So first im REIKARI and im quite new to be honest but i wanted to share some Ideas of mine how I THINK the PP System would be "fairer". Ofc im well aware that alot people won´t be impressed and maybe think it´s garbage. It´s a harsh concept but i tried to balance it as good as i could. I guess the biggest problem is that im not a Pro Player and im not so deep into the system at all. So my sight on the system is quite limited. I also didn´t use decimal numbers.


image upload

I tried to use simple numbers just so it´s easier to understand my thoughts. I wanted to make sure the players will get rewarded for: Map lenght, Numbers of Notes, Accuracy, Combo and Mods.
Its really a harsh unpolished concept but i wanted to share my thoughts with you guys. Hopefully you get my Idea-/thoughts.

Just for comparison, here´s what a 4* would look like (Just remember, numbers aren´t fix)


(PP for Acc is increased, PP for lenght is a bit increased too)
The reward for using Hidden or other Mods are really controversial right now because the map often changes from 4* to 5* but that´s why i called it a harsh concept ^^

I enjoyed thinking about it so that´s the only reason why i made this up. Maybe it helps some other people or the Team to give them some ideas or something.

Well that´s all for the moment, thanks for giving me your attention (and sorry for my grammar and stuff, english isn´t my native language^^)

you probably want to discuss that directly with the others at dev discord (including ppy) at #osu-performance channel
https://discord.gg/ppy

I could do that but im not very good at talking (language barrier) writing is easier for me. I just wanted to know what other people may think about this. If its total BS or somehow usable in some ways.

R E I K A R I wrote:

I could do that but im not very good at talking (language barrier) writing is easier for me. I just wanted to know what other people may think about this. If its total BS or somehow usable in some ways.

Rivals_7 meant writing, not talking.

My thoughts:

- Assigning each mode the same number % pp is not accurate. Depending on patterns, some maps will be harder or easier with one mod than another.
- If that 1* map is one hour long, that's 360pp to start with, which is very wrong
- This system gives more pp the more notes there are in a certain time. Do you ever wonder why 2010ish maps have so many triple note stacks?
- The system is sensitive to highest combo, like the current system. Therefore it still suffers from the middle miss issue, where a miss in the middle will make the map worth a lot less than a miss in the end or beginning. This makes maps with difficulty spikes in the middle much harder. The solution is to count the amount of misses, not combo, so a miss anywhere will be worth the same.

- Assigning each mode the same number % pp is not accurate. Depending on patterns, some maps will be harder or easier with one mod than another.

Yes, you are right - spacing, jumps and specific patterns are not included in this but is this really a bad thing? Mappers would have more freedom and abusing wouldn´t be so easy anymore. Ofc you still could abuse it for example tons of fast streams & multinotes. But the Maps would not be that popular anymore because not everyone is a streaming god and has enough stamina to stream 4 Minutes.
The difficulty is determined by the Star Rating so, if a song is harder or has more difficult patterns - well it has a higher difficulty. Simple as that ^^

- If that 1* map is one hour long, that's 360pp to start with, which is very wrong

The numbers i used are examples - nothing fix but i wanted to make it easy to understand but i still get you and why you shouted out at this point

- The system is sensitive to highest combo, like the current system. Therefore it still suffers from the middle miss issue, where a miss in the middle will make the map worth a lot less than a miss in the end or beginning. This makes maps with difficulty spikes in the middle much harder. The solution is to count the amount of misses, not combo, so a miss anywhere will be worth the same.

It is not like the current system, the player will be rewarded "easier" for his-/her efforts. it counts the highest combo the player reached - but even when you only get a low combo you still get rewarded and the difference between low, middle, high, fc isn´t that much anymore. Ofc it punishes you if you miss some notes but in other ways -->Accuracy.
What i tried to do here is reduce the the difference between low combo, middle, high and full combo. FC gives you still bonus % but lets say you have a really good play but you had unlucky combo breaks.

Like 99% + Acc, 3 breaks no misses. So you still get atleast the 20% (or more) multiplier at the end for your effort and not just nothing. As long as you get rewarded by a combo streak in any way, you cant balance the "miss in the the middle of the song" . You miss in the middle - okay max combo you can reach = 50% . So i really get what you´re trying to say. The only way that comes into my mind is a tolerance miss or a free miss in the middle. But that wouldn´t work..
Like..you reached the middle of the song-/map between 48% to 52% you can miss one note but it doesn´t really count and your combo doesn´t break either...doesn´t work

But thanks for your attention and your thoughts, i really appreciate it ^^

R E I K A R I wrote:

The difficulty is determined by the Star Rating so, if a song is harder or has more difficult patterns - well it has a higher difficulty. Simple as that ^^

This brings to the following point: There is no need to do what you are doing with the pp system. You are making things, way way more messed up than they should be. Do not make pp attempt to the star rating jobs, its only function is to combine the results of various maps in such way it is representative of your skill.

The model should be:

There is a map with patterns, and by analyzing the patterns, difficulty metrics are generated
There is a player, and by playing the map, the player generates a score
Based on the difficulty metrics generated and score generated, generate pp value(s)



Star rating should be looking at the map length, how mods affect patterns, map OD, CS, and all of the other stuff, and PP calculation should only get the calculated metrics and apply them to the resultant player score. Instead, the current system and yours attempts to fix pp by trying to take account stuff the star rating calculation fails to take account.

R E I K A R I wrote:

It is not like the current system, the player will be rewarded "easier" for his-/her efforts. it counts the highest combo the player reached - but even when you only get a low combo you still get rewarded and the difference between low, middle, high, fc isn´t that much anymore. Ofc it punishes you if you miss some notes but in other ways -->Accuracy.
What i tried to do here is reduce the the difference between low combo, middle, high and full combo. FC gives you still bonus % but lets say you have a really good play but you had unlucky combo breaks.

