Tom94 wrote:

Hello everyone,

note that currently the osu!mania ranking is not functional. If you have any suggestions on how the future system should work, then leave it in this thread, please.


I'd like to mention a few things regarding how I plan on tackling this.

The general direction:

1. First of all, everyone can decide on his/her own which maps he deems good or bad, or which ones he/she prefers playing. We shouldn't be forcing certain maps on people by making them worth more for no sensible reason.
2. I'll try to make an algorithm that's fair for both auto-converts and mode-specific maps. We will all see how it turns out, and if it's alright, then I see no reason to reduce the value of auto-converts. As mentioned above you are still free to play the maps you prefer playing. I can however understand where the auto-convert argument is coming from, and it'll definitely be addressed if there won't be a good solution that weights all maps the same.
3. Default X-key modes exist for each specific beatmap for a reason. My initial plan is to give slightly more points to the default X-key mode of a particular beatmap than any custom one. This is not set in stone yet and most likely will change in one way or another while testing.
4. Unfortunately X-key specific ranking is not going to happen. That'd be for one part a way too huge amount of things to manage and for the other would clutter up everything.

Beatmap difficulty:

As for Taiko and CtB I also got an initial difficulty list for osu!mania ready. This is the gamemode I am the least sure about my results and so far only the base algorithm is done with no variations. However, you can already see how low converts are weighted simply because they are so easy, I've included a few of the harder ones:

https://docs.google.com/spreadsheet/ccc ... _web#gid=0

Get the feedback going.

Aqo wrote:

http://osu.ppy.sh/b/231825?m=3 porno map top performance

[Normal] autoconverts beating 90% on mania specific highdiffs of charts

something somewhere went horribly wrong. please fix.

Xcrypt wrote:

I have the same issue, and I know others who have too. Some of the harder maps became unplayable because of the choppiness.
Oh and good job Tom

SM_Bluefusion wrote:

Is it normal that weird auto-converted maps on the highest difficulty can beat any mania map by a ton?

Songs like:
http://osu.ppy.sh/b/152561?m=3
http://osu.ppy.sh/b/198428?m=3
http://osu.ppy.sh/b/152480?m=3

Everyone can recieve tremendous amounts of PP just by ranking them moderatly with any mod and I bet there are way more.

richardfeder wrote:

Just a quick glance >,<
Clearly there is huge issue about low key amount diff earning a relative high difficulty. One thing I want to point out is that you cannot simply put them all under one single rating system although we have to do that in the future anyway. So now in the early stage of ppv2 I suggest make unique list for every key amount or at least don't put something completely different(for instant 8k and 4k) together. The comparison among different key amount are totally none sense for now.

A couple of opinions:

In difficulty 6, take a look at 7k charts.
Unfinished 7k Hard and Eden HD. Those two clearly don't belong the group of difficulty 6.
Eden HD -> although there is no really outstanding difficult part in this chart, parts contain continuously 1/4 in a relatively high BPM and many 4key-press at the same time make this chart extremely hard compare to other maps in the same level.
Unfinished 7k Hard -> 146BPM with some 1/4 and easy 1/2... seems not harder than other maps in this level. Actually not. Pattern make this chart a little bit hard to read and press for players in that level such as 00:20:065 - and 00:25:099 - , and for those who can not master higher level it is extremely hard to earn FC and high acc(OD8 btw).
So what's important? Sharing a similar average density, songs contain remarkable hard part (like burst or strange pattern) are usually deserved a higher difficulty. Everything unique may makes a chart harder than usual.

Xcrypt wrote:

1. rhythmical planet (collab) overrated, there are some pretty hard parts in there, but also quite a bit easy parts that pump up accuracy so it doesn't deserve to be so highly diff judged.
2. imperishable night 2006 (7k insane) overrated, there are some pretty hard parts in there, but also quite a bit easy parts that pump up accuracy so it doesn't deserve to be so highly diff judged.

I judge underrated/overrated based on how hard it is to get a good score, not on a clear basis.

