FFRmas Day 12

[Wayward Vagabond]

i have tons of easy files but you guys rejected them because they didnt fit your definition of a fun file.

[_Zenith_]

Originally Posted by ssbmchamp

(This comment is not directed at you Zenith, but rather all of the difficulty consultants)

LMFAO I may not be an official difficulty consultant but anyone could tell you this day 6 song does not deserve the same rating as Azul (Remix)

Played the day 6 song with cold hands, and as a sightread, came close to a SDG...played Azul immediately after (which i was already very familiar with the file) and got 75+ goods....something's not right with that picture.

If people refuse to increase Azul to at least an 80 (which it should be) then this day 6 song should be 75-77. I would bet anything that a week or so from now you will see much better scores on this day 6 song from the same people (D6 or lower) than you will on Azul.

You guys can analyze and compare these songs to other files all day long and come up with a rating but at the end of the day, what a lot people score on it will ALWAYS be the best indicator of the appropriate rating for a song.

I'll speak in terms of difficulty consultants on their behalf since it's a logical point that would probably be made:

Azul and Waltz are two entirely different skillsets

Waltz is more technical and has multiple sections that are easier to drop PA on
Azul is more consistent and has two parts that are easy to drop PA on

Never forget the AAA-based ratings, Azul is like Serious Shit where you have it or don't (in terms of high D5) and Waltz is like Cirno where you practice the technical parts and then fight to AAA the sections that you weren't practicing.

I'd naturally make the assumption that you'll see more AAAs on Azul than Waltz given a time period, and they both can easily be 79 because different skillsets even when comparing the two that Azul is more consistently dense than Waltz.

[rzr]

Originally Posted by Wayward Vagabond

i have tons of easy files but you guys rejected them because they didnt fit your definition of a fun file.

Would be nice to have more to practice on novice files rather than just filtering anything lower than a 40 and hoping it doesn't have hands.

[MinaciousGrace]

Originally Posted by ssbmchamp

what a lot people score on it will ALWAYS be the best indicator of the appropriate rating for a song.

i suggest you take some classes in applied statistics/statistical modeling and familiarize yourself with the foundational concepts of indirect estimation so you can realize on a conceptual level how stupid this argument is

in other news aaa based ratings still causing more problems than they solve since 2003

[gold stinger]

Originally Posted by _Zenith_

Azul and Waltz are two entirely different skillsets

ty

for me:
Jacks > speedup/slowdown > dense js > bursty > stamina 4+mins

in other news, AAA on waltz but 4g on Azul

[blanky!]

Originally Posted by MinaciousGrace

in other news aaa based ratings still causing more problems than they solve since 2003

I wonder if single digit good based ratings are a better indicator of difficulty, which seem to account for some of the more infamous difficulty spikes of FFR's charts.

eDIT: This reminds me, why are we still counting FC's instead of SDG's in profile stat bars.

[_Zenith_]

SDG ratings wouldn't be bad but AAA isn't at all the worst method either; we try to place ratings on a AAA-to-Division scale (maybe not the main DCs but imo it's what I do) and seeing how the two 79 charts this event have all had AAAs from D6-D7 players.

Closest AAA to Azul is: hitlerisright (questionable) and sweet2kill210 and they aren't SDGs

Closest AAA to Waltz is: sweet2kill210 @ 11-2-0-3

AAA ratings thus far have proven to be really accurate in these cases but older files would potentially be better with SDG-to-division ratings but that's a whole 'nother topic

[MinaciousGrace]

either sdg or % based systems are superior

aaa ratings place a greater emphasis on ntrials versus actual skill level

in other words

its a shittier measurement of skill

[ssbmchamp]

Originally Posted by _Zenith_

I'll speak in terms of difficulty consultants on their behalf since it's a logical point that would probably be made:

Azul and Waltz are two entirely different skillsets

Waltz is more technical and has multiple sections that are easier to drop PA on
Azul is more consistent and has two parts that are easy to drop PA on

Never forget the AAA-based ratings, Azul is like Serious Shit where you have it or don't (in terms of high D5) and Waltz is like Cirno where you practice the technical parts and then fight to AAA the sections that you weren't practicing.

