Forums

Poll

Question: Should the Beginnings 2013 point system be revised?

  • Yes
    - 13 (65%)
    No
    - 7 (35%)

Total Members Voted: 20

Pages: [1] 2
Author Topic: Beginnings 2013 point system revision?  (Read 463 times)
Rocketman
*****

Posts: 518
No tears now; only dreams
It's fairly obvious the point system is flawed. As it is, first place gets 100 points and then for each subsequent place, a person gets one less point until 0. But this basically means that if one doesn't do a run, they'll have no chance of winning (e.g. a person who gets 1st in 4 maps will lose to a person who gets 15th in 5). I can't seem to come up with a good solution to a new point system, if a new one is desired at all.

I thought about things like... adding a player's times then dividing by the their number of runs, so the person who has lowest time wins. But then there may be a map that takes a lot less time than the rest so a person speedrunning that map and only that map might win it all. Or scaling it differently so that 1st place gives 10 points, 2nd gives 9..., 10th gives 1, 11 or later gets 0. But this still has the same problem where a person who doesn't run at least once can be severely set back, though not as much as the 100 point scale.

Link to hiscores (at the moment overall isn't updated): http://bit.ly/HPg5TX


---------------------------
   
Fly
**

Posts: 57
Just call me trey
   
Rocketman
*****

Posts: 518
No tears now; only dreams



That seems flawed though as well though....

Imagine one person gets: 1st, 2nd, 1st, 2nd, 3rd. Another gets 3rd, 1st, 2nd, 1st, 2nd. Now, one of these people is bound to lose over the other purely because of how many people submitted demos... It doesn't seem right.


---------------------------
   
Offline  RNC1839
Maggot
*

Posts: 23
Just average the points for all 5 stages.


---------------------------
   
Fly
**

Posts: 57
Just call me trey



That seems flawed though as well though....

Imagine one person gets: 1st, 2nd, 1st, 2nd, 3rd. Another gets 3rd, 1st, 2nd, 1st, 2nd. Now, one of these people is bound to lose over the other purely because of how many people submitted demos... It doesn't seem right.

I don't think there's a perfect system. The weighting based on number of demos submitted, however, takes things like difficulty of the map. It evens the field a bit more for people who don't submit demos every week. The best system would be to determine an "optimal" best case time for each map, then award points based on how far off the mark each person is. You would have to curve it to fit all of the outliers, and clearly this method would involve a good bit of work to standardize an optimal time. You could also fit the curve to fit the number one time (as opposed to a true "flawless" run), but everything has bias
   
Rocketeer
****

Posts: 424
I think we have to look at the number of send on demos, the more the better.
And then just the average place. If two people tie, The one with the hights place on the hardest stage wins.
Something like that. still pretty vague.


---------------------------
   
Rocketman
*****

Posts: 518
No tears now; only dreams
Just average the points for all 5 stages.

Then a person could get 1st in a single map and not run and still be at least tied for 1st place overall

I don't think there's a perfect system. The weighting based on number of demos submitted, however, takes things like difficulty of the map. It evens the field a bit more for people who don't submit demos every week. The best system would be to determine an "optimal" best case time for each map, then award points based on how far off the mark each person is. You would have to curve it to fit all of the outliers, and clearly this method would involve a good bit of work to standardize an optimal time. You could also fit the curve to fit the number one time (as opposed to a true "flawless" run), but everything has bias

The number of demos submitted doesn't reflect how hard a map is. Grimace got 6 (stage 2), zero got 11 (stage 3). Is one harder than the other? Tbh it's hard to say. It seems as though people will submit a run if they like the map more than anything. And then there are people like Alexgaara who submits hour+ long times (no offense to alexgaara), regardless of how hard the map is. And finding the ideal time to base a point system off is just not feasible.

I think for now I'll reduce the point scale to 10 points for first for now. Still not perfect. I think there's an optimal point system... not a perfect one though.


---------------------------
   
Offline  RNC1839
Maggot
*

Posts: 23
Just average the points for all 5 stages.

