This is a theoretical question, please answer under the assumption that this can happen

If long notes in osu!mania were changed to be exactly 2 combo per LN (press+release) instead of an incrementing combo, would you want that?
Note that this change would incur a score wipe of all scores on all charts ever, both ranked and unranked, that have at least one LN in them.

Even though osu!mania has been established for quite a while, I'm pretty sure there are still things that could be changed. CtB had a pretty significant change late into its period of existence, I don't see why mania can't have it too. But need to see what everybody thinks first.

They should be only one combo upon release... I guess 2 would make more sense than the way you gain combo now though.

And I don't see why it would need a scorewipe; just let people have different combos, it doesn't change accuracy. You can still see people's score that used HR before that was unranked so why delete these scores.

How do you even wipe the scores on unranked charts? They don't have scores to begin with. ??

NO I'm not okay with wiping everyone's scores over something meaningless like this.

‎

It could be worse (std scoring)

The combo number is not used for score, accuracy or pp, so changing how it is calculated doesn't really have much of an effect overall. I don't think that wiping scores for that change is needed.

what why would anyone do that

Aqo wrote:

This is a theoretical question, please answer under the assumption that this can happen

If long notes in osu!mania were changed to be exactly 2 combo per LN (press+release) instead of an incrementing combo, would you want that?
Note that this change would incur a score wipe of all scores on all charts ever, both ranked and unranked, that have at least one LN in them.

Even though osu!mania has been established for quite a while, I'm pretty sure there are still things that could be changed. CtB had a pretty significant change late into its period of existence, I don't see why mania can't have it too. But need to see what everybody thinks first.

You say that it's a good idea, but you don't explain why. WHY should LN's be only worth two? WHY should we give up the scores we worked hard for just to make this arbitrary change without a proper explanation as to the exact reason we SHOULD? I've never heard anyone complain about LN combos being worth too much, ever.

Also, CtB is still a completely broken and garbage excuse of a gamemode so I would hold off on saying "this will be good for mania as it was for CtB".

This informative post will be deleted anyway, just move along

Given that combo doesn't actually affect the scores,, why would a score wipe be needed?

Rori Vidi Veni wrote:

- Make 100% actually mean 100%: Self explanatory. Bonus points if we get an option to display it like in ITG(starts at 0% and is rising)

Rori Vidi Veni wrote:

- EX score(also known as DP score in Stepmania): No fixed ceiling, no combo bonus, and no "miss placement affects how much you lose" bullcrap. Each grade is worth X points and the max score is simply highest grade value times combo. I don't think LR2's 2-1-0-0 is great either, and it would definitely need lots of discusion. Maybe 10-5-2-1-0?

Changing the accuracy formula doesn't require wiping scores, just a pp recalculation.

The current accuracy formula works like EX score, but with other parameters.

Instead of PGreat-Great-Rest = 2 - 1 - 0, it uses MAX-300-200-100-50-Miss = 30-30-20-10-5-0

One possibility is matching the accuracy formula to the base score formula (32-30-20-10-5-0), in that case 100% = 1 million score, and also there is more distinction between scores that just avoid getting many judgments below 300, and scores with great accuracy.

Most players I have seen seem to focus more on the accuracy percentage than the score for measuring their performances in maps. If the accuracy formula was changed to one that matches better the way players perceive their performance in plays, then it should become the main measure of performance for the game overall (ranking leaderboards based on accuracy instead of score, changing the pp formula to weight accuracy more than score).

And maybe wipe the highest combos.

Full Tablet wrote:

The combo number is not used for score, accuracy or pp, so changing how it is calculated doesn't really have much of an effect overall. I don't think that wiping scores for that change is needed.

I think it does... not really sure on how the score mechanics work though...

edisk wrote:

And maybe wipe the highest combos.

Full Tablet wrote:

The combo number is not used for score, accuracy or pp, so changing how it is calculated doesn't really have much of an effect overall. I don't think that wiping scores for that change is needed.

I think it does... not really sure on how the score mechanics work though...

SPOILER

https://osu.ppy.sh/wiki/Osu!mania#Score

The combo counter doesn't matter for score, what matters is the Bonus value during the play. 1 LN has the same weight than 1 regular note for score.

