View previous topic :: View next topic |
Author |
Message |
pucktheduck
Joined: 01 Mar 2008 Posts: 36 Location: Earth
|
Posted: Thu Jul 24, 2008 3:31 am Post subject: |
|
|
Here's an idea (kind of an iffy one). Custom leagues! Each week, four well-charted songs (or fewer, if people find it too time-consuming) will be selected from the custom songs index and posted along with their download links. This way, people will be playing songs that they've never seen before, making things more interesting and also discouraging squeezing for the most part. Of course, those just looking to increase their scores by entering leagues shouldn't do this one.
Like I said, this one is questionable. But if enough people want to do it, and staff are willing to search the index for good charts, then why not? |
|
Back to top |
|
|
IWillKickU
Joined: 06 Dec 2007 Posts: 2830 Location: In the Undertow
|
Posted: Thu Jul 24, 2008 2:34 pm Post subject: |
|
|
The idea for custom leagues has come up before, but it always gets the same response. No, because
A) Custom charts are too easy to modify (ie adding star power phrases were they don't belong).
B) Not everyone has access to the songs, and we DON'T want to encourage illegal downloading of MP3s, or requiring participants to spend more money when the already have the GH product. Same reason there's no DLC.
C) Customs require modified equipement (for most people), which Scorehero does not condone. Running a league that requires you to have a modded PS2 (that's the platform that most custom players use) would be officially condoning mod chips (something SH tries to stay away from for legal reasons).
D) There are simply WAY too many possibilities for custom songs, it would be increadibly time consuming for a mod or league commissioner to sort through them all to find good ones.
E) A player using GH3 customs would have a huge advantage over a player using GH2 customs because of the forgiving timing window. This puts people who don't have GH3 at a huge disadvantage, when leagues NEED to be fair across the board.
F) Because the custom database is so vast, Player A might not be better than Player B, but a custom was chosen that Player A actually wrote, and Player B has never even heard the song before. Player A is at a huge advantage regardless of skill differences.
If I racked my brain a bit, I sure I could come up with reasons G, H, and I, but Deimos said it best in another thread that I guess has been deleted:
Quote: | Right now, custom leagues are not happening. |
If only I was blessed with Deimos's brevity. |
|
Back to top |
|
|
Heh333
Joined: 27 Sep 2007 Posts: 6 Location: Finland
|
Posted: Thu Jul 24, 2008 10:22 pm Post subject: |
|
|
FC or 100% league, little different from the others.
* 4 songs on every week
* All scores submitted must be FC or 100% (rather FC)
* If you can't FC all the songs in that week, you're out
* Songs get harder every week: First week songs easy like mississippi queen, when you were young etc.
and after that a little harder every week
The idea just came to my head, so I thought i'd post it. |
|
Back to top |
|
|
pucktheduck
Joined: 01 Mar 2008 Posts: 36 Location: Earth
|
Posted: Fri Jul 25, 2008 12:31 am Post subject: |
|
|
Oh... That makes sense, though. I had a feeling custom leagues wouldn't work, I just didn't know how.
heh333's idea sounds good. Except it might be a little hard for the novice players. I think I could hold my own on it, though. Also, top players who enter in this league might not get eliminated for a very long time, making this one considerably longer (or shorter, if only noobs sign up) than traditional leagues. This might screw up preparations for the next season. |
|
Back to top |
|
|
Rocker0913
Joined: 10 Jul 2007 Posts: 607 Location: Moncton, NB, Canada. BANDS WANTED HERE!
|
Posted: Fri Jul 25, 2008 8:20 pm Post subject: |
|
|
Heh333 wrote: | FC or 100% league, little different from the others.
* 4 songs on every week
* All scores submitted must be FC or 100% (rather FC)
* If you can't FC all the songs in that week, you're out
* Songs get harder every week: First week songs easy like mississippi queen, when you were young etc.
and after that a little harder every week
The idea just came to my head, so I thought i'd post it. |
I can see this as a very good idea for a rivalry, but not for a league. The reason? Like pucktheduck said, it would either last too short or too long. I see leagues trying to make things making things fair, While clearly, this would only favor the very top participants, while screwing the ones here for fun. What happens if youre out the first week? Youre out and its over that quick. I like the current format of four weeks of songs then playoffs, that way, everyone is guaranteed at least 4 weeks of competition, although your idea is an intriguing one.
