maniaCmiC wrote:

If you follow your line of thinking, then the easier the map / part of the map, the more PP it should give.

This doesn't make sense.

That's not what I'm saying at all. I'm saying that misses on easy parts should make the map give less pp. I'm not saying that hits on easy parts should give more pp.

maniaCmiC wrote:

You somehow managed to make it "make sense", if you ignore that one part is harder than the other.

The reward isn't based on "likelihood of which is going to be more due to luck", it's rewarded because it's actually a harder part. Hitting a harder part rewards more than an easier part, because the easier part is easier to hit. It's pretty straightforward. I encourage you to actually critically think about your argument before thinking I'm ignoring it.

You isolated it into being due to probabilities of missing. Well the 6* part isn't rewarded for likelihood of it not being due to luck, it's rewarded more because it's actually a harder part with higher SR. The 5* part is rewarded less because it's easier to hit.

Seriously people on this forum baffle me with their weird & obscure mental gymnastics. You know why it's "counter-intuitive"? Because things that are obviously incorrect are a lot less likely to be intuitive.

Bad players are more likely to miss easy notes. Therefore more weight should be given to misses on easy parts.

If you are considering hits you need to consider strains. If you are considering misses, you need to consider probabilities instead because no strain was experienced.

If you assigned a negative strain value to misses it would be possible to get negative pp. This makes zero sense.

Monkey D Amri wrote:

Dude what are they doing to flashlight?

what did they do to flashlight? lol

Fxjlk wrote:

maniaCmiC wrote:

If you follow your line of thinking, then the easier the map / part of the map, the more PP it should give.

This doesn't make sense.

That's not what I'm saying at all. I'm saying that misses on easy parts should make the map give less pp. I'm not saying that hits on easy parts should give more pp.

maniaCmiC wrote:

You somehow managed to make it "make sense", if you ignore that one part is harder than the other.

The reward isn't based on "likelihood of which is going to be more due to luck", it's rewarded because it's actually a harder part. Hitting a harder part rewards more than an easier part, because the easier part is easier to hit. It's pretty straightforward. I encourage you to actually critically think about your argument before thinking I'm ignoring it.

You isolated it into being due to probabilities of missing. Well the 6* part isn't rewarded for likelihood of it not being due to luck, it's rewarded more because it's actually a harder part with higher SR. The 5* part is rewarded less because it's easier to hit.

Seriously people on this forum baffle me with their weird & obscure mental gymnastics. You know why it's "counter-intuitive"? Because things that are obviously incorrect are a lot less likely to be intuitive.

Bad players are more likely to miss easy notes. Therefore more weight should be given to misses on easy parts.

If you are considering hits you need to consider strains. If you are considering misses, you need to consider probabilities instead because no strain was experienced.

If you assigned a negative strain value to misses it would be possible to get negative pp. This makes zero sense.

The problem is you don't ever have negative values. The "punishment" is for *not hitting* a note. So if a note is harder to hit, you lose more PP for not hitting it, due to the potential PP you could have gotten FOR hitting it.

If you miss on an easy part you lose less PP because there wasn't as much PP to be gained to begin with. The easy part doesn't give the most PP because it's easy to hit.

You don't seem to understand..

You can't isolate the difficulty of hitting a note from how much it should lower PP. If a note is harder to hit, it should make you lose more of the total PP from the map because that part gives more PP (bc it's harder to hit.)

If we apply your line of thinking, and scale it, you would be saying a missing a bunch on a 100000000* part you should still get more of the map's PP than if you missed on the 1* part.

It doesn't make sense because the 10000000* part is where all the map PP comes from. Missing on the 1* part is not as impactful because it never gave much PP.

You could use your logic to scale a play to infinite PP xd, just increase SR and hit the easier parts, you get more of the maps total PP than if you hit the hard part and missed the easier...

I'm not going any further though. If you want to use your "logic" to try and make it make sense, then LOL.

It doesn't make sense. Try to figure out why what you're saying is counterintuitive, and let me know once you get there. Hint: it's because it's wrong. But you will continue to do mental gymnastics to form it to be "right" so it'll be endless. Try to criticize your own argument here.

this rework is the best thing to ever happen to this game
but is it confirmed happening

maniaCmiC wrote:

The problem is you don't ever have negative values. The "punishment" is for *not hitting* a note. So if a note is harder to hit, you lose more PP for not hitting it, due to the potential PP you could have gotten FOR hitting it.

