Hi ^^
osu!Taiko has recently gotten an update to its SR calculation, yay!
So far, the SR looks fine (with a couple exceptions), but the algorithm that determines PP has some major flaws I believe.
(Edited a couple of times to be more consistent and informative)
-Show how greatly different speeds of play can be worth the same amount of PP
-Show that high length DT maps can boost over massively higher difficulties of play
Galaxy collapse
99.4% FC, worth 501pp
8.37* with 3585 notes (OD:7)
United (L.A.O.S Remix)
99.4% DT FC, worth 493pp
7.34* (DT included) with 2200 notes (OD:7^)
Notes on comparison (example #1)
8.37* = 501pp
7.34* = 493pp
Wait what?
So what we see here is that 1 whole star higher of difficulty is being worth basically the same amount of PP as the map ranked as 1 less star. United is obviously getting some OD changes due to DT, but still, I think the difference between plays like these should be wayyy larger. (DTHD on United will be worth way more than Galaxy Collapse SS nomod)
A little example from these maps to show the difference
- Galaxy Collapse has a 67 length 320BPM part, recognized as the difficulty spike of this map.
- United DT has a 31 length 263BPM part, which is the hardest part in terms of speed.
31 length 263bpm is worth almost the same PP as a 67 length 320bpm under the current pp calculations... Think about that for a minute.
What feels like is wrong currently
-DT on long maps gets too much PP, it doesn't need to be that fast either. This might be because the OD gets high, and the pp system values high OD and length combined?
From what I've seen, 260 ish BPM maps that have a 3+ minute length will be worth more than my other nomod plays ever will be. Higher speed maps (no-mod or low length) gets buried under PP plays like these.
DT: Long maps, medium SR
Currently, you can boost your PP by playing long uncomplicated DT maps. United is the perfect example for this, as it has no rhythm complexities and no big variation in patterns. In my opinion: Maps should only get increase in PP if you do this to them:
- Introduce new patterns
- Increase speed
- Increase length of streams/patterns (soft/hard-capped)
- Increase density
- Very slight PP increase if same patterns are repeated (to test consistency)
- Introduce new rhythms
- Changes in rhythm
- Changes in speed (?)
PP should NOT increase on
- Uncomplicated slow parts
- Solely note count
A little note on HD
Because the difference in PP between very high speed maps is so low, HD becomes essential to keep up in higher levels of play, as it scales so well. Example: A no-mod play of an 8.8* map is worth the same as an 8.0* map with HD (assuming the accuracy is the same). A 0.8* difference is way too big in my opinion. (HD might be balanced in lower levels of play. A soft-cap might fix this problem, or boost high speed)
Also, HD does not increase technical difficulty or timing (same inputs as nomod), same with SV (speed of which notes appear). I think that either HD should not increase PP, OR SV should have an impact on star rating(SR) (SV currently does not).
8.37* (DT included) with 2233 notes (OD:6.5^)
So what's so special about this map? (example #2):
Colorful is the top play of many top players (examples: _yu68, Peaceful, ekumea1123, shake_dan). What does that mean? Does it mean that this is the hardest score with best accuracy all these top players have? Most likely not. Let's take _yu68 as an example. Is his DTHD play on this map the best ranked play he has ever set? Playing 263 bpm, with not very dense 1/6 parts? I'm 99% sure he has set better plays on other maps than that, as I've seen him play 340 bpm semi-dense 1/6 with high accuracy (example: Necrophagiatoast).
I think for top-players competing to become the best, setting the best scores, it has to be very frustrating to see their hardest plays get buried under uncomplicated and slower maps.
Final notes (TL/DR):
- Simple 1/4 maps with high length can be super over-weighted for PP, especially if OD is high.
- High speed PP plays and no-mod PP plays are currently buried under high length DT PP plays.
- High speed maps needs a greater PP or SR difference from lower speed maps. This game only has 2 different inputs basically, so speed is very very important!
- (HD often unbalanced at high SR)
We can make osu!taiko ranking more realistic/better! Think how demotivating for players yearning for improvement and rank, but the only way they can climb the rankings, is by playing long and uncomplicated maps, which gives more PP than anything else..
Please point out errors and come with constructive criticism
Or show me that you agree, if you do^^
osu!Taiko has recently gotten an update to its SR calculation, yay!
So far, the SR looks fine (with a couple exceptions), but the algorithm that determines PP has some major flaws I believe.