Another flaw would be that it would be very difficult to do divisions to associate yourself with people of your own skill. In the current type of league, its easy because there is a qualifying week that makes it easy to put the member into a group associating to his skill level. In this type, however, we can only seperate by those who are eliminated from the leagues each week.
It would be motivating people to FC the songs thought, which cant hurt, but as I said again: In my view, great idea for a rivalry, not so much for a league. _________________
Music is life. Live a fun one. Listen to ska.
Listening to: Dinosaur Jr., Planet Smashers, Streetlight Manifesto, Less Than Jake, Beatles
No banners. Why? Meh. |
|
Back to top |
|
|
Raikri999
Joined: 12 Jan 2008 Posts: 1573
|
Posted: Sat Jul 26, 2008 3:14 am Post subject: |
|
|
Rocker0913 wrote: | Heh333 wrote: | FC or 100% league, little different from the others.
* 4 songs on every week
* All scores submitted must be FC or 100% (rather FC)
* If you can't FC all the songs in that week, you're out
* Songs get harder every week: First week songs easy like mississippi queen, when you were young etc.
and after that a little harder every week
The idea just came to my head, so I thought i'd post it. |
I can see this as a very good idea for a rivalry, but not for a league. The reason? Like pucktheduck said, it would either last too short or too long. I see leagues trying to make things making things fair, While clearly, this would only favor the very top participants, while screwing the ones here for fun. What happens if youre out the first week? Youre out and its over that quick. I like the current format of four weeks of songs then playoffs, that way, everyone is guaranteed at least 4 weeks of competition, although your idea is an intriguing one.
Another flaw would be that it would be very difficult to do divisions to associate yourself with people of your own skill. In the current type of league, its easy because there is a qualifying week that makes it easy to put the member into a group associating to his skill level. In this type, however, we can only seperate by those who are eliminated from the leagues each week.
It would be motivating people to FC the songs thought, which cant hurt, but as I said again: In my view, great idea for a rivalry, not so much for a league. | You also forget that you'd be required to show pictures(Sorta) because I could say I fc'd a song but I didn't really which would exclude some people _________________
|
|
Back to top |
|
|
Heh333
Joined: 27 Sep 2007 Posts: 6 Location: Finland
|
Posted: Sat Jul 26, 2008 6:15 pm Post subject: |
|
|
I thought that the qualifying would be maybe 3 songs, for example slow ride, talk dirty to me and RARAN. If you FC all three, you're in AAA, 2 FC's to AA and one song FC takes you to the A-league, but then again theres the problem about pics |
|
Back to top |
|
|
petererk
Joined: 30 Jun 2007 Posts: 884 Location: That place
|
Posted: Mon Jul 28, 2008 10:40 pm Post subject: |
|
|
I think I'm actually going to put more than the first week's songs up this
time, and from what I saw GH3 was running pretty smooth. So I have
nothing to suggest. Keep up teh funz. _________________
00:57 pkovach2 : Whats a "penis"?
00:57 jnukes2-1 : i'll show you tomorrow
Ewwww....
My Customs |
|
Back to top |
|
|
sonicboom139
Joined: 03 May 2007 Posts: 285 Location: St. Louis, MO
|
Posted: Tue Jul 29, 2008 12:11 am Post subject: |
|
|
With the whole points chart system, I can see the complaints for losing 10 points for one tick. On the other hand, this chart works well for difficult songs such as Raining Blood and The Way it Ends. I have two ideas for ways to solve this.
1) Make a points system based off spread rather than place. This could involve something such as giving the top person 100, the bottom person 0, and putting the rest on a linear fit.
Strengths: So long as the top and bottom scores are reasonable, rewards good squeezing and hard FCs equally.
Weaknesses: If one person puts in a terrible score, possibly for doing the song only once, it screws the system over (basing points of median could solve?), linear fit is not always the best.
2) Using different point scales for different songs. e.g., if under 33% FCed the song, use current scale, if 33% to 67% FCed the song, use a different scale, and if over 67% FCed the song, use a third scale. Whatever scales are appropriate remains to be seen.