If you miss on an easy part you lose less PP because there wasn't as much PP to be gained to begin with. The easy part doesn't give the most PP because it's easy to hit.

You don't seem to understand..

You can't isolate the difficulty of hitting a note from how much it should lower PP. If a note is harder to hit, it should make you lose more of the total PP from the map because that part gives more PP (bc it's harder to hit.)

If we apply your line of thinking, and scale it, you would be saying a missing a bunch on a 100000000* part you should still get more of the map's PP than if you missed on the 1* part.

It doesn't make sense because the 10000000* part is where all the map PP comes from. Missing on the 1* part is not as impactful because it never gave much PP.

As soon as you miss the entire difficulty system breaks down. Here's an example:

You set up cross screen jumps at a extremely high bpm to get massive strain



All you have to do is hit the notes on the left side and you have hit half of the hard part. It would be like 5% of the difficulty for hitting half of the hard notes. You can no longer say "they hit the hard part so they should get more pp" because once you miss (even on the easy part) you cannot determine a hard part anymore. Strains are irrelevant, probability becomes more accurate for skill assessment.

Now back to your example.

maniaCmiC wrote:

You could use your logic to scale a play to infinite PP xd, just increase SR and hit the easier parts, you get more of the maps total PP than if you hit the hard part and missed the easier...

No. Under this miss rework they have to assume the worst strain for misses. Using probabilities to asses misses wouldn't be any different. Hitting the easy 1* parts and missing all the hard 10000000* parts would be severely punished, the reverse would be punished even more. This example basically proves my point because who the hell hits 10000000* parts and misses 1* parts? Plays like that can only be extreme luck and it should be be punished accordingly. Plays like that are exactly why probabilities are important to use as a metric.

You might say: Shouldn't the play that hit the 10000000* parts be rewarded? Even though its extremely lucky, clearly its a better display of skill?

No it should not. PP is used to rank players and so lucky plays are really bad to assess skill. The whole point of pp is to give a reliable metric to compare players. If player 1 can only beat player 2 0.000001% of the time (when he goes god mode and hits 10000000* parts) then should he really be the same rank?

maniaCmiC wrote:

I'm not going any further though. If you want to use your "logic" to try and make it make sense, then LOL.

It doesn't make sense. Try to figure out why what you're saying is counterintuitive, and let me know once you get there. Hint: it's because it's wrong. But you will continue to do mental gymnastics to form it to be "right" so it'll be endless. Try to criticize your own argument here.

You can stop posting whenever you want. You don't need to announce you are walking out of the argument. If you cant point out any flaws in the "mental gymnastics" then how can you label it so? If the logic is so bad then why do you have to resort to labelling the arguments instead of proving them wrong?

Calling my arguments "mental gymnastics" is a cop out for actually understanding the topic and making actual points. By emotionally labelling points of view for any discussion you will create blind spots for your own understanding and derail the discussion.

Fxjlk wrote:

Here's an example:

You set up cross screen jumps at a extremely high bpm to get massive strain



All you have to do is hit the notes on the left side and you have hit half of the hard part. It would be like 5% of the difficulty for hitting half of the hard notes. You can no longer say "they hit the hard part so they should get more pp" because once you miss (even on the easy part) you cannot determine a hard part anymore. Strains are irrelevant, probability becomes more accurate for skill assessment.

This is a very valid point, but I believe there are ways to mitigate that.

Take a True/False test for example. One may think 0% would display no knowledge, 50% would display some knowledge, and 100% would display full knowledge. However, if one would randomly guess the answers, the expected % of them correct would be 50%. Hence, it is 50% that actually displays no knowledge.

How this relates to the example you gave: You will be able to hit ~50% of notes that are supposedly way above your skill level. So the inability to hit more than a certain % of notes hit above some difficulty displays the inability to perform such patterns.

e

maniaCmiC wrote:

Your argument is that consistency is more important, lucky plays shouldn't be rewarded

But consistency is already a part of the system

Consistency is part of the system, at least as it stands now with combo (aside from diff spikes).

With the miss rework we will punish misses on the easy parts less. The fundamental assumption is that these misses don't matter because they are easy but this is wrong. This is a large downgrade in ranking accuracy, no matter what way you look at it.

