Scoring Data in Brutal Difficulty Range

**xXOpkillerXx** · 08-27-2019, 09:42 AM

Re: Scoring Data in Brutal Difficulty Range

Originally posted by Dynam0

My main concern with that approach is pattern manipulation and stamina. Getting those right would be a tough ask imo. I still think it is far less tedious to have subjective difficulty assignments and as Rob said we just need to get the decay part correct. It's not incredibly far off at this point.

Can you give me specific examples ? Either from actual files or just made up sections ? It's hard to get into details without concrete examples.

PS: I already have computed the manipulatable sections mostly. That's dealt with by accounting for the number of frames you have to hit singles as jumps (1 frame being harder to manip, 2 frames is most likely a jump). As for stamina, I explained a basic framework for it; what would you disagree with ?

Edit: I also disagree with "it's not incredibly far off"; far off what ? The only thing the current system can hope to reach is optimal subjective consensus (as in the most people who agree with the difficulties). While that isn't a bad metric per se, it is inevitably flawed and biased, and doesnt solve the problem: variable decay is still just a decay, it doesnt fully compensate for chart structure, it just covers cases where the distribution of ordered note difficulty ressembles one unique function with a modifyable decay. The truth is there can be many more shapes to that distribution.

**Dynam0** · 08-27-2019, 10:28 AM

Re: Scoring Data in Brutal Difficulty Range

Idk man I just think taking this granulated an approach to difficulty is not worth the effort and it's always going to be prone to outliers reliant on fudge-factoring. Who determines how difficult a trill is a 240bpm? Someone who is good at them? The average player? Who is an average player? You see how silly the concept of automating this is? It's entirely subjective. There will still be 97s that feel like 96s and so on based on individual player strengths.

**xXOpkillerXx** · 08-27-2019, 10:40 AM

Re: Scoring Data in Brutal Difficulty Range

Originally posted by Dynam0

Idk man I just think taking this granulated an approach to difficulty is not worth the effort and it's always going to be prone to outliers reliant on fudge-factoring. Who determines how difficult a trill is a 240bpm? Someone who is good at them? The average player? Who is an average player? You see how silly the concept of automating this is? It's entirely subjective. There will still be 97s that feel like 96s and so on based on individual player strengths.

That pattern approach never worked for me. The questions you're asking are legit ones, but that's like surface level. I'd really appreciate to debate more on the points I mentionned but you have to be willing to talk details, otherwise this discussion goes nowhere (which is often the case). I'm not saying a calculator will achieve 100% agreement from everybody on every difficulty, but it would give a much better framework to argue on.

**Dynam0** · 08-27-2019, 10:45 AM

Re: Scoring Data in Brutal Difficulty Range

Well considering my prior post stated that going into said details is not worth the effort, I am out of this discussion :P
Best of luck though, if you do figure out a way to do it then I'll be proven wrong and then some

**SputnikOwns** · 08-27-2019, 01:29 PM

Re: Scoring Data in Brutal Difficulty Range

Correct me if I'm wrong, but isn't it quite rare to have an AAA equal to or above one's rating? The algorithm is excellent as far as I'm concerned -- so long as the song difficulties are correct.

**Matthia** · 08-27-2019, 01:44 PM

Re: Scoring Data in Brutal Difficulty Range

Originally posted by SputnikOwns

Correct me if I'm wrong, but isn't it quite rare to have an AAA equal to or above one's rating? The algorithm is excellent as far as I'm concerned -- so long as the song difficulties are correct.

shit like this happens but only because of how absurdly low the difficulty of this file outside of two or three main hard parts is compared to what it is at currently

Edit: It is very likely that White Walls Part 2 is also seeking a nerf not much in AAA difficulty but rather the How-Easy-This-Can-Be-Abused factor which is the overall discussion we are having at the moment

**xXOpkillerXx** · 08-27-2019, 02:07 PM

Re: Scoring Data in Brutal Difficulty Range

Originally posted by SputnikOwns

Correct me if I'm wrong, but isn't it quite rare to have an AAA equal to or above one's rating? The algorithm is excellent as far as I'm concerned -- so long as the song difficulties are correct.

The thing is they're not, which is why this thread exists

**SputnikOwns** · 08-27-2019, 02:40 PM

Re: Scoring Data in Brutal Difficulty Range

That's totally fine of course. They should be ordered correctly according to top players. No need to adjust the algorithm for 100+ though.

**xXOpkillerXx** · 08-27-2019, 02:45 PM

Re: Scoring Data in Brutal Difficulty Range

Originally posted by SputnikOwns

That's totally fine of course. They should be ordered correctly according to top players. No need to adjust the algorithm for 100+ though.

This is the second problem. As it was said a few times, there are a bunch of files which are hard AAAs but easy SDGs, and that breaks the current system.

**One Winged Angel** · 08-28-2019, 03:40 PM

Re: Scoring Data in Brutal Difficulty Range

I agree with Dynamo, honestly there are far too many variables to deal with to ensure a calculator will work effectively. Manually establishing separate decay formulas for charts that are more accessible to less skilled players will reach higher levels of accuracy and communal consensus without the hassle of dealing with all the intricacies needed to be taken into account for a calculator in a game that encodes charts at 30fps. A 16th roll at 300bpm starting on a quarter note can effectively be jumptrilled because the one frame gaps all coincide between arrows hit with the same hand (1/2 and 3/4), but offset the roll by a 16th and now all the one frame gaps appear at 2/3 and 4/1. Having to account for quite literally the same pattern being variable in difficulty depending on the starting note, on top of everything else that needs to be taken into consideration...I personally wouldn't want to invest any time in that.

