatmap. Moreover, clarifying that only hitobjects are taken into account for this so stuff like metadata or combo colours doesn't get mixed in here would probably make things clearer as well. Also I don't think we need this for anything other than self-plagiarism since the beatmap submission rule is already a thing, so making this intention clear in the rule might clear up some confusion.

Oh, okay so it's like this... and about the stuff that would be unrankable I obviously thought about the mentioned Gunther compilation and Nightcore compilation, both by Andrea.

One question though, would this rule make a whole set unrankable if every difficulty was designed in an almost identical way but with the only obvious differences such as spacing and density alternations? The easiest example would be this: https://osu.ppy.sh/s/860248 I am asking as I've always wanted to create something with as linear difficulty gaps as possible and now I am not sure if it would be okay or not. The whole set uses similar/same sliders and patterns with as few changes as possible.

is this 80% rule over the entire song length? Or just the copied section? Like lets say I map and rank a 1:30 song, and then I include it in a 30 minute song compilation. Do I have to change 80% of that included 1:30 map in the song compilation? Or does just at least 80% of the map have to be new content?

Nao Tomori wrote:

can you give some examples of the types of maps this is targeting? i have a hard time picturing what maps exactly this rule is stopping

-
i say this because there is already precedence for stopping maps that are too similar to other ranked versions - for example monstrata's original taki and mitsuha things, i assume you can use the same logic to stop harumachi in this case since they are very similar; it doesn't need to be a separate, extremely vague rule that will definitely be misinterpreted by anyone who doesn't know the exact situation that the rule is being made for.

It's mostly for cases like https://osu.ppy.sh/forum/p/6089473 or the recent Harumachi case you mentioned

Krfawy wrote:

Oh, okay so it's like this... and about the stuff that would be unrankable I obviously thought about the mentioned Gunther compilation and Nightcore compilation, both by Andrea.

One question though, would this rule make a whole set unrankable if every difficulty was designed in an almost identical way but with the only obvious differences such as spacing and density alternations? The easiest example would be this: https://osu.ppy.sh/s/860248 I am asking as I've always wanted to create something with as linear difficulty gaps as possible and now I am not sure if it would be okay or not. The whole set uses similar/same sliders and patterns with as few changes as possible.

It always will depend on whether there is a same "design" or flat out copypasted patterns. A similar design is a very valid way of creating a spread conceptually. Copypasting? Not so much

tatatat wrote:

is this 80% rule over the entire song length? Or just the copied section? Like lets say I map and rank a 1:30 song, and then I include it in a 30 minute song compilation. Do I have to change 80% of that included 1:30 map in the song compilation? Or does just at least 80% of the map have to be new content?

The entire copied song. In the compilation, that 1:30 song must be "80%" different from the previously ranked one.

Raiden wrote:

tatatat wrote:

is this 80% rule over the entire song length? Or just the copied section? Like lets say I map and rank a 1:30 song, and then I include it in a 30 minute song compilation. Do I have to change 80% of that included 1:30 map in the song compilation? Or does just at least 80% of the map have to be new content?

The entire copied song. In the compilation, that 1:30 song must be "80%" different from the previously ranked one.

Dang. I don't think that is possible. Even if you were to have an entirely different person map it. There is bound to be similarities. and if you have the same person map it, its pretty much impossible in my opinion. 80% is a big portion. It seems more like it'd be forcing the mapper to get a GD'er to circumvent self-plagiarism. And any time you force someone to get another person's help isn't good in my books. I can say without a doubt, if I were to do that exact scenario it'd be impossible for me to make my map 80% different from the previous one, even with a full remap. Especially in gamemodes like taiko and mania where unlike osu! and ctb there isn't the position of the note on screen to add a huge depth of uniqueness for each map.

You could even test if its plausible. Have a decent sized group of people (at least 10) map the same song, and then 2 weeks later have them map the same song without referencing the old map at all. I believe they are bound to be at least 80% similar. Though sadly there would be room for cheating in this sort of test.. But why cheat when there is no benefit?

Naxess wrote:

Anyway 80% change seems pretty unrealistic (at least for taiko and mania) considering that they are mapping the same song, after all.

Yea speaking of mania, I made maps based on other maps before either in my style or as a means of difficulty extension. Here are some examples:

https://osu.ppy.sh/b/1188254 and https://osu.ppy.sh/b/857513 <- my style
https://osu.ppy.sh/b/1093190 and https://osu.ppy.sh/b/802866 <- difficulty extension

Not sure if they satisfy the 80% change threshold


Also I'd like to consider the following edge case: Suppose you are one of the mappers for a song compilation project where your section is 20% of the map that got ranked. Later you decide to map the full version of the song and use that 20% that you mapped for a difficulty and finish the rest 80% of the song which was not present in the song compilation. Are you allowed to rank it?

For the sake of having a style of mapping people want to see an extension of, this will be awful. It's like providing a montage of previews for 5 short movies and then saying the movie will need to be different if you want the full thing.

The idea is good and necessary from my perspective, but I think there are some essential things to be clarified.

The first is how we judge the similarities. No doubt similarity in style is okay since that is quite personal, and direct copy-paste is a no-no. However, how can we handle the case when a mapper copies his work and make some adjustment to the notes? Is it okay to just flip all notes? (I would say no in this case. ) Is it okay to keep rhythm the same but employ different patterns or placement? (My answer is yes) I'm afraid it is hopeless to find a nice standard to quantify the similarities, as mapping is more like "art" instead of "paper", where people can quantify the latter (imagine your BSc or PhD thesis) while having no idea about the first one (actually people are still struggling with deep learning on images). I believe many discussions need to be carried out in order to determine a good standard for this rule.

The second is the "already ranked". I prefer changing it into "already in ranked/qualified/loved section". I believe this is easy to fix.

i think going with a hard number at all for how much copying is ok is a bad idea. like right now, it's taboo to re-rank the same song even without a rule, but with this rule clarifying that it's ok to copy up to 80% of your own maps, you'd be free to just like... copy 79% of it, then map the rest properly. it would still be taboo and i doubt anyone would be stupid enough to do that, but they'd be able to argue that the rc supports their amount of repeated content

the better choice here is to have a guideline saying something like "dont copy paste ur own ranked maps" and leave the discretion of whether or not the guideline applies to the qat and bns. what's considered copying varies between modes, so wording it like this will avoid any of those issues, while still having some rc reference to fall back on when enforcing repeated ranked maps

write a guideline like that, then if people agree to it, we can add it

How about:

1. Avoid re-using your own ranked beatmaps to rank other beatmaps. This is to avoid unnecessary bloating of the Ranked section.

that seems logical.

Okay, now it is sexy.