If pp was looked at more in terms of probabilities, we would not have massive discrepancies in real skill when people have the same pp.

maniaCmiC wrote:

Your example is like getting 50% on a test, by guessing all the answers

It's an extremity that's irrelevant

Yes, hitting 50% of the notes would show you can't play it at all

But hitting 75%, or getting 75% on the test, shows you are probably not just guessing and can do it to at least some extent

I mean what's stopping us from changing the example above so that its close to 75%? you can have a cross screen every 4th note. This will still make a massive difference to strain in the map with no increase in actual difficulty. The percentage is irrelevant, misses will always, reduce information to use to rank players fairly.

You cannot call an extremity irrelevant without any proof. I used your above extremity of 10000000* and 1* notes and it only proved my point. I didn't call it irrelevant, almost all extremities show something and them being an extremity doesn't make them any less relevant (unless you can prove it is so). It may not be as impactful to the same extent but the point still stands.

abraker wrote:

This is a very valid point, but I believe there are ways to mitigate that.

Take a True/False test for example. One may think 0% would display no knowledge, 50% would display some knowledge, and 100% would display full knowledge. However, if one would randomly guess the answers, the expected % of them correct would be 50%. Hence, it is 50% that actually displays no knowledge.

How this relates to the example you gave: You will be able to hit ~50% of notes that are supposedly way above your skill level. So the inability to hit more than a certain % of notes hit above some difficulty displays the inability to perform such patterns.

Given a large enough sample size and no repeat tests that is true. However Osu is made of small samples and tests that can be repeated indefinitely where only top results are used. This means the cutoff needs to be higher than you would expect.

Harder notes are like questions that have 3 answers. A 3 question answer test would show some knowledge with >33% of them correct

This is why more weight should be given to the easier true/false questions (easy notes). The cutoff is higher (50%>33%). However under the miss rework there is no way to weigh misses on the easy or hard parts. All you can do is punish high strain variance across the board.

This is why we need a hybrid system of both misses and combo. Some combo scaling will allow better weights for the parts of the map that are less likely to indicate skill when missed. Additionally rewards can still remain for chokes on maps with high strain variance.

e

Fxjlk wrote:

MBmasher wrote:

im a pp dev too

I know. I respect you guys alot which makes it hard to disagree.

joz12345 wrote:

If you're looking at scientific reasoning for a scoring method, miss count is much more sensible. An analogy: imagine you're trying to figure out the probability of heads for a weighted coin - you can either count the heads/total throws, which will directly give an unbiased estimator for that probability - or you can look at the max number of heads in a row you have achieved per group of n throws. The latter is much more inefficient - you'll need many sets of n throws to get a good estimator looking at "max combo", whereas you'll get a decent estimate from just one set if you look at the head/tail count. The same applies when looking at misses in osu, just the scenario is more complicated.

Determining the probability of hitting notes in osu is not the same as determining the probability of heads or tails for a weighted coin. In the weighted coin example the probability of results stays constant but this is not the case for osu. I explained this in a post in the other thread here. I would also like to add that physical fatigue is also a function of time during the map and lowers hit chance as a function of time. This can be felt during deathstreams for example.

joz12345 wrote:

Theres some bad reasoning in this thread.

1) There's no reason either system can't be balanced - you can always ask "how hard is it to get x misses or y combo on map z?" If you change the underlying measurement, you're changing the game slightly, which you can choose to like or dislike, but it doesn't mean anything has to be unbalanced.

There is no way you can weigh the location of misses without combo. Its not the best way to weigh miss location but its better than nothing. If there is still some combo scaling then at least 1 miss plays in the middle will be worth less that a miss at the end.

joz12345 wrote:

2) There seems to be a lot of discussion about map length - that is an orthogonal concept. Misses/combo are useful to compare plays only on the same map, and it's kinda obvious that plays on the same map are the same length.

Map length is not the same concept, it is just to show how all misses are not equal. Map length has been brought up to show that fatigue is a factor for determining skill.

joz12345 wrote:

3) If notes at the end are harder to hit why should missing them give more pp? Surely if anything, that's backwards - if you miss the hard bit you get less pp.

It is counter intuitive. If you were comparing on a play by play basis and compare individual strains the player that hits on the hard part should be awarded more pp.

However if you are talking with respect to probability, players who miss on easy parts are more likely to be lower skill players as the probability to hit it should be very high. The misses on easy parts have a higher statistical weight than misses on hard parts. if a player misses the easy part rather than the hard part, it is more likely that the player hit the hard part out of luck rather than skill.