That is a problem. hard charts with lower average difficulties usually result in good score and OK acc, but actually to master them you need to enter a much higher level in order to manage those extreme hard part. This cause many underrated cases in the list and we need to pay more attention on the maximum difficulty of a chart to judge its rate.
Same thing also happened in 4k. Eg: Wu Xuan Lan 4k Hard and Insane Techniques 4k Hyper


Speed change and large amounts of LNs makes difficulty rise. We're so Happy obviously easier than other two charts due to this reason.
LNs patterns and speed change should also be considered.

Good Job!

Fudgyking wrote:

The issue here is that DT doesn't increase the score multiplier in mania and that the score is capped at 1 million. This means that even if you make a better score pp-wise with DT and do not beat your standard speed highscore, which is worth less pp, nothing will happen.

This gives an "unfair" advantage to those who didn't make a good standard speed score on a map and can play it with DT instead which gives more pp due to the increased difficulty of the map. Indeed, there are players who could make a decent score with the DT mod, but won't get anything out of it even if it is way harder than in normal speed because they already have a good standard score.

I see no way of solving this issue and keeping the score limit at 1 million at the same time unless it is possible to count only the best performance on a song instead of the best score.

Fafeluke wrote:

How about HD/FL ?
Those mods needs considering too to be honest.

Aqo wrote:

Fafeluke wrote:

How about HD/FL ?
Those mods needs considering too to be honest.

On every single map I played with HD I had the same result as nomod; I don't play HD regularly and have not much experience with the mod but it feels like it makes no difference at all difficulty-wise.

Can't speak for FL since I don't play it but I know people who play FL often end up finding it easier than nomod.

Xcrypt wrote:

rank harder maps.

[TaikoTori] wrote:

Why exactly do we discuss DT usage right now? I feel like that is a point being very unimportant right now. What we should concentrate on is why certain converted maps give such an big amount of pp, and why certain mania-maps are considered easy while they are actually hard and vice versa.

The problem with the converted maps such as Extreme Fantasy might be the note density. I agree to an certain extend that yes, the note density on those maps are higher than on other maps, but note density does NOT equal difficulty.

Same thing with some of the mania-maps, especially Wintersun for example, that (in my opinion) should be placed a lot higher than it already is.
There are certain patterns that do not seem hard, but are really hard in reality due to how exactly they are placed. For example http://osu.ppy.sh/b/330588 [Eien] as a Longnote training map, is very, very easy in terms of noteplacement, but this only counts if you are good with Longnotes in the first place.

In my humble opinion, there are two options:

a) Calculate alot of patterns into your algorithm, and i mean -alot- because there are alot.
b) Have a Mania-Difficulty-Rating-Team, as that may an option as long as there arent alot of maps and they could still catch up.

Atleast thats how i see it.

Xenlon wrote:

I agree,this is a great idea and a rating team would be really nice IF they dont abuse it for themselfes.

arcwinolivirus wrote:

I think fixing the DT bug will be a good start. DT is giving too much PP without score multiplier.

Tom94 wrote:

arcwinolivirus wrote:

I think fixing the DT bug will be a good start. DT is giving too much PP without score multiplier.

It's not a bug. DT is considered as of now, that's intended behaviour. Whether that will change isn't clear yet.

October Scream wrote:

This thread in a nutshell: ppv2 is crap.

DT should not be considered if it should give extra pp or not. It should already not be giving extra pp. The only way it could be remotely considered if it was to give extra pp when you're in the top 8 of a song. The only ones I've seen using DT are the top 10 O!M players, and their non DT scores are a hell of a lot better than there recent DT scores. It would have been best to stick with a broken ppv1 I would say since it didn't give any extra pp to players who used mods (But the star system is broke, which I will talk about in the next paragraph).

October Scream wrote:

And on the topic of Auto Converts; why only count certain auto converts, and deny a team of O!M map judges? The method of suggesting which maps are giving too much pp for their difficulty from other players to you is almost the same as having a map judging team; the only difference is that it's the final opinion of one person instead of multiple people, which is far more accurate than one person or a broken star system.

Drace wrote:

Tom, looking over the list again, it will be impossible to make an even semi-accurate algorithm with only strain values. Its like judging a sport's team on only their muscle mass...

Drace wrote:

You're gonna have to set up a pattern library and put maps against it for any hope of this being successful.

