Momiji wrote:
Involving scorev1 or scorev2 at all is a garbage in, garbage out situation
Momiji wrote:
Involving scorev1 or scorev2 at all is a garbage in, garbage out situation
That's basically what the score system was lmaoB1rd wrote:
The only real solution is to dispense with the PP system altogether. Then instead of people scrambling to get the top PP plays, they will be scrambling to get the best plays as defined by themselves and their peers. In other words, an algorithm challenges the human mind to exploit it. Having no PP system uses the human mind itself as a judge of performance, which is vastly superior to any PP system.
#downwithPP
what you fail to realize is that we don't have data to throw into the equations, so we're stuck with generic forms. unless you have a splendid idea...winber1 wrote:
You can come up with equations all you want acting as if you are coming up with good ideas, but unless you actually show hard data of a equation that actually works reasonably well with hundreds of edge cases, there will only be skepticism.
Any system which hinges on the player-base is bound to encounter popularity issues and outlier issues, sooner or later you will find the outliers just like how people have found outliers in the current pp system.then give us a new pp calculation if you're just going to try and shit on others' ideas
I honestly think the only real solution is to improve the difficulty calculator, or invest some deep learning data scientist to make a computer model that will be able to "predict" what the player base will think of the difficulty of said maps (and any anomalies can be reported and sent to the machine to improve its understanding of difficutly).are you bullshitting lul
Lol, it's impressive how you gave something so obvious and so good that still seemingly 0 people have said before. The disproportions of PP are literally mirrored in the star rating system, of course excluding how low IQ acc PP is (way too separate issue). Thumb upwinber1 wrote:
Literally the only real solution is to make the difficulty calculator better. All other cases have too many edge cases where things can just go wrong. Ranking-based system was ppv1 and it gave too much pp for anime maps because those were by far played much more. Of course scarlet rose was just a huge name in general as well as a few other well known "difficult" maps, so those were relatively adequately weighted for how difficult it was. You could even argue to have a combination of ppv2 and ppv1, but then things like haitai would be even more broken and overly weighted. You can come up with equations all you want acting as if you are coming up with good ideas, but unless you actually show hard data of a equation that actually works reasonably well with hundreds of edge cases, there will only be skepticism.
Hate to break it to you but he's bullshittingMomiji wrote:
Lol, it's impressive how you gave something so obvious and so good that still seemingly 0 people have said before. The disproportions of PP are literally mirrored in the star rating system, of course excluding how low IQ acc PP is (way too separate issue). Thumb upwinber1 wrote:
Literally the only real solution is to make the difficulty calculator better. All other cases have too many edge cases where things can just go wrong. Ranking-based system was ppv1 and it gave too much pp for anime maps because those were by far played much more. Of course scarlet rose was just a huge name in general as well as a few other well known "difficult" maps, so those were relatively adequately weighted for how difficult it was. You could even argue to have a combination of ppv2 and ppv1, but then things like haitai would be even more broken and overly weighted. You can come up with equations all you want acting as if you are coming up with good ideas, but unless you actually show hard data of a equation that actually works reasonably well with hundreds of edge cases, there will only be skepticism.
bruhrepr1se wrote:
what you fail to realize is that we don't have data to throw into the equations, so we're stuck with generic forms. unless you have a splendid idea...winber1 wrote:
You can come up with equations all you want acting as if you are coming up with good ideas, but unless you actually show hard data of a equation that actually works reasonably well with hundreds of edge cases, there will only be skepticism.Any system which hinges on the player-base is bound to encounter popularity issues and outlier issues, sooner or later you will find the outliers just like how people have found outliers in the current pp system.then give us a new pp calculation if you're just going to try and shit on others' ideasI honestly think the only real solution is to improve the difficulty calculator, or invest some deep learning data scientist to make a computer model that will be able to "predict" what the player base will think of the difficulty of said maps (and any anomalies can be reported and sent to the machine to improve its understanding of difficutly).are you bullshitting lul
Facial detection maybe. Identification and landmarking (the latter in particular) are both still rather difficult for computers not to mention in many cases you would like it to be done in real time which isn't very practical in a computational sense. If you have a face in the standard straight on chin level pose then it's usually pretty easy for a computer. Anything more than a slight variation in pose reduces the success rate dramatically even when trained on data with high variation in poses.winber1 wrote:
extremely high if not 100% success rate for a particular "type" of facial image (which of course is because facial is more important than beatmap difficulty out in the real world).