Here is an example:

5 star jumps should have a probability of being hit of >99.9% for 270 pp plays
6 star jumps should have a probability of being hit of >99.5% for 270 pp plays

Compare player 1 who missed a 5 star jump vs player 2 who missed a 6 star jump on a 200 note map with 100 of each note type.

Player 1 has:

99% chance to hit 5 star jumps
100% chance to hit 6 star jumps

Player 2 has:

100% chance to hit 5 star jumps
99% chance to hit 6 star jumps

Since there is a greater discrepancy for player 1 that player 2 from the probability requirements (0.9% vs 0.5%) then player 2 should be awarded more pp. This is because the play by player 1 is more likely to be out of luck rather than skill.


All these gymnastics to basically say "I wanna make it harder for the farmers to bruteforce maps. So let's make it so the longer, easier parts cause you to lose all PP if you miss anything cuz a truly skilled player shouldn't be missing any of it."

What you fail to understand is that the Harumachi Clover farmer has no life. He will eventually have the easy part of that 45 second map memorized by rote, and has no interest in longer 3+ minute maps with the same difficulty spikes but a longer "easy" section. Furthermore, the unskilled player will most likely miss more of the easy section than the skilled player on average, which in a system weighted for misses and not combo, works exactly as intended.

Would such a system make it slightly easier for farmers to gain pp? Sure. But in relation to the number of players it would boost to the appropriate ranking, I think it's worth it. Making it so Mr. DT farmer can now miss 2 notes and still get most of their PP in exchange for regular non-farmers/long map players not having to tear their hair out playing the same map over and over in order to get any pp sounds like a good deal to me.

Fxjlk wrote:

MBmasher wrote:

im a pp dev too

I know. I respect you guys alot which makes it hard to disagree.

joz12345 wrote:

If you're looking at scientific reasoning for a scoring method, miss count is much more sensible. An analogy: imagine you're trying to figure out the probability of heads for a weighted coin - you can either count the heads/total throws, which will directly give an unbiased estimator for that probability - or you can look at the max number of heads in a row you have achieved per group of n throws. The latter is much more inefficient - you'll need many sets of n throws to get a good estimator looking at "max combo", whereas you'll get a decent estimate from just one set if you look at the head/tail count. The same applies when looking at misses in osu, just the scenario is more complicated.

Determining the probability of hitting notes in osu is not the same as determining the probability of heads or tails for a weighted coin. In the weighted coin example the probability of results stays constant but this is not the case for osu. I explained this in a post in the other thread here. I would also like to add that physical fatigue is also a function of time during the map and lowers hit chance as a function of time. This can be felt during deathstreams for example.

joz12345 wrote:

Theres some bad reasoning in this thread.

1) There's no reason either system can't be balanced - you can always ask "how hard is it to get x misses or y combo on map z?" If you change the underlying measurement, you're changing the game slightly, which you can choose to like or dislike, but it doesn't mean anything has to be unbalanced.

There is no way you can weigh the location of misses without combo. Its not the best way to weigh miss location but its better than nothing. If there is still some combo scaling then at least 1 miss plays in the middle will be worth less that a miss at the end.

joz12345 wrote:

2) There seems to be a lot of discussion about map length - that is an orthogonal concept. Misses/combo are useful to compare plays only on the same map, and it's kinda obvious that plays on the same map are the same length.

Map length is not the same concept, it is just to show how all misses are not equal. Map length has been brought up to show that fatigue is a factor for determining skill.

joz12345 wrote:

3) If notes at the end are harder to hit why should missing them give more pp? Surely if anything, that's backwards - if you miss the hard bit you get less pp.

It is counter intuitive. If you were comparing on a play by play basis and compare individual strains the player that hits on the hard part should be awarded more pp.

However if you are talking with respect to probability, players who miss on easy parts are more likely to be lower skill players as the probability to hit it should be very high. The misses on easy parts have a higher statistical weight than misses on hard parts. if a player misses the easy part rather than the hard part, it is more likely that the player hit the hard part out of luck rather than skill.

Here is an example:

5 star jumps should have a probability of being hit of >99.9% for 270 pp plays
6 star jumps should have a probability of being hit of >99.5% for 270 pp plays

Compare player 1 who missed a 5 star jump vs player 2 who missed a 6 star jump on a 200 note map with 100 of each note type.

Player 1 has:

99% chance to hit 5 star jumps
100% chance to hit 6 star jumps

Player 2 has:

100% chance to hit 5 star jumps
99% chance to hit 6 star jumps

Since there is a greater discrepancy for player 1 that player 2 from the probability requirements (0.9% vs 0.5%) then player 2 should be awarded more pp. This is because the play by player 1 is more likely to be out of luck rather than skill.