Drace wrote:

Also a difficulty rating team won't be hard to run and manage at all. The only thing is it might be slightly challenging to get it started.

Tom94 wrote:

Of course nothing stops you or anyone from providing a list of easy and hard patterns.

Tom94 wrote:

Everything is done automatically, which ensures, that every map gets the exact same treatment.

Drace wrote:

Also a difficulty rating team won't be hard to run and manage at all. The only thing is it might be slightly challenging to get it started. It'd just be an extra step between bubble and rank, it will more accurate than any algorith, and a number of members will make it not biased. There's also no reason for this to take months haha

October Scream wrote:

Tom94 wrote:

Everything is done automatically, which ensures, that every map gets the exact same treatment.

Weren't we trying to avoid something like that before? Pretty soon, some people will start saying that what you're using is broken as well.

October Scream wrote:

Tom94 wrote:

Of course nothing stops you or anyone from providing a list of easy and hard patterns.

This is like what you said to me about opinions. For some people, they find jackhammers incredibly easy, but stairs hard, vice versa.

October Scream wrote:

^ Get it started sooner and it will work easily. There should be multiple rules for applying though, like being above rank #200, skill in modding, have made a map, and/or are a BAT member. Even some low rank people could help by testing out easy maps and such. How a process could work is first, the map should be qualified. Once that happens, about 7 users should play each diff (not all having to be the same players each diff) with no mods, a vision mod, or hard rock (No key mods, random, or easy mods), and should get a C or greater. They would use a forum to fill out that would judge by a scale of 1-25 or any other number on things such as patterns, note frequency, chords, freezes, etc. They'd submit their opinion to Tom94 (Since he's running it mostly ATM) and then a simple math equation could be done to find an average of everyone's submitted fourm.

Tom94 wrote:

There won't be any rating team, due to the immense amount of time that'd be required to manage it, sorry. Give the algorithm chances, write which maps you think are rated wrong as some others already did.

I already have ideas on how to circumvent those high-density patterns from being weighted too much.

xxbidiao wrote:

Imperishable night 2006 is underrated. Its later half is crazy.

xxbidiao's thought wrote:

Imperishable night 2006 is underrated. Its later half is crazy. That's why I can't get good score on this song amongst other song of same "difficulty".

Tom94 wrote:

I already have ideas on how to circumvent those high-density patterns from being weighted too much

October Scream wrote:

Tom94 wrote:

Of course nothing stops you or anyone from providing a list of easy and hard patterns.

This is like what you said to me about opinions. For some people, they find jackhammers incredibly easy, but stairs hard, vice versa.

October Scream wrote:

Tom94 wrote:

Everything is done automatically, which ensures, that every map gets the exact same treatment.

Weren't we trying to avoid something like that before? Pretty soon, some people will start saying that what you're using is broken as well.

October Scream wrote:

Drace wrote:

Also a difficulty rating team won't be hard to run and manage at all. The only thing is it might be slightly challenging to get it started. It'd just be an extra step between bubble and rank, it will more accurate than any algorith, and a number of members will make it not biased. There's also no reason for this to take months haha

^ Get it started sooner and it will work easily. There should be multiple rules for applying though, like being above rank #200, skill in modding, have made a map, and/or are a BAT member. Even some low rank people could help by testing out easy maps and such. How a process could work is first, the map should be qualified. Once that happens, about 7 users should play each diff (not all having to be the same players each diff) with no mods, a vision mod, or hard rock (No key mods, random, or easy mods), and should get a C or greater. They would use a forum to fill out that would judge by a scale of 1-25 or any other number on things such as patterns, note frequency, chords, freezes, etc. They'd submit their opinion to Tom94 (Since he's running it mostly ATM) and then a simple math equation could be done to find an average of everyone's submitted fourm.

xxbidiao wrote:

Tom94 wrote:

There won't be any rating team, due to the immense amount of time that'd be required to manage it, sorry. Give the algorithm chances, write which maps you think are rated wrong as some others already did.

I already have ideas on how to circumvent those high-density patterns from being weighted too much.

I believe other people like Drace said what I just want to said below.