This informative post will be deleted anyway, just move along

Rori Vidi Veni wrote:

I'm still gonna say it indirectly affects score until some score screenshots prove otherwise.

Also, your link shows just how overcomplicated the scoring is. "HitValue", "HitBonusValue", HitBonus", "HitPunishment", "ModMultiplier" and "ModDivider". Why do you need so many variables?

1. For the first question you just need to read carefully https://osu.ppy.sh/wiki/Osu!mania#Score and maybe look at a replay, on Ln's combo is trivial, at the end of the day, you are given a judgement on a LN equivalent to a single note plus a bigger combo proportional to the lenght of the said LN.
2. Again, read carefully the link above and you will understand why the mode needs so many variables, given that we also have a bonus sistem that rewards you for beeing accurate and not missing, NOT for keeping a high combo.
All that said, the changes proposed by OP would indeed be a nice change but are far to trivial to warant a full wiping of the scores.

This informative post will be deleted anyway, just move along

To be honest I would also prefer a metric that doesn't involve combo at all but I doubt it will happen.

Rori Vidi Veni wrote:

rewards not missing = combo based

you are half right, but the the way it works is that you start with max bonus per note. 300 and above will keep it at max while 200 and below will decrease it, miss returning it to 0. If you break combo on a LN and get a 200 judgement the bonus will only go down by 4 while the combo goes down to 0, thus combo based=/=reward for not missing. As you can see, breaking combo is nowhere near as dramatic as it is for standard. Really, all this is explained in the link I provide, just read and try to understand.

I agree with the change, since the Star-Rating system doesn't increase correctly the stars by number/length of LNs.

Rori Vidi Veni wrote:

Also, your link shows just how overcomplicated the scoring is. "HitValue", "HitBonusValue", HitBonus", "HitPunishment", "ModMultiplier" and "ModDivider". Why do you need so many variables?

Because names are better than single letters the way math formulas work.

Since there's a lot of confusion here, let me try to explain the exact way this math formula works. There are 2 numbers that are added together to figure out how many points a note is worth when you hit it. The first one is based entirely on your accuracy, the second one is a combination of your accuracy and how consistent you are.

Let's start with the simpler one, the one based completely on your accuracy.

wiki wrote:

BaseScore = (MaxScore * ModMultiplier * 0.5 / TotalNotes) * (HitValue / 320)

What does this actually mean? Well, looking at the formula I can tell that this number is calculated every time you hit a note.

Let's break it down. The first term of this equation calculates the maximum number of points that a note is worth. The simple way to do this would be to simply divide the maximum score a song is worth (1 000 000) by the total number of notes in a song.

But wait, some mods (like easy/halftime/nofail) actually affect the maximum score you can get, and that needs to be added into the calculation. So if you need to reduce the maximum score to 500 000 (50% of 1 million), what you get is

MaxScore * ModMultiplier / TotalNotes

. But wait, before we forget, this is actually only half the score calculation, so we need to make sure to always multiply this by 0.5 (50%), so that's how we get the first half of the calculation:

wiki wrote:

BaseScore = (MaxScore * ModMultiplier * 0.5 / TotalNotes)

So far, we've only calculated the maximum score that a note is worth. This is only useful if a player gets a 1 000 000. the second term of this equation solves just how many points a judgement worse than MAX is actually worth. If you calculate it out it works out to this:
MAX = 320/320 = 100%
300 = 300/320 = 93.75%
200 = 200/320 = 62.5%
100 = 100/320 = 31.25%
50 =50/320 = 15.625%

Let's actually do a calculation here. Suppose you decide to play a song with 500 notes, and no mods, and the first note you hit is a 200.
The calculation becomes:
BaseScore = (1 000 000 * 1.0 * 0.5 / 500) * (200/320)
BaseScore = (500 000 / 500) * (200/320)
BaseScore = (1 000) * (200/320)
BaseScore = 1 000 * 0.625 (remember, 62.5% = 0.625)
BaseScore = 625 points for this note.

If the note was instead a 300, you would end up with 1000 * 93.75% or 937.5 points.

If the song had 750 notes instead of 500, the calculation would look like this:
BaseScore = (1 000 000 * 1.0 * 0.5 / 750) * (200/320)
BaseScore = (500 000 / 750) * (200/320)
BaseScore = (666) * (200/320)
BaseScore = (666) * 0.625 (remember, 62.5% = 0.625)
BaseScore = 416.66

But wait, like we said earlier, this is only half the actual score calculation!