What Dynamo was fudging around with in Excel looks interesting and I might mess around with that later.

Tangential to that, I've looked at suggestions and scoreboard data and am interested in feedback for this structure of 98+:

A few notes:

* denotes charts that would benefit from separate equivalency decays. I'm not suggesting they should share identical formulas, but a manipulation of the current formula is required.

La Camp's scoring data after a decade in game suggests it can hang with the 99s. Structurally I may not entirely agree with that shift but it fills out the 99s a bit better for the time being. I know there's more charts queued that will end up in this range so stuff will probably move a bit.

I'd rather move M8BT up than AT down. I don't think a gap of two difficulty points exists between those two charts which is why AT didn't also move down.

Punkture is Punkture. There's players that think it shouldn't even be a 95 but an overwhelming majority thinks otherwise. Scoring data is vomit and nothing else exists to compare it to. 97 at the min works, 98 is likely okay.

_.Pulse isn't here because I'm projecting a shift to 97. If people think 98 is more appropriate then sure.

I wanted to bring some 101s up to lessen how many charts exist in that tier but none of them feel as difficult as what's included in the tier above. I feel there's a hard line between the 101s and 102s. Also considered breaking up the 101s into two tiers and pushing 102+ to 103+ but the 101s truly feel quite similar in difficulty to me, and I don't think there's a two point gap between anything that would be considered a lower end 101 and the current 102s. Think we might just need more charts in the 102/103 range (DZ resubmit Apocynthion..)

Feedback appreciated.

**SK8R43** · 08-28-2019, 04:12 PM

Re: Scoring Data in Brutal Difficulty Range

I dont think a perfect rating system can exist, ever.

Everyone has different strengths and weaknesses which affect the difficulty of a song viewed by them. Some people think songs are easy while others think the same song is hard. I have pretty much a better score on every 86 than i do on the 85's for instance. I feel like they could be switched and still be fine.

I'm not really sure that its possible to win in this situation. I see that OWA mentioned Punkture and that it could be a 98 but since jacks are my strength i feel like thats way to high of a level. Ive gotten to the end with a low SDG fc multiple times. Ive also SDG the ending in multiple runs. my score on it isnt the best, but ive gotten better PA 1 or 2 miss runs. Compared to other 95-98s my score on this song is quite good.

If you guys could figure out something that works better that would be sweet, but i think its near impossible to get it perfect.,

**xXOpkillerXx** · 08-28-2019, 04:18 PM

Re: Scoring Data in Brutal Difficulty Range

Originally posted by One Winged Angel

I agree with Dynamo, honestly there are far too many variables to deal with to ensure a calculator will work effectively. Manually establishing separate decay formulas for charts that are more accessible to less skilled players will reach higher levels of accuracy and communal consensus without the hassle of dealing with all the intricacies needed to be taken into account for a calculator in a game that encodes charts at 30fps. A 16th roll at 300bpm starting on a quarter note can effectively be jumptrilled because the one frame gaps all coincide between arrows hit with the same hand (1/2 and 3/4), but offset the roll by a 16th and now all the one frame gaps appear at 2/3 and 4/1. Having to account for quite literally the same pattern being variable in difficulty depending on the starting note, on top of everything else that needs to be taken into consideration...I personally wouldn't want to invest any time in that.

Could you elaborate on this ? I'm really not sure what you mean there.

Also, the calc I have works on .swf, not on .sm, so the whole conversion point is irrelevant; the difficulty is computed with the resulting converted chart.

**One Winged Angel** · 08-28-2019, 05:12 PM

Re: Scoring Data in Brutal Difficulty Range

Charts like Husigi where over 90% of the notes are understandable at a much lower skill level than the difficulty rating would suggest, but mastery of that last < 10% requires being able to play the difficulty range they sit in.

**AutotelicBrown** · 08-28-2019, 06:26 PM

Re: Scoring Data in Brutal Difficulty Range

The decay should change according to how hard is getting a reasonable non-AAA score in the spiky section(s), and how hard the rest of the file is to AAA.

Anyway, an approach like Dynam0's formula is probably the most pragmatic solution. My suggestion would be trying to operate over the raw numbers of goods and then using the formula that already exists.

For instance, if the current formula is f(x, A) where x is the raw number of goods and A the current AAA-based difficulty, change it to some f(g(x,B),A).
g(x,B) being some formula where g(x,0) = x (I) and g(x,B) > x for B > 0 (II).
A simple example would be g(x,B) = ((5*x)^(1+c*B))/5, with c being some arbitrary constant to make the range of B suitable to show in the game interface (preferably an integer), and the 5 factor being there just to avoid oddities with 0.2~0.8 values.

(I) - So that by default any song would have B=0 and keep the usual formula
(II) - So silly stuff would be assigned some appropriate B value as needed to force a faster decay

I think this would be fairly intuitive to work with but I'm too lazy to test it out

On a side note, making a general difficulty calculator is trying to solve an even harder problem and trying to make inferences based on the inevitably noisy results.
While I don't think making one is a lost cause, it's way out of scope for what FFR warrants.

**One Winged Angel** · 09-30-2019, 08:13 PM

Re: Scoring Data in Brutal Difficulty Range

Discuss?

Obviously not in charge of difficulties or have final say for this but a lot of thought went into this list and will back up any placement decisions if necessary. I don't think anything on this list is off more than 1 point.

Scoring Data in Brutal Difficulty Range

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment