A new beginning for our SM community.

[customstuff]

Originally Posted by Shikari

Please, do that. I appreciate that you came up with it on your own, without anyone asking, really. If you can do it, I'll add it to the Tier Points system.

Alright, I'll try my best. I can't guarantee I'll be perfectly accurate, but I'll do what I can.

So should I basically add a notepad file in the pack with 1-100 difficulty ratings then?

[Xiaounlimited]

Original post about NPS rating system:
http://www.keybeatonline.com/forums/...08&postcount=1

This works incredibly well for 6k. I don't know how it will translate to 4k due to the variance in densities (a raw 10nps file on 4k will be incredibly harder than a raw 10nps difficulty on 6k). The amount of fudging required for 4k may be unreasonable due to the presence of patterns not seen as prominently in 6k (jacks, chordjacks, bursts). I still prefer the NPS scale, but a subjective rating system might be viable here as well.

I don't agree with the concept of a 1-100 system. Look at KBO's proposed 1-50 system: there was absolutely no line between the different levels other than relative difficulty. When there's a paradigm shift later on and even harder files are introduced into the scene, you have to shuffle the difficulties of files around. That becomes a pain in the ass, trying to determine something along the lines of, "Is this 85 now a 79, or a 76?" Such minute increments in difficulty become almost worthless, IMO. There's also the fact jumping up difficulties feels much less rewarding to a player.

Anyway, this is just about ratings. Shouldn't get too hung up on it. It's probably the least important thing right now.

[customstuff]

I do sort of agree with Xiaounlimited. I think something like a 1-20 difficulty system would work better.

[Shikari]

Originally Posted by Xiaounlimited

I don't agree with the concept of a 1-100 system. Look at KBO's proposed 1-50 system: there was absolutely no line between the different levels other than relative difficulty. When there's a paradigm shift later on and even harder files are introduced into the scene, you have to shuffle the difficulties of files around. That becomes a pain in the ass, trying to determine something along the lines of, "Is this 85 now a 79, or a 76?" Such minute increments in difficulty become almost worthless, IMO. There's also the fact jumping up difficulties feels much less rewarding to a player.

Here's my argument: How many files does KBO have? How many files does FFR have? Having a small range of difficulties is easy for them to keep, but look at Stepmania, look at the multitude of files and packs we have out there. 1-100 seems completely fair too me. Basing achievements on this range seems easier, since we can award people who are improving based in these numbers. I mean, a newcomer starts with 1-10 files, gets better, start AAA'ing them easily, then go for 11-20, then 21-30, and so on. We can plan achievements based in every multiple of 10, so people can keep playing and improving to achieve these tasks. About changing difficulties based on players' skills, like FFR did and still does, I don't think it's right. It'd be unfair with people that's totally new and unused to the game, that's why we need to come up with a fair system for all, and I really, really don't believe that's it's too hard.

Also, please, don't think I'm overplanning this difficulty/TP system. I think we all agree that making an unique diff system will be easier to compare files, because, to new people, a 5 (Xoon rating) being harder than 12 (old-school rating) seems a little nonsense, and a TP system will atract new people to the game, and give reasons for nowadays players to keep playing Stpemania.

[sakura080789]

Originally Posted by Shikari

Here's my argument: How many files does KBO have? How many files does FFR have? Having a small range of difficulties is easy for them to keep, but look at Stepmania, look at the multitude of files and packs we have out there. 1-100 seems completely fair too me. Basing achievements on this range seems easier, since we can award people who are improving based in these numbers. I mean, a newcomer starts with 1-10 files, gets better, start AAA'ing them easily, then go for 11-20, then 21-30, and so on. We can plan achievements based in every multiple of 10, so people can keep playing and improving to achieve these tasks. About changing difficulties based on players' skills, like FFR did and still does, I don't think it's right. It'd be unfair with people that's totally new and unused to the game, that's why we need to come up with a fair system for all, and I really, really don't believe that's it's too hard.