The second half is a bit more complicated, but not nearly as complicated as it sounds at first.

wiki wrote:

BonusScore = (MaxScore * ModMultiplier * 0.5 / TotalNotes) * (HitBonusValue * Sqrt(Bonus) / 320)
Bonus = Bonus before this hit + HitBonus - HitPunishment / ModDivider
Bonus is limited to [0, 100], initially 100.

The first part should look pretty familiar. The first term of the formula is identical to the formula we used just above. Just like above, if you use easy/nofail/halftime, your maximum score is cut in half. The second term here is a bit more complicated. This part is the reason that people are arguing about whether combo is used in your score or not. The ingame combo number is not actually used here. A0nother thing people have been arguing about is how other mods, such as hardrock, doubletime, etc. affect your score in mania. This part of the score calculation is where they are calculated.

The game tracks how consistently you can hit 300's or MAXs with this "Bonus" variable. You start the song with 100 bonus. As long as you hit 300 or MAX, you never lose any bonus, but as soon as you hit a 200 or lower, your bonus will be reduced. If you miss, your bonus is set to 0.

This "Bonus" variable is where mods such as HardRock will affect your score. These mods reduce the amount of bonus that is subtracted when you hit a note and get a 200 or less. For example, if you hit a 200, you will lose 8 out of your bonus. If you hit a 200 while using DoubleTime, you will lose 8/1.1 or 7.27 instead.

So let's explain the second term here now that we know what bonus actually means.

wiki wrote:

(HitBonusValue * Sqrt(Bonus) / 320)

This ALMOST looks like the first equation. And it does the same thing, it calculates the actual percentage of maximum score that whatever note you just hit is worth. If you look up the value of HitBonusValue for each judgement, you'll notice that the maximum number is 32. This is because sqrt(100) (the maximum value Bonus can be worth) is 10. This means that if you've hit all MAX or 300's, and then you hit a 200, that 200 is going to be worth (16 * 10 / 320) = (160/320) = 50% of maximum score in for this half of your total score.



So let's combine everything together. Earlier we calculated how many points a 200 is worth for the first half of your score, and I just explained how the second half works, so let's say that the first note you hit on a song is a 200, and let's actually calculate exactly what it's worth:

BaseScore = (1 000 000 * 1.0 * 0.5 / 500) * (200/320)
BaseScore = (500 000 / 500) * (200/320)
BaseScore = (1 000) * (200/320)
BaseScore = 1 000 * 0.625 (remember, 62.5% = 0.625)
BaseScore = 625 points

BonusScore = (1 000 000 * 1.0 * 0.5 / 500) * (16 * sqrt(100) / 320)
BonusScore = (500 000 / 500) * (16 * sqrt(100) / 320)
BonusScore = (1 000) * (16 * sqrt(100) / 320)
BonusScore = (1 000) * (16 * 10 / 320)
BonusScore = (1 000) * (160 / 320)
BonusScore = (1 000) * 0.50
BonusScore = 500

625+500 = 1125 points for hitting a 200 on the first note of a song with exactly 500 notes in it.

Now let's see what happens if you hit a 300 after that 200:

BaseScore = (1 000 000 * 1.0 * 0.5 / 500) * (300/320)
BaseScore = (500 000 / 500) * (300/320)
BaseScore = (1 000) * (300/320)
BaseScore = 1 000 * 0.9375
BaseScore = 937.5 points

BonusScore = (1 000 000 * 1.0 * 0.5 / 500) * (32 * sqrt(92) / 320)
BonusScore = (500 000 / 500) * (32 * sqrt(92) / 320)
BonusScore = (1 000) * (32 * sqrt(92) / 320)
BonusScore = (1 000) * (32 * 9.59 / 320)
BonusScore = (1 000) * (306.93 / 320)
BonusScore = (1 000) * 95.91
BonusScore = 959.1

937.5 + 959.1 = 1896.6 points

If instead of a 200, your first note was a 300, it would be worth 1937.5 points instead. That means that the first 200 cost you 40.9 points off your score for that 300.

