nerfing fast non-spaced(overlapped) streams

[GoldenMine0502]

a basecode to nerf acute high bpm stream and doubletap.
because no one could do doubletap on spaced streams.
finally it is multiplied on speedBonus.

but this code make some speed players discourage.
it makes hidamari no uta be taken about 150pp.
and other stream maps be taken about 0-5pp in average.

We could consider to apply this code on angleBonus.

[peppy]

Please change the title to explain what this is doing. Changes made should not be "nerfing" a single beatmap – rather applying a concept.

Documentation? Statistics? Reference to discussion? Anything?

[GoldenMine0502]

In average, the streams have a lot of percentage to hit with doubletap

I think the streams make a lot of normal dt maps be buffed, and hidamari also.

As we only nerf non-spaced and fast stream, we willn’t affect on normal spaced streams like sidetracked day, or etc.

Statisticsally, it makes hidamari no uta be taken about 150pp.
and other stream maps be taken about 0-5pp in average.

we also could consider use this on angleBonus to nerf acute stream buff like https://osu.ppy.sh/ss/17104627/6175 (i didnt apply)

[GoldenMine0502]

I did a lot of changes on other repo
please check this statistics as only VARIANCE

Apol - Hidamari no Uta (-Keitaro) [Expert]
234pp -> 234pp
371pp -> 371pp +HDHR
759pp -> 627pp +HDDT

detail records on below comment

[peppy]

!pp check osu

[GoldenMine0502]

i think it needs more discuss about how should we handle half stacked note

[GoldenMine0502]

i updated my code to get better pp

the statistics for four players related this patch:
https://docs.google.com/document/d/1erjgkqcQOZLWrov-Td-z0WerszFoD5RWoA3xEUKqmPE/edit

(ezdt hidamari no uta player)
User: Anze
Live PP: 11647.7 (including 419.2pp from playcount)
Local PP: 11380.4 (-267.3)

(speed player)
User: aetrna
Live PP: 18736.5 (including 348.8pp from playcount)
Local PP: 18445.7 (-290.8)

(hddt player)
User: mrekk
Live PP: 22127.6 (including 385.4pp from playcount)
Local PP: 22049.1 (-78.5)

(standard stream player)
User: Nameless Player
Live PP: 14002.7 (including 378.6pp from playcount)
Local PP: 14001.7 (-1.0)

as a result, it makes hidamari no uta decreases 150pp,
aetrna plays nerfed 20pp in average,
and other players no affection.

we could consider to alleviate 50 nerf. it needs discuss.

[Wieku]

This code has many issues and has some false claims.
It's reinventing the wheel (using in-place Lerp instead of using Interpolation.Lerp, even if it was used a bit higher), Clamp function is reinvented and it's incorrect causing all >200BPM streams to be nerfed.
And Anze's Hidamari no Uta play is already nerfed by 141.5pp prior to this change:
Anze-master.txt

[GoldenMine0502]

i understand why i alleviates my code much but nerfs a lot of

but it nerfs normal 1/6 (even jumps) patterns i think

maybe applying both things, the weaker would be better

[Wieku]

But this code nerfs all >200BPM streams, not only non-spaced ones. This PR is a wrong way to approach such problem anyways. The problem with Hidamari no Uta is not that stream at the end, but the fact the rest of the map is too easy compared to it.

[GoldenMine0502]

okay, i have slight anxiety on 400bpm+ jump maps, but the solution looks not bad

originally the code is for nerfing acute stream spam

[GoldenMine0502]

i found fatal error on my code and i fixed.

it nerfs anze's hidamari no uta 90pp now (without all >200bpm stream nerf)
and totally it nerfs 200pp+.

in addition, i nerfed the pattern acute stream spam like https://osu.ppy.sh/ss/17106883/b4ed
but this is slight variance.

(if you are depressed, find my stupid code error lol)

this statistics remarked lerp code
anze_aetrna_records2.txt

[GoldenMine0502]

reason why i'm nervious about the lerp code

for example, https://osu.ppy.sh/beatmaps/1954859?mode=osu
this map is 330bpm jump (165bpm) but it is nerfed cuz of it
live 409pp -> 365pp

[peppy]

Please uncommit this one.

[peppy]

Actually this whole PR needs some serious merge resolution (the same code has since been changed by other difficulty changes).

[GoldenMine0502]

this is now up to date.

[stanriders]

Unfortunately this proposal is flawed from the start - you're trying to combat doubletapping by nerfing every stacked stream, without considering if it was doubletapped or not in any way. This issue is not something that can be dealt within difficulty calculation - if a stream is doubletappable that doesn't mean it's easier to play and therefore should be nerfed. There are already measures in performance calculation against doubletapping, as well as at least one pending (#15035) so I don't think we even need something extra in that field.
There are a couple of code issues as well, but they don't really matter if conceptually this should not be merged in the first place. I don't think I should close this PR though, that's up to @smoogipoo

[GoldenMine0502]

In the case of nerfing doubletap, this is not ideal solution and no one could present the best solution as you know. and this is not for only doubletap, but also DT farming maps.
This code doesn't nerf all streams because it just nerfs only speedbonus.

Do you really think past progression is always correct? If then, why we change the algorithm?
why we are struggling stemina on speed value?
speed value is only for physical value. not?

in ONLY the writed code, i acknowledge it needs to modify something. but this code have probablity for nerfing acute stream patterns, or etc. for example, at #15323, i use similar code on NoteVariance.

in newest speed, i can't find any anglebonus so i couldn't apply it.

as a result, i use similar code on other parts, so up to smoogipoo.

[stanriders]

"dt farming maps" is a very broad term. What maps are you talking about? Do they even need to be nerfed? Will they be nerfed by your solution? How much should they be nerfed and why? Are they the only maps that gonna be affected by this nerf? There are a lot of questions regarding reasoning of this PR

[GoldenMine0502]

okay agreed.

i checked whether this nerfs some alternative maps but it works
when all reviews are completed, i will change all of them.

i will close it.