Also, please, don't think I'm overplanning this difficulty/TP system. I think we all agree that making an unique diff system will be easier to compare files, because, to new people, a 5 (Xoon rating) being harder than 12 (old-school rating) seems a little nonsense, and a TP system will atract new people to the game, and give reasons for nowadays players to keep playing Stpemania.

with all these different changes to the rating system people either want 1-100, 1-50 , 1-20. why not vote on it?

[Oni-Paranoia]

@Xiao - You wouldn't have to shuffle difficulties if you did things right the first time. Make a ridiculously hard song a 100 (preferably something impossible for this era). That way, people work there way up rather than re-rating files in the future.

People misunderstand how to even rate a file correctly. Your rating is 1-100 meaning 1 is the easiest, 100 is the hardest. If a 100 can be passed, then the scale is either A) Going to grow like ITGs 1-12 scale did or B) Files are going to be re-rated and achievement crap is going to get confusing like FFR.

@Sakura - A lot of people don't understand what contributes to a rating system. I don't see why it should be voted upon, I think what everyone needs to do is layout the pros and cons so people can discuss without people restating already said facts.

[Wineandbread]

Originally Posted by Shikari

Here's my argument: How many files does KBO have? How many files does FFR have? Having a small range of difficulties is easy for them to keep, but look at Stepmania, look at the multitude of files and packs we have out there. 1-100 seems completely fair too me. Basing achievements on this range seems easier, since we can award people who are improving based in these numbers. I mean, a newcomer starts with 1-10 files, gets better, start AAA'ing them easily, then go for 11-20, then 21-30, and so on. We can plan achievements based in every multiple of 10, so people can keep playing and improving to achieve these tasks. About changing difficulties based on players' skills, like FFR did and still does, I don't think it's right. It'd be unfair with people that's totally new and unused to the game, that's why we need to come up with a fair system for all, and I really, really don't believe that's it's too hard.

Also, please, don't think I'm overplanning this difficulty/TP system. I think we all agree that making an unique diff system will be easier to compare files, because, to new people, a 5 (Xoon rating) being harder than 12 (old-school rating) seems a little nonsense, and a TP system will atract new people to the game, and give reasons for nowadays players to keep playing Stpemania.

The only thing about using 1-100 that I dislike, regardless of the number of songs that stepmania has, is that it will generate a lot of debate, especially since not everyone is equally good at hitting every type of pattern

"Uh guys... watf starship.6 is a 61? Push that shit down to 59 man the bursts are ez"
"No way dude I think it should be rated higher, the intersparsed color notes are definitely like +2 from where it's supposed to be"
"What, and rate it higher than cities of the future? You're crazee pce"

The arguments will be worthless. When the scale is smaller, it's a lot harder to argue up and down on it, regardless of player ability.

I feel like the 1-100 is just too dynamic of a range, and you might not even end up using every single value. Like you might as well just list the song's raw NPS/Fudge-factor and put the difficulties like that. Tell me a single "1" song we have right now that's even worth playing that isn't like 2 notes long or less than 15 seconds.

You could argue that the more difficult songs that not everyone can play need to be scaled respectively (for the sake of example let's just say like Impulse Beginner is a 20. There are obviously a lot of more difficult songs out there now that you might need to differentiate between) but I guess that will be dependent on whether or not players want a low difficulty cap or a high one.

[sakura080789]

Originally Posted by Oni-Paranoia

@Sakura - A lot of people don't understand what contributes to a rating system. I don't see why it should be voted upon, I think what everyone needs to do is layout the pros and cons so people can discuss without people restating already said facts.

hmmm alright i see.

[Reincarnate]

I don't know what happened here but the 1-100 system is being totally bastardized.

The idea is that you set 1 to the easiest song, 100 to the hardest song, and then weight the hardness of a song as a function of the average performance on that file further weighted by the average skill level of the players that play it.

In other words, every file's difficulty is market-weighted. You can also implement Bayesian weighting to shell out low-N high-variance issues that'd otherwise cause large difficulty-value swings.