Then a person could get 1st in a single map and not run and still be at least tied for 1st place overall


That's why I said ALL 5 stages. If someone gets 1st on one map and don't submit for the others they get 20 points. If someone gets 2nd on all maps they get 99 points at the end of the comp.


---------------------------
   
Offline  The BQE
Soldier
***

Posts: 135
combination of treynolds idea and later stages should be weighted higher (getting 1st place on stage 5 should be worth way more than 1st place on stage 1).
« Last Edit: November 30, 2013, 09:05:35 PM by The BQE »
   
Rocketman
*****

Posts: 518
No tears now; only dreams
Just average the points for all 5 stages.

Then a person could get 1st in a single map and not run and still be at least tied for 1st place overall


That's why I said ALL 5 stages. If someone gets 1st on one map and don't submit for the others they get 20 points. If someone gets 2nd on all maps they get 99 points at the end of the comp.

That's really no different than looking at the total sum of their scores though... you're just changing things by a factor of 5.

I've thought of an idea:
Use the 100 point scale still. At the end, average the total points per player by however many maps they speedran. And then provide an additional point bonus for each map they did speedrun. So if a player speedruns all 5, gets 2nd on all 5, they'd get (99*5)/5 + 5 = 104 points. If a player does 4 maps and gets 1st on all 4, they'd get (100*4)/4 + 4 = 104 (which would tie with the person who got 2nd on all 5). Someone who gets 5th on all 5 will get (96*5)/5 + 5 = 101. A person who gets 1st, 2nd, 3rd, 4th, 5th will get (100+99+98+97+96)/5 + 5 = 103. To me this feels like a pretty solid ranking system.

The 1 point bonus could be adjusted if it's found to be not optimal (maybe a 2 or more point bonus instead of just 1?). Perhaps a greater weight on the bonus point for the later stages? Tell me what you guys think.
« Last Edit: November 30, 2013, 09:36:04 PM by Afterglow »


---------------------------
   
Offline  RNC1839
Maggot
*

Posts: 23
That's really no different than looking at the total sum of their scores though... you're just changing things by a factor of 5.

Yeah I realized that a few minutes after posting...



Anyway I have more ideas.

Adding on to the weighting it idea. The total number of submissions is how many points #1 gets then decrement from there down to 1 point. Then multiply the score by the stage number minus one. So the math would look something like this:

Stage 1. Fuck stage 1.
Stage 2. 20 submissions, 20 possible points. Multiply by 1. #1 gets 20 points. #2 gets 19 points...
Stage 3. 15 submissions, 15 possible points. Multiply by 2. #1 gets 30 points. #2 gets 28 points...

Stage x. n submissions, n possible points. Multiply by x-1. #1 gets n*(x-1) points. #2 gets (n-1)*(x-1) points...


With a constantly diminishing number of participants this would work fine. Also because the best are always submitting there's no issue of someone stealing first.



OR we could do it golf style and if someone doesn't submit they forfeit to last place +1 or any other number really. So 5 would be the best score possible (quba). I like this one the most actually.


---------------------------
   
Rocketman
*****

Posts: 518
No tears now; only dreams
With a constantly diminishing number of participants this would work fine. Also because the best are always submitting there's no issue of someone stealing first.

OR we could do it golf style and if someone doesn't submit they forfeit to last place +1 or any other number really. So 5 would be the best score possible (quba). I like this one the most actually.

The thing is that it's not constantly diminishing... it's a bit unpredictable... from 6 to 11 for demo? I don't know man, I'm really skeptical about factoring in the number of submissions. Merely being able to complete a map hardly means anything anyways imo. Anyone can submit a multi-hour time.

I also don't believe in heavily weighting the later stages. It can be just as hard to speedrun an easy map as a hard map. Perhaps some weight to the later ones, but not a huge amount.

Not too sure about what you mean about golf score thing. Anyways the thing I want to change with a point system revision is to enable someone who misses a stage but still does really well (like top 3 in the rest) will still have a chance of winning. From the sounds of it, the golf thing isn't fixing that (as you're placing someone who misses a stage into last place???) As it is, once you miss a stage, you pretty much lose all chance of placing highly, which is I don't think is good.