Or you can just count number of misses and make it a steep curve adjusted to emulate the difficulty similar of FCing a map.

Combo:
- Representative of player's peak skill
- Is subject to more chance, and as a result will produce varied results with every play
- Map's difficulty curve throughout time is based on this formula:

Misses:
- Representative of player's average skill
- Is a better determinant of player's overall skill since it should produce consistent results with every play
- Map's difficulty curve throughout time is based on this formula:

R E I K A R I wrote:

Like 99% + Acc, 3 breaks no misses. So you still get atleast the 20% (or more) multiplier at the end for your effort and not just nothing. As long as you get rewarded by a combo streak in any way, you cant balance the "miss in the the middle of the song" . You miss in the middle - okay max combo you can reach = 50% . So i really get what you´re trying to say. The only way that comes into my mind is a tolerance miss or a free miss in the middle. But that wouldn´t work..
Like..you reached the middle of the song-/map between 48% to 52% you can miss one note but it doesn´t really count and your combo doesn´t break either...doesn´t work

I hope you realize that combo causes enough issues and that we are better without it

Hmm well i get you but if we leave things how it is, there will be always maps that abuse patterns and almost all maps look the same. Your method even pushes pp abusive patterns even more. Mappers will figure out what they need to do and there we have the same maps over and over again.
I also try to solve this somehow, its not that these maps are bad but repetetive

Ofc my way to solve this isn´t flawless too. You could push the overall pp with lots of notes = streams & multinotes but honestly streaming is more difficult than jumpy patterns and if some mappers would try to abuse it..well i think the maps wouldn´t be that popular.

To solve the problem with combo well.. i have no answer to it. Like i said as long as combo is the main factor of gaining pp you can´t balance it . If you miss in the middle well thats it. That´s why i think my way would be better (as long as we keep combo).

Im sorry if my grammar or my statements are a bit weird or not quite understandable, its really hard to argument about something and you cant express yourself to the fullest xD

R E I K A R I wrote:

Your method even pushes pp abusive patterns even more. Mappers will figure out what they need to do and there we have the same maps over and over again.
I also try to solve this somehow, its not that these maps are bad but repetetive

I ask that you explain and give an example of how my method pushes pp abusive patterns even more since I fail to see how that is the case.

"There is a map with patterns, and by analyzing the patterns, difficulty metrics are generated
There is a player, and by playing the map, the player generates a score
Based on the difficulty metrics generated and score generated, generate pp value(s)"

So mappers will analyze which patterns gives the most pp-/highest score and we are at the start again or where we are right now. Thats what i understood ^^

R E I K A R I wrote:

So mappers will analyze which patterns gives the most pp-/highest score and we are at the start again or where we are right now. Thats what i understood ^

The only way patterns would give the most pp-/highest score is that the difficulty metric say the map is hard when it is not. Fix what calculates the difficulty metrics. Fix star rating.

You think it's the pp that makes maps overrated or underweighted right now? Actually that's is only 1% of the problem. You need to first fix star rating before you fix pp, or you will try to "fix" pp when pp itself uses broken star rating values. That's how it works right now.

Well kinda... star rating is kinda embeded into the pp formula, and well, it's just a mess. The star rating you see ingame is not really the same thing being used for pp, which is why I said to calculate them separately as difficulty metrics to pass the values down for calculation.

https://github.com/ppy/osu-performance/ ... dScore.cpp
^ This code is the current pp formula. It calculates metrics for speed, aim and acc, but does it with the player's score implicitly (meaning all in one). I only said to calculate metric first, then apply them to player scores to calculate pp.

• pp(difficulty(map), score)

and not

• pp(map, score)

If you take pp formula and plug in the same relative player results for all maps, what you essentially get is star rating. Your pp formula calculates difficulty metrics for acc, highest combo, map length, and number of notes. Therefore, the star rating would be based on just these difficulty metrics, which I think is very wrong.

What if, get this, it took the average combo of the play. Let's say a person had 3 misses, one at 142x combo, one at 241x combo, and one at 189x combo, the average combo there would be 190.666666667, rounded to 191, is what the pp calculations will work with

Zozimoto wrote:

What if, get this, it took the average combo of the play. Let's say a person had 3 misses, one at 142x combo, one at 241x combo, and one at 189x combo, the average combo there would be 190.666666667, rounded to 191, is what the pp calculations will work with

Wouldn´t that be even worse? ^^ I mean lets say the map has 700 notes in total, your first combo breaks after 650 notes and then you hit the other 50 notes, 650 + 50 /2 = average 350. That would be even worse for the players ^^ Because as soon as you miss a single note - it doesn´t even matter WHEN you miss it, your play would be destroyed ^^

"The only way patterns would give the most pp-/highest score is that the difficulty metric say the map is hard when it is not. Fix what calculates the difficulty metrics. Fix star rating."

I dont think so. Mappers know which specific patterns are worth more pp than others. Spacing, Jumps, specific distances and so on. I don´t really think the Star rating is that bad right now.

We could also check for when the miss happened. Which object.

EDIT: But then again, a miss early in the map would be worth way less than a miss later in the map.
So we could look for the amount of notes that weren't missed.

Zozimoto wrote:

We could also check for when the miss happened. Which object.

EDIT: But then again, a miss early in the map would be worth way less than a miss later in the map.
So we could look for the amount of notes that weren't missed.

But that would not solve the problem of your suggestion. In my example the player has not even missed a note and had almost an Full combo but with your method even that near FC would be worthless even with no misses ^^