I'm not saying the algorithm is "wrong". It's always "correct" and reflect one kind of difficulty - based on a certain norm.
Actually, we everyone have a norm. For example, when I say that

xxbidiao wrote:

Imperishable night 2006 is underrated. Its later half is crazy.

I implied that

xxbidiao's thought wrote:

Imperishable night 2006 is underrated. Its later half is crazy. That's why I can't get good score on this song amongst other song of same "difficulty".

You should get what I mean - It's nonsense at all to the community, because other people may find it easy or even harder - in one word, have different thoughts. And it's all about the norm we set up during our years of rhythm gaming experience.

For instance, I started rhythm gaming on Taiko no Tatsujin and DJMAX / DDR, which is all note-biased. Without good experience on slider-style song, I'm good on IIDX style songs, but poor at O2-style songs.
However, Fate_Yakumo, another player who is of O2Jam expert, find these slider songs easy - He can S a song of O2Jam style that I can't pass. When coming to IIDX songs, instead of one-side winning, we are just at the same level, something 70% vs 30%.
You guess what? I always overrate slider-biased song, and he always overrate note-biased ones.

This example reflect 2 common norm in 7/8K: IIDX norm and O2Jam norm. There are also many more, and every key amount have multiple norms. Sadly they are not enough compact with each other. And whether you admit it or not, you are setting up your own norm.

Just like what you have said,

Tom94 wrote:

I already have ideas on how to circumvent those high-density patterns from being weighted too much

You implied that "I'll degrade these players who is good at high-density pattern playing" (To some extent moving away from IIDX norm to O2Jam norm).

So this is why I keep saying that "This difficulty number does nothing with actual difficulty at all" and "We should try to admit there are weighted songs".
Considering the diversity of mania players, however you modify your algorithm, there are always some people winning, and some people losing.

My suggestion is, however hard it is, and however offence you made, a certain norm should be set up. Whether you choose BMS norm (note biased, density and jackhammers counts ) or O2Jam norm (slider biased, coordination counts) or even DDR norm (hand speed counts) or a mixture of them, Don't hesitate to say that "I mean to hurt you" because what norm you choose you hurt some people who have ability that the norm doesn't emphasize on. What's more, the process of letting players finding their weighted song is fun, isn't it?

-----------------

For the DT stuff, I actually have more to say.
Why do people come here to protest about DT score problem?
Actually it's not about scoring at all (People can play DT on the first run).
DT getting higher weight neglects the design of the osu!mania system.

You may have noticed that osu!mania has the most strict judgment with MAX even on OD0 (Actually OD0 and OD10 has no difference on timing of MAX). This is intended because most people agree that osu!mania should be different to other modes by having judgment emphasized, not combo. (o!m once calculate score just like o!standard and current scoring system is later decided.) The philosophy inside is that 1 more MAX is good.

However having DT heavier weighted, though completely reasonable, is going against this. People may achieve a just-so-so S score to win more PP then a SS with 90% MAX ratio. This put osu!mania far away from it's emphasize on judgment, to the extent that people just DTing every song regardless of judgment, and people no longer care about MAX ratio. This is surely what people don't want to see. (Feature requests on DT bonus score has been rejected multiple times both by faculty and players)

So unless ppy somewhat change the mechanism to make DT version of the beatmap a whole new one and have its separate scoreboard, I suggest no bonus for DT for now.

Thing O Doom wrote:

Sorry to bother, I'm a noob and don't really have a place in these discussions,

but recently since the new system came out, I've been playing harder maps (That I get a terrible score in).
I'll get like a B or C on a map, then come back to it later and beat my personal highscore on the map, then I visibly watch my ranking drop when the screen flash updates in the song stats screen. I know the system is unstable but I don't remember this being a thing in the old system, correct me if I'm wrong. This makes me not want to play online and improve w.w""

milky228 wrote:

Thing O Doom wrote:

Sorry to bother, I'm a noob and don't really have a place in these discussions,

but recently since the new system came out, I've been playing harder maps (That I get a terrible score in).
I'll get like a B or C on a map, then come back to it later and beat my personal highscore on the map, then I visibly watch my ranking drop when the screen flash updates in the song stats screen. I know the system is unstable but I don't remember this being a thing in the old system, correct me if I'm wrong. This makes me not want to play online and improve w.w""

This happened to me in the old system and I put it down to people that were lower ranks than me moving up and therefore putting me down, I assumed this only updated after i'd played my first song. There also seems to be a bug on other gamemodes where peoples PP is going down after beating there personal best, I expect this will be patched soon if it's also affecting o!m.