This informative post will be deleted anyway, just move along

It's not as complicated as it looks when you're actually doing the math. Perhaps the consistency section could be combined into the accuracy calculation, but the math is actually quite easy to understand.

If you want to keep some sort of consistency element to scoring, and you also want to keep mods affecting the score a bit, there's almost no way this could be any simpler.

Let's simplify:

MaxNoteScore = (MaxScore * ModMultiplier * 0.5 / TotalNotes)
ScorePercentage = (HitValue / 320)
BonusPercentage = (HitBonusValue * Sqrt(Bonus) / 320)

Now you get TotalScore = MaxNoteScore * NotePercentage + MaxNoteScore * BonusPercentage

If you can design a system that
* Takes mods into account
* Has a maximum score of 1 000 000 per song
* Takes both accuracy and consistency into account
* And is simpler mathematically to calculate

Be my guest.

Rori Vidi Veni wrote:

rewards not missing = combo based

When someone says combo based, it's assumed they mean based on the combo listed ingame. By that definition, mania scoring is NOT combo based, because the bonus score is not based on your actual combo number, but an invisible variable instead.

You are however right in the sense that mania scoring uses consistency as well as accuracy as the scoring method.

This informative post will be deleted anyway, just move along

Rori Vidi Veni wrote:

Bobbias wrote:

If you can design a system that
* Takes mods into account
* Has a maximum score of 1 000 000 per song
* Takes both accuracy and consistency into account
* And is simpler mathematically to calculate

I have to quote myself yet again, don't I?

Rori Vidi Veni wrote:

- EX score(also known as DP score in Stepmania): No fixed ceiling, no combo bonus, and no "miss placement affects how much you lose" bullcrap. Each grade is worth X points and the max score is simply highest grade value times combo.

I've been saying all this time that 2) and 3) are what makes mania scoring bad. Do you have medical problems with reading comprehension or something? I'm not dissing you, I'm genuinely curious.

To be honest, I don't think mania scoring is as bad as you believe it to be. You also missed my point. I was stating that mathematically speaking, those variables are the bare necessity to model the scoring system. In the most literal sense possible it's not overly complicated based on it's intended design. I was not discussing the merit of that design choice at all.

I'll let you in on a little secret about osu. Peppy is not going to change the system just because a bunch of people on the forum come up with some idea of some perfect scoring system. If any changes are made, it will be because peppy decided he's not satisfied with the current mechanics and can justify a score wipe to change it (which it most likely would require).

And as far as I'm concerned, people should calm down and stop taking o!m so seriously anyway. If it somehow gets better, that's great, but if you've been paying any attention to how osu's developed you'll know that any significant changes to o!m will be extremely rare and probably have nothing to do with the community's wishes.

I should also point out that I've been a very vocal critic of osu!mania's issues since the game mode was released. I've predicted many of the current issues plaguing the community as well. I'd love for a saner scoring system to be implemented, but that conflicts with the current design philosophy.

If you can design a system that
* Takes mods into account
* Has a maximum score of 1 000 000 per song
* Takes both accuracy and consistency into account
* And is simpler mathematically to calculate

My idea involves knowing the difficulty of a certain point of the map. Suppose you know D_t (difficulty at point t). Then

Where D_t is the difficulty at time t, Acc_t is the score attained from a note at time t and C is the coefficient needed to normalize the score to 1,000,000 points. Instead of the score spread having a logistic growth distribution, this will weigh harder parts more than easier parts while still preserving the 1,000,000 score. The difficulty of a pattern at a certain point will be affected by mods, so that's accounted as well. The main difficulty in this is knowing the difficulty of the map at a certain time though.

If you can convince peppy to somehow completely change how difficulty calculations work, that might be a viable option. Good luck with that.

Also, how does your proposed calculation handle multiple notes at the same time? Sum them into Acc_t, or calculate this for each note in parallel?

It takes account the score or accuracy of each individual note as if they were not parallel. Also another interesting thing to note (not sure how good of an idea it is). Since this formula doesn't care about accuracy or score based on the score to accuracy conversion you wrote, Acc_t can also be the % accuracy in terms of hitting dead center 0ms. The score points would still be there as a visual display, but the scoring will be more accurate since you won't have integral values anymore (0, 50, 100, 200, 300, MAX).