What’s even the point of this being an online game if there is no ranking/scoring system? Might as well just play offline 100% of the time as my plays won’t even matter and I’ll get slightly better performance (game stability) from playing offline.B1rd wrote:
The only real solution is to dispense with the PP system altogether. Then instead of people scrambling to get the top PP plays, they will be scrambling to get the best plays as defined by themselves and their peers. In other words, an algorithm challenges the human mind to exploit it. Having no PP system uses the human mind itself as a judge of performance, which is vastly superior to any PP system.
#downwithPP
If only there was a way to judge the skill or players and plays without an arbitrary number...Edgar_Figaro wrote:
What’s even the point of this being an online game if there is no ranking/scoring system? Might as well just play offline 100% of the time as my plays won’t even matter and I’ll get slightly better performance (game stability) from playing offline.B1rd wrote:
The only real solution is to dispense with the PP system altogether. Then instead of people scrambling to get the top PP plays, they will be scrambling to get the best plays as defined by themselves and their peers. In other words, an algorithm challenges the human mind to exploit it. Having no PP system uses the human mind itself as a judge of performance, which is vastly superior to any PP system.
#downwithPP
I know “removing ranking system” sounds like a good idea to people but this only stems from long periods of disgruntlement from the flaws in the systems. Osu is a competitive game and many players (myself included) find the competitive aspect to be a massive draw. If there was no ranking system I tbh wouldn’t play half the songs on Osu as they aren’t my preferred music choice, Also I’d care a lot less about getting better as “what’s the point” and I’d play this game a lot less than I currently do.
TL:DR: Ranking systems are a major aspect of players desire to play games and shouldn’t be removed just to appease the hipsters
there isn't... LoL uses MMR, Dota2 uses ELO, CS:GO uses Glicko2, ESEA uses RWS, CEVO uses efficacy, Hearthstone uses MMR, SC2 uses ELOB1rd wrote:
If only there was a way to judge the skill or players and plays without an arbitrary number...
Then please enlighten me, because I don't have a single clue of what it could be while you seem to have a pretty good ideaB1rd wrote:
There is.
B1rd wrote:
There is.
"I have discovered a truly marvelous proof of this, which this margin is too narrow to contain." -Pierre de FermatBoaring wrote:
Then please enlighten me, because I don't have a single clue of what it could be while you seem to have a pretty good idea
Not sure about Stepmania but I know Beatmania uses the Dan System, where players are required to pass a course consisting of 4 maps that each test different skillsets of the game. If they successfully pass the course they rank up and they do this up until they reach Kaiden, which is the highest rank you can acquire.B1rd wrote:
As far as I know, other rhythm games like Stepmania or Beatmania don't have any rankings like osu! does, so you can't complain that it would ruin the game.
A true ranking system should be well ordered. What you described is only a partial order and that's assuming it's even transitive. I'm not saying you're wrong about removing the ranking aspect, just that what you described does not produce a ranking.B1rd wrote:
<snipped>
this leads to subjectivity as to what constitutes a good player. especially with the number of players complaining that DT is not a skill, and jumps aren't skill (which... is retarded). while the pp system isn't perfect it clearly shows that cookie is better than i amB1rd wrote:
Well it shouldn't be that hard to figure out considering I just told you what the method was. That method being your brain. I don't need to look at the rankings to say that Cookiezi is the best player. I can look at Freedom Dive HDHR and multitude of other scores he was made and use that to determine he is the best player.
Of course it's not easy to neatly rank people from 1-100,000,000, but that's how it should be, because the game is so complex with all the skills that is very difficult to make an objective ranking of overall skill. You can't look at any one player in the top 100 and say for sure that he is worse than all those above him and better than all those below him, because the PP system only measures a specific skillset. And it's not hard even without the PP system to rank players according to a few singular aspects. However we shouldn't pretend that those few aspects that are measures by the PP system create an objective ranking of skill, and then only compete in those few skills, such as we do now.you just wrote that you determined cookiezi to be the best player
Just because you get rid of the PP system, doesn't mean it gets rid of the competitive aspect of the game; it's just that instead of competing on farm maps like Daidai Genome and Haitai, they will be competing on actual good maps. As far as I know, other rhythm games like Stepmania or Beatmania don't have any rankings like osu! does, so you can't complain that it would ruin the game.then what constitutes a good map? whatever answer you give is subjective and everyone's opinions vary, you can't just redo the pp system to have "actual good maps" and satisfy everyone. undoubtedly someone will say "your 'good maps' are shit, give me the old pp system"
I really meant that you can judge player skill without a ranking system, though you can still rank players based on certain criteria, you just don't get to universalise your arbitrary ranking system. Although you could say that judging people by their scores is still judging people based on numbers, but no reason to get overly semantic.chainpullz wrote:
A true ranking system should be well ordered. What you described is only a partial order and that's assuming it's even transitive. I'm not saying you're wrong about removing the ranking aspect, just that what you described does not produce a ranking.B1rd wrote:
<snipped>
After having experienced a rhythm game arcade environment and the mentality it breeds I personally think osu would be better off without rankings. There is no mid-map retry in arcade versions and people mainly play to beat themselves as opposed to others. The mentality is never "yay my rank went up." Rather, it's something more like "holy shit I finally did it."