PyaKura wrote:

Hey,

Just checked the page of my new mapset that I submitted yesterday (http://osu.ppy.sh/b/380286&m=3).
The diff rating went from Easy and Normal to Hard and Insane for the first two diffs. Does this have anything to do with the changes being made with the diff rating and ranking system ?

milky228 wrote:

Thing O Doom wrote:

Sorry to bother, I'm a noob and don't really have a place in these discussions,

but recently since the new system came out, I've been playing harder maps (That I get a terrible score in).
I'll get like a B or C on a map, then come back to it later and beat my personal highscore on the map, then I visibly watch my ranking drop when the screen flash updates in the song stats screen. I know the system is unstable but I don't remember this being a thing in the old system, correct me if I'm wrong. This makes me not want to play online and improve w.w""

This happened to me in the old system and I put it down to people that were lower ranks than me moving up and therefore putting me down, I assumed this only updated after i'd played my first song. There also seems to be a bug on other gamemodes where peoples PP is going down after beating there personal best, I expect this will be patched soon if it's also affecting o!m.

Tom94 wrote:

Dropping your pp by beating your score is extremely unlikely in o!m. You'd need to get just barely more score with a lot worse accuracy for that to happen.

The rank drop most likely occured due to other people passing you, as it always has been already.

Thing O Doom wrote:

Ok, I think I can debunk this now, this morning I played a map and it shows my rank as ~7990, this was an unranked map so nothing was affected, this being just to make sure I didn't get overly surpassed while sleeping.

Full Tablet wrote:

Thing O Doom wrote:

Ok, I think I can debunk this now, this morning I played a map and it shows my rank as ~7990, this was an unranked map so nothing was affected, this being just to make sure I didn't get overly surpassed while sleeping.

I don't think playing an unranked map triggers a rank refresh.

October Scream wrote:

http://osu.ppy.sh/u/86188 <-- This is bullshit. No one would play dt back then, now everyone is, and he's sprung up to the top because of key mods and dt. I don't care as much about dt, I care about the key mods coming back. So many players are better than that, and I don't know why they can't be up there?

October Scream wrote:

http://osu.ppy.sh/u/86188 <-- This is bullshit.

No one would play dt back then, now everyone is, and he's sprung up to the top because of key mods and dt. I don't care as much about dt, I care about the key mods coming back. So many players are better than that, and I don't know why they can't be up there?

Tom94 wrote:

October Scream wrote:

http://osu.ppy.sh/u/86188 <-- This is bullshit.

No one would play dt back then, now everyone is, and he's sprung up to the top because of key mods and dt. I don't care as much about dt, I care about the key mods coming back. So many players are better than that, and I don't know why they can't be up there?

Can you name any players who are better than Staiain at stepmania style maps? What gives you the right to say which kind of skill is worth the most? I can understand if you disagree, that's totally valid, but try being less offensive. It's not like your opinion is worth more than that of other's.

#1 Judging from the general opinion here the DT bonus for pp will get removed for now. It might return when there has been found a solution regarding nomod plays overwriting DT, but as it is now it doesn't make much sense to have DT give a bonus.

#2 Keymods don't decrease pp gain at the moment. This will change.

October Scream wrote:

Right now, I can compare Furry. There are more players out there though, and the only reason Furry isn't high is because he is playing unranked maps it seems to me.

October Scream wrote:

I've spectated him, and he's good. May pass Staiain. He has dt ranks on autoconverts, but 4k, just like Staiain.

AlexDest wrote:

Any videos?

October Scream wrote:

AlexDest wrote:

Any videos?

His scarlet rose dt pass with 4k.

ZXCV wrote:

If you really think Scarlet Rose DT pass is impressive, then I suggest you go watch more videos, especially Staiain's. Your basis on what defines a "pro" player is severely flawed.

