11 Jul 2023
Community Meeting Recap
by osu! team

Community Meeting Recap

by osu! team

Catch up on the latest community meeting and participate in discussing the final evolution of osu!'s scoring system!

A bit over a week ago, a community meeting was held to discuss how to proceed with scoring in osu!(lazer) with various people in the community.

If you missed the previous meeting, here's the quick summary of where we have been:

  • How the leaderboards will be combined was discussed but no definitive conclusion was reached.
  • Leaderboard resets were planned to be performed multiple times until the scoring system would be finalised.
  • Mod multipliers for speed adjustment mods were decided to be rounded to the nearest 0.1x change in speed.
  • The question of whether immediate priority should be put on getting osu!(lazer) ready for tournament use was met with a clear "no".
  • The scoring system used in osu!(lazer) was discussed, with the general sentiment being that it had many issues.

And here's where we are now:

  • The discussed changes to mod multipliers for speed adjustment mods have been implemented.
  • Further internal discussion has led to the decision that leaderboards will be either wholly or partially wiped just once when the new scoring system is finalised, instead of multiple times.
  • ScoreV2 is now implemented as the current scoring system in osu!(lazer). It will be adjusted one final time and then locked permanently, after which all existing scores will be converted to this system.

It's that last point that motivated this meeting. ScoreV2 has proven to be a better scoring system as it has been used for tournament play for years, but there may be issues that could be addressed. What adjustments to make, if any, is the big question.

You can binge the entire VOD below:

Much of the conversation this time surrounded a long-debated topic: combo. Specifically, its significant — perhaps too significant — influence on score. These potential changes for the osu! game mode were discussed:

  • Directly decreasing the weight of the first few misses, since even one miss can have a huge impact on score.
  • Increasing the proportion in which accuracy affects score compared to combo. This was based on a Twitter poll by peppy.
  • Drastically decreasing the value of high combos and aggressively scaling combo's influence based on accuracy. This change has been submitted for review on GitHub already.

There were some other issues brought up as well:

  • Hit values for slider ticks and some other judgement types currently don't match osu!(stable), which will be corrected.
  • Slider ends (which currently are almost completely optional to hit), slider hit accuracy and other changes to game mechanics have been contentious. These will need more discussion.
  • osu!mania PERFECT judgements are currently required for 100% accuracy, making it an incredibly difficult feat. Also to be discussed separately.

In short, ScoreV2 is the future — or at least an improved version of it. But in order to make sensible adjustments, focus should be spent on identifying problematic leaderboards and outlier scores that "feel wrong", to identify underlying issues and experiment with applying changes to get those scores closer to expectations.

This is where you come in. We're looking for feedback on the current ScoreV2 system in order to nail down any remaining issues. The coming months will see the last steps taken to lock in scoring and make scores permanent in the new client.

Keep in mind that there are as many opinions on scoring as there are players, and any adjustments will therefore focus on aligning things "close enough" to fix widely perceived issues, rather than debating over fine details.

A lot of extensive discourse about scoring has been going on for months already, e.g. in this GitHub thread concerning ScoreV2 as it currently stands. If want to contribute, post your well-thought-out ideas, aggregated survey findings and feedback for other proposals there.

—osu! team

Comments27

Sort by

Shouldn't we also set up the final mod multiplier numbers first before wiping all scores? Wiping all scores and not doing it again when there's still the task of setting the correct score multiplier for mods is kinda concerning and it'll just mess up the leaderboards again. Also because I think the current mod multipliers for certain mods aren't what they're supposed to be for fair leaderboard placement. I think everything that is related to score should be set in stone first before the one-time score wipe happens so that when it does occur, the leaderboards afterwards will be what people wanted it to be.

Walavouchey ominoussage

mod multipliers can be changed at any point later and scores recalculated without loss

ominoussage Walavouchey

Oh great. I just thought they might've forgotten about it. Thanks.

There is a pretty clear necessity for more debates before any change since there are a lot of points to be made and this is a decision that affects literally everyone. It's as some people said: changing the scoring system just one more time to never do it again while there is obvious confusion is just, straight up dumb. Let's not ignore the fact that this can impact players in such a way that they might even quit the game, in case they feel unsatisfied and unheard. It's a huge thing. Just, please, consider having more debates and keep sharing news with us.

Walavouchey KimAriel

there will be more meetings, yes

ominoussage KimAriel

The reason why they only want one more time to change the scoring system because scorev2 is already favored upon by the majority of players and it's been the scoring system used in tournaments. I'm hoping they don't mess it up though because I can't imagine scorev2 getting better than it already is.

spelling error: "this is were you come in"

The "weight of the first few misses" being too high is not the issue with combo scoring. Misses mattering arbitrarily more based on where they happened in the map and the way they are spread out is the problem.

There is no point in "identifying problematic leaderboards and outlier scores" because this problem happens on essentially every leaderboard in the game. If you look down far enough on leaderboards with a good amount of plays, you'll almost always find scores that are placed below other scores that are objectively worse than it because the player missed in an arbitrarily less optimal spot.

I cannot support the idea of adjusting the score system only one more time and then never again when there isn't even an agreement on what we are trying to fix. That seems crazy to me.

Walavouchey Sleepteiner

Misses mattering arbitrarily more based on where they happened in the map and the way they are spread out is the problem.

So in other words, combo should have 0 influence like in mania?