The Dan system would probably be alright too since, again, it's more about beating yourself than trying to rank you against everyone else.
Yeah, because player skill is somewhat subjective. There are so many skills in the game that it's hard to say that any certain skill is better. Although you can have a in-depth discussion about that.repr1se wrote:
this leads to subjectivity as to what constitutes a good player. especially with the number of players complaining that DT is not a skill, and jumps aren't skill (which... is retarded). while the pp system isn't perfect it clearly shows that cookie is better than i am
you just wrote that you determined cookiezi to be the best player
i beg to differ. the top ranks have their variety of skillsets, especially rohulk and idke
do i think that pp overvalues certain aspects of the game? yes. do i think that pp measures ONLY one skillset? no.
then what constitutes a good map? whatever answer you give is subjective and everyone's opinions vary, you can't just redo the pp system to have "actual good maps" and satisfy everyone. undoubtedly someone will say "your 'good maps' are shit, give me the old pp system"
i tried searching for beatmania and stepmania leaderboards and didn't find anything. osu is the only game i know of with a leaderboard this hardcore and that's what sets it apart.
where have i made that claim? i only mentioned that too many bitches whine about DT without recognizing that it actually does take skill to play. and i'm not even a DT player.B1rd wrote:
What you're saying is that you want the game to prioritise certain aspects of the game, like aim and speed, over others. Well if people don't want to recognise skills like DT, then so what? You shouldn't try to force other people to respect you.
I think people who complain about DT aren't actually saying that DT takes no skill, but rather it takes no skill relative to other skills that give the same PP, which is pretty much true. You already admitted that the PP system isn't an objective criteria of skill as you denied Cookiezi to be the best when he has the most PP. So you should be all for getting rid of the PP system so that players with unique skillsets aren't overshadowed by players with slightly more PP.where have i made that claim? i was pointing out that you wrote that you can't determine who's the best, but you also write that cookie was the best
The problem with the PP system is that is is so pervasive and affects people who don't care for it. Yeah, I do believe that some maps are objectively better than others; I don't regard copy paste PP maps to be as good as maps by Lan Wings for example. And it's the result of the PP system that people who don't like PP mapping and aren't interest in competing in that specific skillset are marginalised.just about all of the top players don't care for pp. they play for fun and pp is gained on the side.
You have to realise that mapping is basically like a market place. Mappers like to make maps that are popular and well received by others. Therefore, mappers generally cater to what the community likes. With the existence of the PP system, the majority of players play for PP, therefore they play maps which give lots of PP and have skillsets catered towards the skills which give PP. Therefore, there is a large demand for PP maps, so the majority of mappers will make PP maps, and even the mappers who don't explicitly map for PP will have their map design largely influenced by the meta and the skillsets of the majority of players.that's simply not true... there are mods and ranking criteria for a reason. i could pp map for this map but it won't get ranked because it doesn't meet criteria simply because it doesn't fit the song
That's why you don't see many AR7 or 8 maps these days, and in fact its quite difficult to get a map ranked that goes against the meta.because ar7 and ar8 are where players still learn the game mechanics, and ar9 is where reading actually begins
That's why it's not valid to say that if the PP system were to go away that what is currently regarded as PP maps will disappear and the players will be left wanting. My argument is, in essence, that the PP system means that players play maps that give PP rather than what they consider good or fun to play. If it were to go, then players would play and compete on maps that they and the community in general thought were good and fun, and therefore the demand and thus supply of maps would shift towards what the community regarded as good and fun. It would also create a more diverse range of maps that focused on larger degree of skillsets, and there wouldn't be a ranking system that arbitrarily chose a few skills to be the most important ones.and if the community thinks that copypasta jump maps are fun?