I can clear Scarlet Rose HDDT on 7K, and so can tons of other players. Does this make us all top #50 material?

ZXCV wrote:

Good, so you agree that passes on DT autoconverts are completely obsolete when it comes to gauging a player's skill.
Case closed.

October Scream wrote:

ZXCV wrote:

Good, so you agree that passes on DT autoconverts are completely obsolete when it comes to gauging a player's skill.
Case closed.

Didn't we decided that already? That's why it's not giving pp now.

October Scream wrote:

I find it moronic how you guys are mainly blaming me and aiming most threats at me. Many other users have spoken about this, and are tired of it.

I'm just the one that spoke up. This is all of their opinions plus mine, so go scout them out and threaten them.

Bites wrote:

October Scream wrote:

I find it moronic how you guys are mainly blaming me and aiming most threats at me. Many other users have spoken about this, and are tired of it.

I'm just the one that spoke up. This is all of their opinions plus mine, so go scout them out and threaten them.

october scream, you are ignoring my post and every point we have made.

THIS IS FROM YOUR USERPAGE

if you are going to spearhead a movement to get keymods unranked, you should expect that we will confront you about these issues we have with your argument. we have not made a SINGLE threat, nor are we blaming you. we are simply asking you to explain yourself and refute what we are saying.

until you are willing to do so, this is the last i have to say on the matter

October Scream wrote:

I'm aware of that petition. I started it because no one else would.

If you want me to explain, fine. I didn't like it when key modders would pass other players, but I have no use of that petition now actually since it's been fixed already.

Bites wrote:

this is the last i have to say on the matter

October Scream wrote:

I'm aware of that petition. I started it because no one else would.

If you want me to explain, fine. I didn't like it when key modders would pass other players, but I have no use of that petition now actually since it's been fixed already.

Bites wrote:

both DT and keymod converts will return to the PP algorithm once tom94 figures out what to do with them, and this has been stated explicitly

Halogen- wrote:

I am glad that DT and key-mod conversions will return to the rankings at the least. I will agree as a osu!mania player who could have probably taken Staiain's speed route to reach the top 10-20 despite not being able to proficiently play above 6k -- the PP scaling could have definitely been improved a bit.

I wanted to make a suggestion about separating performance points by keys in osu!mania since you're practically playing different games when you add keys, but... one of the toughest parts about scoring in osu!mania is that most players who can play higher key beat-maps are (usually) also capable of playing lower ones as well, making separated scoreboards/ranks by key a bit redundant for those who can play 8k all the way down to 4k.

PyaKura wrote:

Halogen- wrote:

I am glad that DT and key-mod conversions will return to the rankings at the least. I will agree as a osu!mania player who could have probably taken Staiain's speed route to reach the top 10-20 despite not being able to proficiently play above 6k -- the PP scaling could have definitely been improved a bit.

I wanted to make a suggestion about separating performance points by keys in osu!mania since you're practically playing different games when you add keys, but... one of the toughest parts about scoring in osu!mania is that most players who can play higher key beat-maps are (usually) also capable of playing lower ones as well, making separated scoreboards/ranks by key a bit redundant for those who can play 8k all the way down to 4k.

I'm a 7K player, I suck at everything which is not 7K, sorry.

October Scream wrote:

I've spectated him, and he's good. May pass Staiain. He has dt ranks on autoconverts, but 4k, just like Staiain.

Bites wrote:

both DT and keymod converts will return to the PP algorithm

falkon13 wrote:

So just to check this at this moment in time playing a map with a keymod will give you 0pp?

Staiain wrote:

To me it just looks like October Scream is just a slow pp farmer that wants to ruin the game and stop actual skilled people from being able to get good ranks, rather than just suck it up and improve his own skill

Halogen- wrote:

Staiain wrote:

To me it just looks like October Scream is just a slow pp farmer that wants to ruin the game and stop actual skilled people from being able to get good ranks, rather than just suck it up and improve his own skill

Ouch.

Although, it's a quite deserved response considering that this thread was quiet until he called the situation bullshit and had a hissy-fit, haha