I'd naturally make the assumption that you'll see more AAAs on Azul than Waltz given a time period, and they both can easily be 79 because different skillsets even when comparing the two that Azul is more consistently dense than Waltz.

I appreciate the response. I guess my problem i still have with the ratings is this:

you made a great point in identifying both files have 2 parts where it's easy to drop PA and both are different skill sets. I couldn't agree with you more! However I feel as if the skill sets required in Azul are more demanding, justifying a higher rating.

The 2 parts in Waltz are early on and the last 3rd of waltz is a cake walk.

The 2 parts in Azul are literally right at the end and are very stamina draining.

Originally Posted by MinaciousGrace

i suggest you take some classes in applied statistics/statistical modeling and familiarize yourself with the foundational concepts of indirect estimation so you can realize on a conceptual level how stupid this argument is

in other news aaa based ratings still causing more problems than they solve since 2003

I actually have taken applied statistics and did very well in the class thank you very much.

you don't need to take a class to look at hard, factual data, to say "hmmmm this song is harder than that one" and make an educated guess on what the rating should be.

the songs may have different skill sets but if on one song 10 people AAA it and on the other 40 people AAA it, common sense tells me that one skill set is much easier to acquire than the other, therefore justifying a lower rating. sometimes common sense is better than trying to sound like some mathematical genius that's over analyzing the shit out of something that doesn't need to be this complicated.

[psychoangel691]

Originally Posted by Wayward Vagabond

i have tons of easy files but you guys rejected them because they didnt fit your definition of a fun file.

Don't lump me in that comment.




I didn't read in depth all the difficulty discussion going on, personally I don't rate things based on AAA difficulty because that means a file that's easy with one hard pattern can be jumped up significantly. With raw scoring being the standard now I think that going by AAA difficulty is not the best measure as well.
I generally have differing opinions on the files difficulty than what is put, but it's not easy to just rate a file when no one's played it yet either. The consultants do the best they can and if things have to be adjusted they'll be adjusted.

[_Zenith_]

Originally Posted by MinaciousGrace

either sdg or % based systems are superior

aaa ratings place a greater emphasis on ntrials versus actual skill level

in other words

its a shittier measurement of skill

% based system in fiffer is currently broken (AA is something like 99+% which is janks while A is anything from 90-98%)

it's true that SDG based system would outrule AAA here although switching the entire system would take a large chunk of time as there are 1900+ songs in-game with more definitely coming between now and projected time of finishing the system switch; not in the least to say it wouldn't be worth it.

@sweet, I look at Gunkienen Tarasuigin when I look at Azul, both are roughly the same sketch of each other with the exception that Azul is broken JS while Gunk has more of a defined burst-JS section spread throughout entire chart.

at the lowest Waltz would be 77-78 still because of how the intro is fast and dense while the trilling section asks to be extremely technical (in terms of D5 skill) even if the outro is stupid simple

[ssbmchamp]

Originally Posted by _Zenith_

% based system in fiffer is currently broken (AA is something like 99+% which is janks while A is anything from 90-98%)

it's true that SDG based system would outrule AAA here although switching the entire system would take a large chunk of time as there are 1900+ songs in-game with more definitely coming between now and projected time of finishing the system switch; not in the least to say it wouldn't be worth it.