[quote="joz12345"]4) Fatigue isn't required to make notes at the end of a map harder to get. If you grind a map for an FC for a set period of time, you can get a lot more attempts in if hard parts are at the beginning of a map - this concept has already been incorporated into a few pp systems in the past, e.g. mine and then delta_t's, but neither has been merged since it's harder to document/balance more complicated stuff.

Xexxar did mention your system would be better to balance chokes. However until a system is able to be merged we should keep combo to some degree. This miss rework will fix some underrated chokes but it would fix then more fairly with combo still having weight.

All these gymnastics to basically say "I wanna make it harder for the farmers to bruteforce maps. So let's make it so the longer, easier parts cause you to lose all PP if you miss anything cuz a truly skilled player shouldn't be missing any of it."

What you fail to understand is that the Harumachi Clover farmer has no life. He will eventually have the easy part of that 45 second map memorized by rote, and has no interest in longer 3+ minute maps with the same difficulty spikes but a longer "easy" section. Furthermore, the unskilled player will most likely miss more of the easy section than the skilled player on average, which in a system weighted for misses and not combo, works exactly as intended.

The fact of the matter is, you don't know how long it would take for an unskilled player to rote memorize the easy sections of a farm map. A skilled player might consistently miss ~2 notes on the easy section, while the unskilled player doing it for the first few times might miss 10+ before they manage to memorize it after 100 plays. By making the easy sections worth all the PP of a map, you turn both of these players into farmers who earn basically no pp for their plays despite one player being far superior to the other. The skilled player may have to farm less for the FC, but he still has to farm a map that he should have no business needing to farm - and all for the sake of making the unskilled player's job slightly more difficult by permitting zero mistakes instead of 1-2.

Would such a system make it slightly easier for farmers to gain pp? Sure. But in relation to the number of players it would boost to the appropriate ranking, I think it's worth it. Making it so Mr. DT farmer can now miss 2 notes and still get most of their PP in exchange for regular non-farmers/long map players not having to tear their hair out playing the same map over and over in order to get any pp sounds like a good deal to me.

Fxjlk wrote:

Given a large enough sample size and no repeat tests that is true. However Osu is made of small samples and tests that can be repeated indefinitely where only top results are used. This means the cutoff needs to be higher than you would expect.

Harder notes are like questions that have 3 answers. A 3 question answer test would show some knowledge with >33% of them correct

This is why more weight should be given to the easier true/false questions (easy notes). The cutoff is higher (50%>33%). However under the miss rework there is no way to weigh misses on the easy or hard parts. All you can do is punish high strain variance across the board.

Top results are considered to be outliers or close to being outliers. Players retrying a bunch of time until they get a score unreasonably above their average skill is an issue. More over, while the difficulty of a pattern increases linearly with respect to their average skill, I've observed difficulty increases logarithmically with respect to samples that display best skill. There might be some way to convert best skill to average skill, but it already makes an assumption that all scores showcase best skill, which is not always the case.

I opt to lean toward interpreting every score the player sets to be sampled from their average skill. If they want to retry the map multiple times to boost their apparent skill, then so be it, but then they shouldn't be surprised to learn that their evaluation is overinflated.

Circling back to % threshold, this does up the threshold by some amount. Taking the best result of a true/false test every 50 samples will make you end up with an expected threshold quite higher than 50%. Applying this concept to pp, it would indeed appear inflated. But again, if players do this then they shouldn't be surprised to find that it is inflated.

It's like playing with DT. You get phenomenal pp gains at first until you farm everything out and hit a solid wall. In this case, the wall is the logarithmic improvement curve you chose for yourself rather than a linear one. So doing this will bite back the player eventually. I'd label this as "it all balances out at the end".

maniaCmiC wrote:

I think you're confused and hyper-analyzing. Maybe you should think critically about what you're saying.

capefear56 wrote:

All these gymnastics to basically say

Maybe you guys should learn basic statistics before criticizing my analysis as "mental gymnastics" or "hyper analyzing". All you are really telling me here is that you are unable to understand or don't really care about the details. If you don't care to understand what the rework is even doing then don't post, its as simple as that.

If you cant explain what you mean your posts are meaningless

capefear56 wrote:

All these gymnastics to basically say "I wanna make it harder for the farmers to bruteforce maps. So let's make it so the longer, easier parts cause you to lose all PP if you miss anything cuz a truly skilled player shouldn't be missing any of it."

That's the direction we have currently been going with the ranking system already. Short maps are worth less with the same difficulty due to being bad at measuring skill. If misses on easy parts are also bad for measuring skill, they should be punished too.

capefear56 wrote:

What you fail to understand is that the Harumachi Clover farmer has no life. He will eventually have the easy part of that 45 second map memorized by rote, and has no interest in longer 3+ minute maps with the same difficulty spikes but a longer "easy" section. Furthermore, the unskilled player will most likely miss more of the easy section than the skilled player on average, which in a system weighted for misses and not combo, works exactly as intended.

The fact of the matter is, you don't know how long it would take for an unskilled player to rote memorize the easy sections of a farm map. A skilled player might consistently miss ~2 notes on the easy section, while the unskilled player doing it for the first few times might miss 10+ before they manage to memorize it after 100 plays. By making the easy sections worth all the PP of a map, you turn both of these players into farmers who earn basically no pp for their plays despite one player being far superior to the other. The skilled player may have to farm less for the FC, but he still has to farm a map that he should have no business needing to farm - and all for the sake of making the unskilled player's job slightly more difficult by permitting zero mistakes instead of 1-2.

Would such a system make it slightly easier for farmers to gain pp? Sure. But in relation to the number of players it would boost to the appropriate ranking, I think it's worth it. Making it so Mr. DT farmer can now miss 2 notes and still get most of their PP in exchange for regular non-farmers/long map players not having to tear their hair out playing the same map over and over in order to get any pp sounds like a good deal to me.

So you want to make the game give more pp even though you know its less accurate for pp?

You do realize that EVERYONE will have an easier time getting pp right?. This means any gains made through chokes will be made by everyone. This means in the end only farmers will gain ranks, we have a system downgrade for ranking accuracy and farmers are more overweighed. This just makes the problems for non-farmers/long map players even worse since they will be punished even more for not retry spamming maps.

abraker wrote:

Top results are considered to be outliers or close to being outliers. Players retrying a bunch of time until they get a score unreasonably above their average skill is an issue. More over, while the difficulty of a pattern increases linearly with respect to their average skill, I've observed difficulty increases logarithmically with respect to samples that display best skill. There might be some way to convert best skill to average skill, but it already makes an assumption that all scores showcase best skill, which is not always the case.

There is no real way to differentiate best skill from average skill when you are looking at plays on the same map. However certain maps and plays are more likely to be best skill rather than average skill. FCs are more likely to be average skill, long maps are more likely to be average skill. This is why they need to be rewarded more in order to rank a player fairly. If you want to convert best skill to average skill, you would need to determine the degree that a map is measuring best skill or average skill. This is probably determined by the notes hit, map length, map type etc

abraker wrote:

I opt to lean toward interpreting every score the player sets to be sampled from their average skill. If they want to retry the map multiple times to boost their apparent skill, then so be it, but then they shouldn't be surprised to learn that their evaluation is overinflated.

Circling back to % threshold, this does up the threshold by some amount. Taking the best result of a true/false test every 50 samples will make you end up with an expected threshold quite higher than 50%. Applying this concept to pp, it would indeed appear inflated. But again, if players do this then they shouldn't be surprised to find that it is inflated.

That's a terrible idea. You should never leave it to the players to ensure that they are ranked fairly. Ideally the path of min/maxing should be a healthy way of playing the game. Rewarding retry spamming is bad game design.

Additionally you allow people to break the system to get an unfair advantage. The system should be designed against min/max behavior because it will rank more players closer to where they deserve to be. That's the whole point of the pp system.

Finally, if you don't care about player choices then why are we punishing farm maps in previous reworks? Why not leave the system broken? Isn't it the players fault for playing overweighed maps?

Its not the players fault. Its the pp systems fault

abraker wrote:

It's like playing with DT. You get phenomenal pp gains at first until you farm everything out and hit a solid wall. In this case, the wall is the logarithmic improvement curve you chose for yourself rather than a linear one. So doing this will bite back the player eventually. I'd label this as "it all balances out at the end".

Except when a player has reached their peak and the best way to rank fairly among other top players is to roll dice in order to play beyond their actual skill. Yeah this is not the way the game should be. We cant eliminate chance altogether but we should at least factor it into ranking.

Fxjlk wrote:

That's a terrible idea. You should never leave it to the players to ensure that they are ranked fairly. Ideally the path of min/maxing should be a healthy way of playing the game. Rewarding retry spamming is bad game design.

It's a problem that both combo and misses face with no clear solution given the information available from scoring data. If scoring data included number of retries the player did, perhaps it could be mitigated.

Fxjlk wrote:

Additionally you allow people to break the system to get an unfair advantage. The system should be designed against min/max behavior because it will rank more players closer to where they deserve to be. That's the whole point of the pp system.

I wouldn't call it an unfair advantage. They would spend time trying to mine out skill for short term gains to be stuck at same skill for longer period of time.

Fxjlk wrote:

Finally, if you don't care about player choices then why are we punishing farm maps in previous reworks? Why not leave the system broken? Isn't it the players fault for playing overweighed maps?

There is a difference between specific maps making it easier to get pp gains, and every map being subject to a specific issue equally.

abraker wrote:

It's a problem that both combo and misses face with no clear solution given the information available from scoring data. If scoring data included number of retries the player did, perhaps it could be mitigated.

It is a problem that is much more relevant to the miss system than the combo system. The combo system is more punishing to breaks on easy notes at the start of the map and so lucky plays are punished harder and you are more likely to see average skill rather than peak skill. Average skill is a more accurate measure because osu is a game with infinite retries and only the best scores count.

Scoring data should never take into account retries on the map. It should take into account the abusability (luck influence) of the map instead. You could get around retry counts by practicing songs offline then play them online which is again, promoting weird player behavior.

abraker wrote:

I wouldn't call it an unfair advantage. They would spend time trying to mine out skill for short term gains to be stuck at same skill for longer period of time.

Retrying maps does not give you skill it gives you pp. pp does not equal skill. They are correlated. If a system can be gamed for short term gains then that is an indication of an inaccuracy of skill measurement in the system. If a proposal is going to make this worse, it should definitely be carefully considered.

Fxjlk wrote:

There is a difference between specific maps making it easier to get pp gains, and every map being subject to a specific issue equally.

The miss rework is going to make all maps more subject to the issue I discussed earlier. Allowing pp on plays where players miss on easy notes (due to being earlier in the map) will make the system more subject to luck.

e

maniaCmiC wrote:

You're writing whole essays and I just can't be asked responding to every claim over a game for weebs

I'm not here to respond to your whole Harry Potter book bro, just read what you're saying and you can dismantle it yourself

We are talking about a complex topic like ranking systems. What did you expect bro? Of course it will be detailed.

If you cant read or understand more than one paragraph then you should probably not be arguing about topics like this. Just accept it and move on.

Fxjlk wrote:

abraker wrote:

It's a problem that both combo and misses face with no clear solution given the information available from scoring data. If scoring data included number of retries the player did, perhaps it could be mitigated.

It is a problem that is much more relevant to the miss system than the combo system. The combo system is more punishing to breaks on easy notes at the start of the map and so lucky plays are punished harder and you are more likely to see average skill rather than peak skill. Average skill is a more accurate measure because osu is a game with infinite retries and only the best scores count.

There should be an expected combo as much as there is an exoected number of misses. Probalistically speaking, both are subject to outliers that display better than average performance as a result. I am looking at this from a 1 combo break or 1 miss perspective. You can have 100/1000 combo or 900/1000 combo, but both correspond to one miss, and it is a matter of retrying the map enough times to get rid of it.

Fxjlk wrote:

Scoring data should never take into account retries on the map. It should take into account the abusability (luck influence) of the map instead. You could get around retry counts by practicing songs offline then play them online which is again, promoting weird player behavior.

Fair point about the work around, but I'd argue that "luck influence" and retries are two sides of the same coin. The more you retry the more likely you are to get that lucky play.

Fxjlk wrote:

abraker wrote:

I wouldn't call it an unfair advantage. They would spend time trying to mine out skill for short term gains to be stuck at same skill for longer period of time.

Retrying maps does not give you skill it gives you pp. pp does not equal skill. They are correlated. If a system can be gamed for short term gains then that is an indication of an inaccuracy of skill measurement in the system.

Yea "mine out skill" was probably a bad way of putting it. "Mine out a number" might have been better.

But anyway, as mentioned in a previous point, combo should be subject to same conditions. Players can retry multiple times to get an smaller combo as much as they can less misses.

Fxjlk wrote:

abraker wrote:

There is a difference between specific maps making it easier to get pp gains, and every map being subject to a specific issue equally.

