PainState
Joined: Apr 04, 2007
|
 Posted:
Oct 29, 2014 - 20:24 |
 
|
I did and I stand by my thesis. |
_________________
Comish of the:  |
|
Dunenzed
Joined: Oct 28, 2011
|
| Posted:
Oct 29, 2014 - 20:26 |
|
|
Christer
Joined: Aug 02, 2003
|
| Garion wrote: | | If there are 4 teams on game finder 3 are cpomb and 1 is elf the elf team is still playing against that cpomb teams even though the algorithm has made it more unlikely to get drawn. |
In that case, yes. However, the reality doesn't match this example. You have four coaches, and roughly 2 teams per coach on average. Having a bias in the scheduler does have an effect on the matches even if it's small.
The mirror match bias was set at 3% reduced chance of a mirror match, which had a net effect of 3% too few mirror matches compared to expected. A slight bias towards pushing over-represented races to each-other would essentially do just that on average. It would directly affect the matches that were scheduled.
People who joined with a chaos and a dark elf team would be more likely to play chaos vs nurgle than chaos vs skaven - and the scheduler might find more suitable matches with his dark elf team instead.
Essentially, it would directly affect the number of games played by the more popular races, and at the same time make the environment "feel" more diverse by those who aren't playing with them.
It would really have an effect. |
|
|
PainState
Joined: Apr 04, 2007
|
| Posted:
Oct 29, 2014 - 20:38 |
|
| Dunenzed wrote: | | Then your thesis is unsupported by the data. |
I think you are reading the data to support your thesis that Iam a numbskull is wrong.
  |
_________________
Comish of the: |
|
JackassRampant
Joined: Feb 26, 2011
|
| Posted:
Oct 29, 2014 - 20:39 |
|
Kinda likin' that combined with the suitability penalty for low-suitability matches. Would solve most of the existing Box gripes, I betcha. |
_________________
Lude enixe, obliviscatur timor. |
|
huff
Joined: Dec 19, 2009
|
| Posted:
Oct 29, 2014 - 20:46 |
|
| harvestmouse wrote: | | On topic though, how is TV a better match maker than recent WDL record? I see no reason why it would be, and blackbox confirms to me my opinion. Most naysayers are those with great records. |
Well going from a TV based MM to a recent record would just change the dynamic, for better or worse I couldn't say, but I can say I like the sound of playing relativly close TV games (outside of stunty teams) compared to to say a 130TV team on a 5 game win streak vs a 200 TV team on a 5 game winstreak. And with no ceiling for the scheduler that matchup is possible if both teams have played 30 or more games. |
|
|
xnoelx
Joined: Jun 05, 2012
|
| Posted:
Oct 29, 2014 - 20:47 |
|
| PainState wrote: | | Dunenzed wrote: | | Then your thesis is unsupported by the data. |
I think you are reading the data to support your thesis that Iam a numbskull is wrong.
|
Eh? I think you just said exactly the opposite of what you meant, proving that thesis right... |
_________________
 Nerf Ball 2014 |
|
Garion
Joined: Aug 19, 2009
|
| Posted:
Oct 29, 2014 - 20:51 |
|
| Christer wrote: | | Garion wrote: | | If there are 4 teams on game finder 3 are cpomb and 1 is elf the elf team is still playing against that cpomb teams even though the algorithm has made it more unlikely to get drawn. |
In that case, yes. However, the reality doesn't match this example. You have four coaches, and roughly 2 teams per coach on average. Having a bias in the scheduler does have an effect on the matches even if it's small.
The mirror match bias was set at 3% reduced chance of a mirror match, which had a net effect of 3% too few mirror matches compared to expected. A slight bias towards pushing over-represented races to each-other would essentially do just that on average. It would directly affect the matches that were scheduled.
People who joined with a chaos and a dark elf team would be more likely to play chaos vs nurgle than chaos vs skaven - and the scheduler might find more suitable matches with his dark elf team instead.
Essentially, it would directly affect the number of games played by the more popular races, and at the same time make the environment "feel" more diverse by those who aren't playing with them.