And yes I said that Cookiezi is the best player - because he is. No one can match Freedom Dive HDHR, no one can match a plethora of his other plays either. He has amazing aim, the best streaming consistency in the game - not only that, he has great reading, and an exceptional ability to play difficult patterns and has very good acc. You might find players that are better than him in one or two aspects, but there is no one who is so exceptionally good in so many areas. He is in a league of his own.again, i didn't confirm or deny cookie being good or not.
Apology accepted.repr1se wrote:
i'm sorry
Are we talking about the Buddhist type of enlightenment? As a Christian I'd advise you not to let your soul be corrupted by heathen religions.repr1se wrote:
i have been enlightened.
Realise what?repr1se wrote:
i now realize that
Agreed, that was my main point.repr1se wrote:
the pp system is completely broken and should be removed as a whole for the benefit of us that enjoy maps that don't get much pp.
Okay bye.repr1se wrote:
jk later
Wait, who is this directed at?repr1se wrote:
bitch
Yeah, I tried solving that problem in my edit. Would really like to hear what you think of it since you have experience in trying to create a statistical PP system.Full Tablet wrote:
Using the average score is not a good indicator of the difficulty of the map, easier maps tend to have relatively few good players attempting plays (mostly people who hunt for score ranks only) and a lot of players who aren't good at the game in general.
Wow! Really nice. Have you made a post explaining this algorithm anywhere, or do you have the code on github or something?Full Tablet wrote:
Here is an example of another ranking system (for osu!mania) based on statistical data only:
t/329678
It is based on calculating "Difficulty Curves" for each beatmap ("Player Skill" required to achieve certain score in the beatmap), and simultaneously fitting and those curves and the "Player Skill" values each player has (based on the scores they have set). It requires an enormous amount of computing power to calculate, though (here it took several months to calculate the rankings with only ~2000 players, with the time required to calculate increasing approximately quadratically with the amount of players and linearly with the amount of beatmaps).
Cool! But I think statistical measures will do better at capturing what "difficulty" is than direct measures because the direct measures will have to deal with the near-infinite complexity of the real world.repr1se wrote:
there are other things that the pp system leaves out, like unstable rate and overall stability of the player. i'm going to propose some generic equations where specific numbers can be filled in later yada yada yada
this proposition encourages stability in accuracy and increases the amount of pp you get from high OD (like HR):
https://gyazo.com/ab41ade7e2796cafc99fd3c5b2f6276b
this proposition reduces the amount of pp awarded in jump maps, to align with stream and technical maps, as well as reducing the pp of slider-only maps:
https://gyazo.com/96a13706bf510199adfae034a234e471
this proposition increases the pp value of technical maps.
https://gyazo.com/591ea23d810d681155da6e7d93d88b71
do i think these changes have to be implemented? no. but i think it does address the concerns of people whining about the current pp system.
Just by picking some arbitrary values and judge whether they look about right. You can fiddle with the numbers here.repr1se wrote:
k_1 and k_2 are constants that determine the range of PP values.
-how will this be determined?
Huh, interesting. That might work better than using ScoreV2 since PPv2 seems to scale accuracy better. Correct me if I'm wrong, but my impression is that the difference between 99 % and 100 % acc is a lot more on PPv2 than on ScoreV2. This is as it should be because going from 89 % to 90 % accuracy is a lot easier than going from 99 % to 100 %.Momiji wrote:
Mine solution - Do that, then turn ppv2 into SCORE, and then filter ppv2 values through the ppv1 amplification
I really like the idea of basing PP values on player ratings and would like to see it developed further. It might just work.ManuelOsuPlayer wrote:
Or a PP system made by players ratting. Where players rate maps under their skills to don't exploit it. For example my lowest toprank PP It's 89pp. I should be able to rate maps what i can get an S to give from 1pp to 88pp.
Meh. I like having a quick and fairly reliable way to determine how good random players on the forums are at aim. It's a pretty good lowest estimate for how good someone is at aim. If someone is 6K PP, I know they're better than me at aim; but if someone is 1K PP, I don't know whether they're better than me at aim.B1rd wrote:
#downwithPP
Hmpfh. There's no contradiction between trying to gain PP and enjoying the game. Likely there are very many players who think osu! is fun because they like trying to get more PP than their neighbour.Husky wrote:
Remove pp.
pls enjoy game
Mio Winter wrote:
[i]
make it (MapRank/N)*(1/ln(sameStuff))?
kricher wrote:
Mio Winter wrote:
[i]
make it (MapRank/N)*(1/ln(sameStuff))?