The end result is that difficulty is derived from something objective.

[Shikari]

Okay, seems like coming up with an unique system for KB files will be difficult, but I still want to try it, for reasons already stated, like TP system and scores database. And, since lots of people have different opinions, why don't we try to gather all these ideas together and set a rating system to work with? Or we can wait for the SSC dev team to actually build and code this scores database, and then think about this system later.

[Reincarnate]

The problem is that we can look at DDR and suddenly slap "10" to something like Max 300, but once people start dominating it, we have to start adding more and more levels and it becomes more and more outrageously silly ("Yeah this song is totally a 22").

Furthermore, having a shifting difficulty rating system ensures that you'll never run into retarded long-tail problems.

It also removes subjectivity. "Oh, this file should totally be rated higher -- it has trills." "Well, I think it should be rated lower because trills are easy." Market-weighting averages all this crap out. So if MOST people suck at trills, the rating will take it into account so there's no argument.

Having a large rating scale from 1-100 gives you plenty of diversity/data, but you can also subcategorize 1-100 into X bins (e.g. 5, 10, etc) if you really want to shell out the minute variance. But my only problem with this is that having larger-range categories usually give less information, and I think 1-100 scales provide plenty of valuable information (a 67 vs. a 68 might be relatively small, but comparing a compressed 6 vs. 7 might be profound, and we may come to expect the same differential in other 6's versus 7's, which brings us right back to perceived difficulty issues to begin with).

In other words, stop arguing with statistics. >:O Numbers don't lie. The system requires no manual management or subjective ratings. The difficulty system creates itself based on how people are doing on files and who's playing them.

[Reincarnate]

Additional notes:

This is also why a notes-per-second metric is a poor idea. It may be a reasonable heuristic on the whole (as higher-TPS songs TEND to be harder), but it is inferior, as it fails to account for songs of high variance (say an easy song with one insanely hard part, or a song that is fast but has easily-hittable patterns).

TPS is an easy, fine metric to use... but it has obvious shortcomings and it'll still result in disagreements for unusual songs. Difficulty is not limited to mere speed alone. You must take into account every other facet of a song. The music, the sync, the background, the pattern types, the length, etc. Instead of trying to subjectively pick what we think is the most appropriate heuristic, just take them all into account via market-weighting. Easy to do, and more accurate.

[aj187]

I feel compelled to bring up http://monoglider.web.fc2.com/ddrlv/ which basically re-rates DDR songs on a scale from 1-100. That's all.

[who_cares973]

lets just steal kbo's rating system

[fido123]

If you guys need a database/web-programmer guy let me know. I know MySQL, ASP, and PHP and I'm currently learning mod_perl. I don't really wanna make the graphics, or do a whole lot of HTML TBH but I really enjoy doing that kinda stuff and have already made login/account systems and file databases.

Add me at mspalmer12@gmail.com if you want to discuss some of the technical aspect of this site.

[who_cares973]

quick question, what is the site going to be called? stepmaniaunited?

[Kraezymann]

I thought about using an anagram for Stepmania so I looked it up on wordsmith.org/anagram and typed in Stepmania and the first one displayed was

"Seaman Pit"

[Kraezymann]

PS the second one was "Seaman Tip"

http://wordsmith.org/anagram/anagram...nia&t=1000&a=n

[Reincarnate]

Can someone else please acknowledge what I've just said about the difficulty system and express some thoughts?

[Shikari]

Originally Posted by Reincarnate

It also removes subjectivity. "Oh, this file should totally be rated higher -- it has trills." "Well, I think it should be rated lower because trills are easy." Market-weighting averages all this crap out. So if MOST people suck at trills, the rating will take it into account so there's no argument.

In other words, stop arguing with statistics. >:O Numbers don't lie.

But the problem is to come up with these statistics to rate the song properly. Taking in account pass/AA/AAA rate is good, because it's based on real data of player's skills and performance, but it's hard to find it for almost every song used in this system, imo.