There is no point in "identifying problematic leaderboards and outlier scores" because this problem happens on essentially every leaderboard in the game.

Even so, the process would still be the same. The point is that changes should be based on where and why the leaderboards are wrong instead of how the scoring equations should work, because there are too many opinions on that. I recommend you listen from 00:26:33 to 00:33:15 in the VOD.

I cannot support the idea of adjusting the score system only one more time and then never again when there isn't even an agreement on what we are trying to fix. That seems crazy to me.

See my other answer, but I'll add that lazer won't ever go live if we can't make scores permanent, and they can't continuously be updated to ever-changing scoring systems because it's a lossy conversion. There's a very good reason for why the leaderboards never switched to ScoreV2 in stable.

Sleepteiner Walavouchey

So in other words, combo should have 0 influence like in mania?

I personally think so, yes, but even if I didn't want that, dramatically reducing the impact of combo is likely the only way to address the problem of certain misses having too much impact.

Also, correct me if I'm wrong, but I'm pretty sure that mania does have a bit of combo influence?

The point is that changes should be based on where and why the leaderboards are wrong instead of how the scoring equations should work, because there are too many opinions on that. I recommend you listen from 00:26:33 to 00:33:15 in the VOD.

I was at the meeting. The "scorev1'd" concept that was brought up almost always refers to late 100s being too impactful compared to early 100s. It isn't related to misses. It isn't the same problem and, again correct me if I'm wrong, but I'm pretty sure that doesn't happen in scorev2 basically at all anyway, so it isn't relevant to the discussion.

Treating the fundamental issues of combo scoring as some mysterious edge case is very misleading. There is no need to search for leaderboards that have examples of it. Just pick a selection of the most popular maps, run a program to find scores that have more total misses, 100s, and 50s but are higher on the leaderboard than scores that have significantly fewer mistakes and adjust the combo/accuracy spilt until the number of those cases goes down significantly. I predict that it'll become clear that we would need to reduce the impact of combo very dramatically to even begin to address the issue of certain misses mattering too much.

See my other answer, but I'll add that lazer won't ever go live if we can't make scores permanent, and they can't continuously be updated to ever-changing scoring systems because it's a lossy conversion. There's a very good reason for why the leaderboards never switched to ScoreV2 in stable.

I'm not saying that we should keep messing with score forever. I'm saying that making one change and then never again is bizarre considering the state of the debate. There has been nowhere near enough discussion about this issue as demonstrated by a concerning lack of understanding of what the problems even are among the people making the decisions. Also, the community meetings and this news post are clearly not working as an adequate call for feedback from the community because of how few people are engaging and how many different types of players are not being represented. This all feels so short-sighted and risky.

Are we really going to tell all the players who might dislike whatever final adjustment is made "sorry, if you wanted to voice your opinion on the future of osu, you had to keep up with peppy's twitter account and the front page of the osu website, so you are too late"?

Yeah, we didn't switch to scorev2 in stable because it was generally agreed upon that the downsides were too much, but now all of a sudden the development team seems to have completely changed their minds (from an outside perspective) and now scorev2 is the only option and we have to commit to it NOW even though the vast majority of the problems that people had with the lazer scoring system would be immediately fixed if the impact of accuracy was increased. I just can't understand the rational behind almost any of this new direction.

isn't perfect the 300 in mania with 320 being max/marvelous?

Walavouchey Remyria

not in osu, see https://osu.ppy.sh/wiki/en/Gameplay/Judgement/osu!mania
it's "perfect" in lazer and "max" on the website (and a rainbow 300 in stable)

ERA Pheon Walavouchey

as if they edited that wiki article... the best judgement (which the entirety of the mania community still calls max/marvelous because noone likes the new judgements, i'll stop ranting though because last time we all did those in charge barely cared about what we had to say it feels like) used to be called "Rainbow 300 (MAX)" on that article, so the confusion is justified :shrug:

man

[-Omni-] l-_-l

wo MAN

I have a question, now that the scores are going to be capped to 1,000,000 + mods how is the account level and the ranked score going to work? Maybe counting the scores like the 'classic' option that lazer offers or they are going to count just the standardised score as both the ranked score and the account level? Since that would take forever to level up unless you nerf the amount of points needed to level up. I'm not sure if I'm explaining myself well.

as you say, there's "standardised" and "classic" scoring displays which are interchangeable and preserve leaderboard ordering between each other, so in theory they could be a setting on the website just like they are rn in lazer. when calculating the total ranked score for the score leaderboards, it'll use classic scoring so score farmers can still do their thing

for levels it's the same story, unless the level formula gets reworked

ZebbyX Walavouchey

Thanks for you answer, seems like a good path to take in my opinion, hope they do that in the end!

Admitted not much thought put into this but having scoring locked permanently after a single change seem pretty unwise, compared to iteratively improving the algorithm.

scoring is not designed to be a flexible system. it's just a metric that compares scores on the same map leaderboard. it's basically a completion bar - when you get max score you're done with the map.

and it's not trivial to migrate scores over to the new system. converting scorev1 -> scorev2 will incur minor deviations compared to if the same score was actually set in the new system, because not all replay data is stored on the servers

pp is flexible, however, and has evolved through many iterations over the years to cover difficulty in a way scoring can't and shouldn't

kokdeleted

Classic score is still a thing for those people...

thank you so much i went to sleep and missed 2 minutes of it

/

thank you so much i went to sleep and missed 2 minutes of it