Strengths: Uses a variety of approaches to tackle different problems
Weaknesses: Depends on the scales, also could be difficult to set up
Amirite? _________________
|
|
Back to top |
|
|
Ieattables
Joined: 31 Jul 2008 Posts: 56
|
Posted: Thu Aug 14, 2008 1:00 am Post subject: |
|
|
I've been experimenting with different formulae for the score system taking current scores into account, and I've made a fomula which I think gives a fairer points spread. On close songs (I believe I used the S9 AAA1 league scores for MHL), it gives very close the the current amount of points - a distribution which is quite fair to everyone really. But on songs with more variation, it awards players who might have scores very close together with more points and players significantly behind them with less. The example I used was from S9 A3 (I think) scores for Monsters.
The main problem with the formula is that it's a bit of a beast, to the point where this is going to go over most people's heads until they read it a few times.
Basically, you need the following calculations:
Standard Deviation of scores.
Average of Score (call it A)
With the Standard Deviation (Call it sd), do the following calculation: 1-sd/[number of scores submitted]. You should get a value between 0.5-1, very close to 1 for close songs and around 0.6-0.8 for more varied songs (I had 0.970 and 0.712). Call this value D.
There are also:
x, the normalised rank
S, the users score
H, the high score (1st places score).
Now, the formula:
= (62.5x + 7.5x^4 + 30x^7) * D + Max ((H - (S - A)) / A, 0) * (1 - D) * 100
The easiest way of understanding it will probably be to take a few leagues, apply to formula to them and see what happens to the scores. Also, I would recommend that users who only submit a completion % should be ignored in calculating the average and standard deviation, but their scores should be treated as 0 when applying the formula. This has the added benefit of giving a significant advantage to players who complete the song (in my second set of data, if the player in last had failed the song he would have gotten 5 less points).
The only problem with the formula (asides from being even more complex ) is that it awards slightly more points in general. Not many, 0-3 on close songs and about 2-10 more in less close songs. But it's a general increase in points across the board especially around the bottom. I'm sure that can be fixed though by tweaking the initial formula (the (62.5x + 7.5x^4 + 30x^7) bit).
Edit: Okay, here's some examples, Sonicboom. This is the data I mentioned above. I had a bit of trouble getting it to align correctly, but it should be okay. All values are rounded to 3 d.p, as well.
Code: | Data set 1 (IIRC, MHL from AAA1)
Score Position Normal New Difference
193994 1 100 100 0
193962 0.947 85.799 86.226 0.427
193754 0.868 69.717 70.624 0.908
193754 0.868 69.717 70.624 0.908
193736 0.789 57.990 59.250 1.260
193546 0.737 51.801 53.245 1.444
193338 0.684 46.513 48.112 1.600
193278 0.632 41.870 43.608 1.738
193230 0.579 37.681 39.545 1.864
192904 0.526 33.806 35.781 1.975
192813 0.474 30.143 32.227 2.084
192630 0.421 26.622 28.809 2.187
191994 0.368 23.192 25.472 2.280
190057 0.316 19.821 22.172 2.351
189544 0.263 16.486 18.929 2.443
188772 0.211 13.173 15.704 2.531
185745 0.158 9.873 12.456 2.582
180684 0.105 6.580 9.181 2.601
170039 0.053 3.290 5.822 2.532
H = 193994
D = 0.970
A = 190409.158
Data Set 2 (IIRC, Monsters from A2/3)
Score Position Normal New Difference
345000 1 100 100 0
340004 0.923 80.269 85.371 5.102
305679 0.846 66.046 71.212 5.166
255253 0.769 55.484 57.763 2.280
245223 0.692 47.279 50.745 3.466
240890 0.615 40.540 45.439 4.899
240835 0.538 34.678 41.261 6.583
224903 0.462 29.320 35.574 6.254
219505 0.385 24.240 31.324 7.084
215580 0.308 19.306 27.351 8.045
211957 0.231 14.445 23.466 9.021
200131 0.154 9.620 18.641 9.021
141783 0.077 4.808 8.353 3.545
H = 345000
D = 0.712
A = 245134.077
|
I also forgot to mention what the Max ( ) bit is for: If someone does particularly badly; or fails the song but still submits it, this ensures they don't get a negative from the average - which would end up giving them possibly something like -40 points instead of 2 or 3.