---------------------------
   
Fly
**

Posts: 57
Just call me trey
The number of demos submitted doesn't reflect how hard a map is. Grimace got 6 (stage 2), zero got 11 (stage 3). Is one harder than the other? Tbh it's hard to say. It seems as though people will submit a run if they like the map more than anything. And then there are people like Alexgaara who submits hour+ long times (no offense to alexgaara), regardless of how hard the map is. And finding the ideal time to base a point system off is just not feasible.

I think for now I'll reduce the point scale to 10 points for first for now. Still not perfect. I think there's an optimal point system... not a perfect one though.
More demos were submitted this week on account of time off of school from thanksgiving, and more collective time for running
   
Offline  RNC1839
Maggot
*

Posts: 23
The thing is that it's not constantly diminishing... it's a bit unpredictable... from 6 to 11 for demo? I don't know man, I'm really skeptical about factoring in the number of submissions. Merely being able to complete a map hardly means anything anyways imo. Anyone can submit a multi-hour time.

Fair enough but I still consider finishing a map to be something, but not groundbreaking by any means. Still deserving of points, but not 80+ for a several hour run.

I also don't believe in heavily weighting the later stages. It can be just as hard to speedrun an easy map as a hard map. Perhaps some weight to the later ones, but not a huge amount.

Weighting the maps would be a lot easier if we could agree on a number that defines how difficult a map is. BUT that's entirely impossible because it's trying to objectify subjective data.

Not too sure about what you mean about golf score thing. Anyways the thing I want to change with a point system revision is to enable someone who misses a stage but still does really well (like top 3 in the rest) will still have a chance of winning. From the sounds of it, the golf thing isn't fixing that (as you're placing someone who misses a stage into last place???) As it is, once you miss a stage, you pretty much lose all chance of placing highly, which is I don't think is good.

Personally I think people should be penalized if they don't submit and can definitely finish the map. This feeling mostly comes from my one week rant but that's just me. To make it fair for a person who does well but can't finish stage # for whatever reason is the difficult bit. And there really isn't a way to do that without either ruining the not as good people's scores or massively inflating their scores.

The golf score would basically make it so a person's place is their score. So to fix the golf idea I think averaging a person's total golf score by their number of runs then the closest to 1 at the end wins. BUT this is the same as not using golf scores and still averaging with their total number. Both of these lead to a chance at stealing first (or whatever). So that wouldn't work very well either.


With the bonus point thing I think this is a good idea the only issue is how big the bonus' will be... I think basing the stage bonus off some logarithm or exponent of something. Or asking everyone what the bonus should be based on difficulty, average the answers, round down, and that's the bonus.


Fortunately I have an even better idea.

Fuck points. Give cookies.


Seriously though. I had an idea that seems really stupid... Anyway create a voting system. Make it so people would see the times and places for everyone (google docs), then rank everyone by how well the voter thinks the others did overall. So voters would choose overall 1st, 2nd, 3rd, etc. I think this is stupid, but could work, I highly doubt it though. Also this would just turn into a popularity contest very quickly.


---------------------------
   
Offline  QuBA
Soldier
***

Posts: 134
I've thought of an idea:
Use the 100 point scale still. At the end, average the total points per player by however many maps they speedran. And then provide an additional point bonus for each map they did speedrun.

That is really bad as well, cause then sending in a bad time could potentially give you less points than not sending it in, that shouldnt happen.

If you really wanna change it just do something like 1000*0.75^(your place). That way the point difference between 1st and 5th etc. is big enough to potentially make up for a missed map.

To be fair, the system works fine if everyeone submits 5 demos, that was pretty obvious from the start. So everyone who didnt submit demos just doesnt seem to care enough.
   
0 Members and 1 Guest are viewing this topic.
Pages: [1] 2
« previous next »


spirit Powered by SMF 2.0 RC5 | SMF © 2006–2011, Simple Machines LLC