(Edited a couple of times to be more consistent and informative)
Example #1: Galaxy Collapse vs United
This example is used to-Show how greatly different speeds of play can be worth the same amount of PP
-Show that high length DT maps can boost over massively higher difficulties of play
Galaxy collapse
99.4% FC, worth 501pp
8.37* with 3585 notes (OD:7)
Additional Info
-BPM Varies: 270, 300, 320
-Hardest part: A 67 note-length 320bpm stream (Right after a while of 320bpm playing) non-repeating pattern
-1/4 Rhythm
-Length: 6:44
Beatmap: beatmapsets/328117#taiko/727903
(Watch a replay, or play the map if you want better understanding)
-Hardest part: A 67 note-length 320bpm stream (Right after a while of 320bpm playing) non-repeating pattern
-1/4 Rhythm
-Length: 6:44
Beatmap: beatmapsets/328117#taiko/727903
(Watch a replay, or play the map if you want better understanding)
United (L.A.O.S Remix)
99.4% DT FC, worth 493pp
7.34* (DT included) with 2200 notes (OD:7^)
Additional Info
-BPM:263
-Hardest part: Up to 31 note-length of 263bpm patterns
-1/4 Rhythm
-Almost only odd note-length patterns
- Length 3:41 (DT included)
Beatmap: beatmapsets/559371#taiko/1183029
(Watch a replay, or play the map if you want better understanding)
-Hardest part: Up to 31 note-length of 263bpm patterns
-1/4 Rhythm
-Almost only odd note-length patterns
- Length 3:41 (DT included)
Beatmap: beatmapsets/559371#taiko/1183029
(Watch a replay, or play the map if you want better understanding)
Notes on comparison (example #1)
8.37* = 501pp
7.34* = 493pp
Wait what?
So what we see here is that 1 whole star higher of difficulty is being worth basically the same amount of PP as the map ranked as 1 less star. United is obviously getting some OD changes due to DT, but still, I think the difference between plays like these should be wayyy larger. (DTHD on United will be worth way more than Galaxy Collapse SS nomod)
A little example from these maps to show the difference
- Galaxy Collapse has a 67 length 320BPM part, recognized as the difficulty spike of this map.
- United DT has a 31 length 263BPM part, which is the hardest part in terms of speed.
31 length 263bpm is worth almost the same PP as a 67 length 320bpm under the current pp calculations... Think about that for a minute.
What feels like is wrong currently
-DT on long maps gets too much PP, it doesn't need to be that fast either. This might be because the OD gets high, and the pp system values high OD and length combined?
From what I've seen, 260 ish BPM maps that have a 3+ minute length will be worth more than my other nomod plays ever will be. Higher speed maps (no-mod or low length) gets buried under PP plays like these.
DT: Long maps, medium SR
Currently, you can boost your PP by playing long uncomplicated DT maps. United is the perfect example for this, as it has no rhythm complexities and no big variation in patterns. In my opinion: Maps should only get increase in PP if you do this to them:
- Introduce new patterns
- Increase speed
- Increase length of streams/patterns (soft/hard-capped)
- Increase density
- Very slight PP increase if same patterns are repeated (to test consistency)
- Introduce new rhythms
- Changes in rhythm
- Changes in speed (?)
PP should NOT increase on
- Uncomplicated slow parts
- Solely note count
A little note on HD
Because the difference in PP between very high speed maps is so low, HD becomes essential to keep up in higher levels of play, as it scales so well. Example: A no-mod play of an 8.8* map is worth the same as an 8.0* map with HD (assuming the accuracy is the same). A 0.8* difference is way too big in my opinion. (HD might be balanced in lower levels of play. A soft-cap might fix this problem, or boost high speed)
Also, HD does not increase technical difficulty or timing (same inputs as nomod), same with SV (speed of which notes appear). I think that either HD should not increase PP, OR SV should have an impact on star rating(SR) (SV currently does not).
Example #2: Colorful
Colorful8.37* (DT included) with 2233 notes (OD:6.5^)
Additional Info
- 263 BPM
- Mostly 1/4. Small parts of 1/6
- Hardest part: Somewhat dense 263bpm small streams with 1/6 parts
- Diff spikes ^
- Some parts of 263bpm death streaming (60 ish note length)
- Length 03:33 (DT included)
- Beatmap: beatmapsets/899738#taiko/1879405
- Mostly 1/4. Small parts of 1/6
- Hardest part: Somewhat dense 263bpm small streams with 1/6 parts
- Diff spikes ^
- Some parts of 263bpm death streaming (60 ish note length)
- Length 03:33 (DT included)
- Beatmap: beatmapsets/899738#taiko/1879405
So what's so special about this map? (example #2):
Colorful is the top play of many top players (examples: _yu68, Peaceful, ekumea1123, shake_dan). What does that mean? Does it mean that this is the hardest score with best accuracy all these top players have? Most likely not. Let's take _yu68 as an example. Is his DTHD play on this map the best ranked play he has ever set? Playing 263 bpm, with not very dense 1/6 parts? I'm 99% sure he has set better plays on other maps than that, as I've seen him play 340 bpm semi-dense 1/6 with high accuracy (example: Necrophagiatoast).
I think for top-players competing to become the best, setting the best scores, it has to be very frustrating to see their hardest plays get buried under uncomplicated and slower maps.
Final notes (TL/DR):
- Simple 1/4 maps with high length can be super over-weighted for PP, especially if OD is high.
- High speed PP plays and no-mod PP plays are currently buried under high length DT PP plays.
- High speed maps needs a greater PP or SR difference from lower speed maps. This game only has 2 different inputs basically, so speed is very very important!
- (HD often unbalanced at high SR)
We can make osu!taiko ranking more realistic/better! Think how demotivating for players yearning for improvement and rank, but the only way they can climb the rankings, is by playing long and uncomplicated maps, which gives more PP than anything else..
Please point out errors and come with constructive criticism
Or show me that you agree, if you do^^