Last edited by Ieattables on Thu Aug 14, 2008 2:36 pm; edited 1 time in total |
|
Back to top |
|
|
sonicboom139
Joined: 03 May 2007 Posts: 285 Location: St. Louis, MO
|
Posted: Thu Aug 14, 2008 3:02 am Post subject: |
|
|
That idea definitely deserves a test run in S10, assuming it can be done without killing Deimos. This formula could be the new standard in leagues and I am strongly in favor of it. I think it would be easier to understand, however, if you used an example of how the formula works to give readers a clearer idea of what you're talking about (even I'm not totally sure, but the general idea and variables taken into account sound great.) _________________
|
|
Back to top |
|
|
googleimage
Joined: 23 Aug 2007 Posts: 1229
|
Posted: Sat Aug 16, 2008 8:13 pm Post subject: |
|
|
Do you have a dataset of two songs such that the number of submitted scores are exactly the same? Even if you could throw in a few hypothetical scores in the mix that would be really helpful. It looks like the gist of the idea is that the spread of the submitted scores would have an effect on the overall point allotments. I would expect that the majority of the score distributions would be left-skewed (i.e. the average score would be much lower than the median, as the top half is bunched up), and that this would be more pronounced in the squeeze-fests. I'm interested in seeing what effect it actually has on the scores. _________________
Hey, Lindsay Lohan - "drink Canada Dry" is a slogan, not a dare! |
|
Back to top |
|
|
Ieattables
Joined: 31 Jul 2008 Posts: 56
|
Posted: Sat Aug 16, 2008 11:46 pm Post subject: |
|
|
Okay, I've been experimenting a little more, and have come up with the following: f(x)=30x+42.5x^2+10x^4+17.5x^7.
Compared to the old formula, a few more points are given around the top (actually, just under a point). Around the x=0.8 mark the points are equal, but then this formula starts giving less points with somewhere around x=0.38 being the biggest gap of a little under 6.2 points. Then the gap begins to close up, with obviously x=0 being identical for both.
Now, when you combine it into my formula above, things become a little more interesting:
= (30x + 42.5x^2 + 10x^4 + 17.5x^7) * D + Max ((H - (S - A)) / A, 0) * (1 - D) * 100.
Having looked through my test data, this does seem a much fairer formula. People very close to the top in less close songs will get more points - but still quite a bit less than the leader. In very close songs, there's not a lot of change, although people lingering around 2/3s of the way through the pack lose a couple of points.
I'll get a few more examples using this formula sometime tommorow, possibly.
Edit: Not enough hours in the day to do that, unfortunately. However, if the Planning Committee like the idea I'm sure they have the resources to test the formula themselves and see what they think of the results since I'm going to be away for a while now.
Also, for the non-maths nerds who have read this formula, seen the examples and are still thinking 'what on earth is this formula about?', here's a brief description:
Your points are still based mainly on rank
Some points will be awarded based on your score.
The more varied the scores submitted, the more points that will be awarded for score. Conversely, in very close songs score will have a minimal effect, making placement more important.
First place still gets 100 points and everyone else gets less. |
|
Back to top |
|
|
Banditman31
Joined: 18 Aug 2008 Posts: 115 Location: Parker, Colorado
|
Posted: Mon Aug 25, 2008 2:28 am Post subject: |
|
|
i would suggest for a twist,
a guitar duet,
1 person is strumming, and ur pushing the buttons
Ps. if this is a bad idea, dont bash me, just giving a suggestion _________________
|
|
Back to top |
|
|
SeanDread
Joined: 27 Feb 2007 Posts: 30
|
Posted: Mon Sep 01, 2008 6:07 am Post subject: |
|
|
How about a Hendrix League? Participants would play each song in both lefty flip as well as the standard right-handed position, and then average their best scores in each playing style to determine their score on the songs for the week. |
|
Back to top |
|
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
Copyright © 2006-2024 ScoreHero, LLC
|
Powered by phpBB
|