@sweet, I look at Gunkienen Tarasuigin when I look at Azul, both are roughly the same sketch of each other with the exception that Azul is broken JS while Gunk has more of a defined burst-JS section spread throughout entire chart.

at the lowest Waltz would be 77-78 still because of how the intro is fast and dense while the trilling section asks to be extremely technical (in terms of D5 skill) even if the outro is stupid simple

I agree that 77-78 would be a more appropriate rating for Waltz

[_Zenith_]

am awaiting rob to come in here and go

"quit bickering nerds"

"walts is 79 deal with it "

<3

[MinaciousGrace]

Originally Posted by ssbmchamp

you don't need to take a class to look at hard, factual data, to say "hmmmm this song is harder than that one" and make an educated guess on what the rating should be.

the songs may have different skill sets but if on one song 10 people AAA it and on the other 40 people AAA it, common sense tells me that one skill set is much easier to acquire than the other, therefore justifying a lower rating. sometimes common sense is better than trying to sound like some mathematical genius that's over analyzing the shit out of something that doesn't need to be this complicated.

do you really want to get into this with me

@zenith technically no development or effort being put into ffr is "worth it" from any conventional standpoint but that certainly hasn't stopped people from spending hours working on it and it's no reason to cling to an antiquated rating system that can be vastly improved with minimal effort

[One Winged Angel]

Originally Posted by MinaciousGrace

either sdg or % based systems are superior

aaa ratings place a greater emphasis on ntrials versus actual skill level

in other words

its a shittier measurement of skill

AAA-based difficulties may not be as desirable for the algorithm used for skill rating, but when the end goal for scoring on a file for 95+% of players on FFR is to AAA it, and token content unlock conditions are predicated on AAA'ing files of specific difficulty, it certainly seems to be the most appropriate way to ascertain difficulties. If neither of those two conditions existed, and if the FFR difficulty ceiling wasn't so piss poor, I would be much more in favor of a % based system.

Given the nature of files on SM it makes a fuckton more sense to create an algorithm that assesses difficulty based on how hard something is to AA or reach whatever %. FFR apparently has letter grading conditions for scoring too. AA is 99% raw score, A is 90 or 95% idfk, the point is literally no one gives a shit and I've seen maybe three people post about breaking a certain percentage threshold on a file. They all play to ultimately AAA it.

[hi19hi19]

I suggested waltz at 78. Does feel easier than Azul to me, but since this is FFR and FFR only cares about AAAs, I admit it's harder to reliably AAA.
If you're fast enough to do Azul, you pretty much can AAA it no problem, which isn't really what's going on in waltz


Besides ratings on FFR ¯\_(ツ)_/¯ anyway lol

[MinaciousGrace]

yeah the problem with % ratings in ffr is you'd have to redesign the entire scoring system from the ground up to make it worth using so i wouldn't advocate that particularly

i still think moving to sdgs makes more sense than not doing so even without implementing an automated system; it aligns with the somewhat shifting focus away from AAA focused meta, and also is healthier in the long term.

[YoshL]

I think in the end, the community has grown small enough that a change like moving away from AAA based score mindsets to SDGs or even 99% raw score isn't that bad. Sure, many skill token unlocks and other such things are based on AAAs, but I think in the end, dwelling on the past and making attempt to keep things the same as they always were isn't necessarily the right way to do things?

starting with a 60 fps engine that converts directly from the .sm to get accurate charts (hue)

[One Winged Angel]

also I replayed waltz and I'm changing back to 78

I have a ton of things I could talk about for a proper way of structuring and assessing difficulties on FFR but quite frankly (and other people are well aware) I don't care that much at this point and a large reason as to why I'm even still here is because I haven't found anyone to pass off doing song difficulties to (although corn and I tend to rate unreleased files under 89ish nearly identically a large majority of the time but I'm sure he also doesn't care). Changing to SDG-based difficulties wouldn't result in more than a +/-2 change for the hardest third of files in game, obviously barring any files with extreme spike spurts for 50 arrows or something. They would just need to be combed through again which really wouldn't be that time-consuming by anyone competent (read: unbiased). Under 66 is a fucking mess like I've said many times before and needs a complete restructuring anyways.

[psychoangel691]

Yeah all the lower ratings are really screwy. I don't think anyone really wants to devote the time right now to try to adjust everything.