The miss rework is going to make all maps more subject to the issue I discussed earlier. Allowing pp on plays where players miss on easy notes (due to being earlier in the map) will make the system more subject to luck.

This response fails to address the point of that reply and reiterate a different argument you have. I fail to see how the issue of retrying is related to difficulty in beginning of the map.

maniaCmiC wrote:

It's osu!, it's not that "I can't read or understand more than one paragraph" it's that I don't GAF to respond to your 300-page essay MF The Hobbit books on this shit.

My responses to you are not even that long. You don't need to make an excuse on why you stopped replying. Just stop replying.

maniaCmiC wrote:

> "then you should probably not be arguing about topics like this"

Yes, prestigious topics like osu! should only be discussed by genius intellectuals like yourself, Fxjlk.

Bro, go touch grass, please. Look at yourself, lol

If you cant read what I'm typing, why cant you just not reply instead of whining that you cant read it?

I don't go around complaining that responses are too short or dumb because I'm a 99999 IQ gigabrain intellectual. This quote "you should probably not be arguing about topics like this" is the response to you judging my posts as too long. You are the one whos being pompous and judgmental here.

abraker wrote:

There should be an expected combo as much as there is an exoected number of misses. Probalistically speaking, both are subject to outliers that display better than average performance as a result. I am looking at this from a 1 combo break or 1 miss perspective. You can have 100/1000 combo or 900/1000 combo, but both correspond to one miss, and it is a matter of retrying the map enough times to get rid of it.

However misses are less likely to give pp under a combo system rather than a miss system. Additionally the chokes for combo are more likely to be made by better players since hit chance declines over time. Additionally chokes can be made of more than 1 miss. This is why the miss rework should retain at least some combo scaling

abraker wrote:

Fair point about the work around, but I'd argue that "luck influence" and retries are two sides of the same coin. The more you retry the more likely you are to get that lucky play.

I agree. Though I would say the luck influence is different for misses compared to combo.

abraker wrote:

But anyway, as mentioned in a previous point, combo should be subject to same conditions. Players can retry multiple times to get an smaller combo as much as they can less misses.

They are definitely different metrics though and both are good in certain situations. Why not use both?

abraker wrote:

This response fails to address the point of that reply and reiterate a different argument you have. I fail to see how the issue of retrying is related to difficulty in beginning of the map.

My response isn't in reference to retries. It is in reference to the luck influence being higher for notes at the start of the map compared to notes near the end of the map at the same strain. Notes at the end are harder due to the strain already experienced by the player. Since the luck influence is higher, misses on easy notes at the start and the middle should be punished more. The way to do this is to keep some combo scaling

@maniaCmiC That's enough

Fxjlk wrote:

Additionally the chokes for combo are more likely to be made by better players since hit chance declines over time.

Chokes can happen even by lower skilled players playing low difficulty maps, not just "better players". Chokes happen on maps that are at the right difficulty for the player of some skill, low or high, such that the player is able to almost FC it.

Fxjlk wrote:

Additionally chokes can be made of more than 1 miss. This is why the miss rework should retain at least some combo scaling

So 2 misses then? I can argue the same thing. It's a matter of playing the map enough time to reduce 2 misses to 1 miss.

Fxjlk wrote:

They are definitely different metrics though and both are good in certain situations. Why not use both?

There are challenges getting combo to work properly with the information available from scoring data. Until scoring data includes information about location of combo breaks and the strain at those combo breaks, it's inaccurate guess work. All relevant info we have is

1. Largest combo achieved
2. Number of notes in the map
3. What the highest strain is
4. Number of misses and slider breaks

We know how many combo breaks exist from miss and slider break count. Location of combo break can be determined only if there is 1 combo break. For anything more it's impossible with the given info, save for few rare special cases. We know the combo length for one of the combo breaks, and only one. So there is no information regarding which combo break that is for if there are multiple combo breaks.

The highest strain is just a number. There is no information regarding where it occurs, so we can only assume that one of the combo breaks happen on it. If there is 1 combo break, that gives a definite assumption where the highest strain is. If there are multiple combo breaks, it's impossible to assume where the strain is because it's impossible to know where the combo breaks are.

In short, current info we have only works for 1 combo break case and nothing more.

1. Fc still better than choke.
2. Diffspikes is a problem now.
3. I am sanic.
4. Some random changes can interest you, why not?

i fucking lost on every rework, and this one is going to make me worse again....