It would really have an effect. |
Interesting, you are no doubt correct as I don't have the data you have. My experiences are usually different though, but that is because I play most of my games between 9am and 15:00 Tuesday to Thursday which are not peak times. So the number of teams is usually very low. I would like to see how the change did effect things though, maybe I am just being pessimistic in that I feel without coaches changing their minds set and trying a wider variety of races the top end of black box will remain cpomb heavy. Though I'd be happy to be proven wrong  |
_________________
 |
|
PainState
Joined: Apr 04, 2007
|
| Posted:
Oct 29, 2014 - 21:05 |
|
| xnoelx wrote: | | PainState wrote: | | Dunenzed wrote: | | Then your thesis is unsupported by the data. |
I think you are reading the data to support your thesis that Iam a numbskull is wrong.
|
Eh? I think you just said exactly the opposite of what you meant, proving that thesis right... |
We are both numbskulls, so we are both right....now do I have to shift my steely gaze upon you sir? Throw out some written wit that will reduce you to ashes? |
_________________
Comish of the: |
|
xnoelx
Joined: Jun 05, 2012
|
| Posted:
Oct 29, 2014 - 21:08 |
|
| PainState wrote: | | xnoelx wrote: | | PainState wrote: | | Dunenzed wrote: | | Then your thesis is unsupported by the data. |
I think you are reading the data to support your thesis that Iam a numbskull is wrong.
|
Eh? I think you just said exactly the opposite of what you meant, proving that thesis right... |
We are both numbskulls, so we are both right....now do I have to shift my steely gaze upon you sir? Throw out some written wit that will reduce you to ashes? |
Well, there's a first time for everything, I guess. |
_________________
Nerf Ball 2014 |
|
koadah
Joined: Mar 30, 2005
|
| Posted:
Oct 29, 2014 - 21:11 |
|
| Garion wrote: | | Christer wrote: | | Garion wrote: | | If there are 4 teams on game finder 3 are cpomb and 1 is elf the elf team is still playing against that cpomb teams even though the algorithm has made it more unlikely to get drawn. |
In that case, yes. However, the reality doesn't match this example. You have four coaches, and roughly 2 teams per coach on average. Having a bias in the scheduler does have an effect on the matches even if it's small.
The mirror match bias was set at 3% reduced chance of a mirror match, which had a net effect of 3% too few mirror matches compared to expected. A slight bias towards pushing over-represented races to each-other would essentially do just that on average. It would directly affect the matches that were scheduled.
People who joined with a chaos and a dark elf team would be more likely to play chaos vs nurgle than chaos vs skaven - and the scheduler might find more suitable matches with his dark elf team instead.
Essentially, it would directly affect the number of games played by the more popular races, and at the same time make the environment "feel" more diverse by those who aren't playing with them.
It would really have an effect. |
Interesting, you are no doubt correct as I don't have the data you have. My experiences are usually different though, but that is because I play most of my games between 9am and 15:00 Tuesday to Thursday which are not peak times. So the number of teams is usually very low. I would like to see how the change did effect things though, maybe I am just being pessimistic in that I feel without coaches changing their minds set and trying a wider variety of races the top end of black box will remain cpomb heavy. Though I'd be happy to be proven wrong |
Of course high TV Box will remain CPOMB heavy. The point is to make it a little more bearable for those who do not wish to go CPOMB. |
_________________
[SL] + Official Stunty teams. Progression KO. Old & new teams welcome. 29th May! |
|
Endzone
Joined: Apr 01, 2008
|
| Posted:
Oct 29, 2014 - 22:35 |
|
There are two points that have been remade over the last few pages which I think are important.
1. Tweaking the scheduler to match clawpomb / big 3 races together slightly more often would have an affect. Christer makes this quite clear above.
2. At higher TV you naturally get more of certain races due to the game rules (this is not just in blackbox). Some races just can't get or stay that high very easily, other races sweet spot earlier so are less attractive to coach at high TV.
The point I want to make is that these two don't contradict each other. The scheduler change would, I think, increase the diversity of races at higher TV, however attrition and 'attractiveness' will continue to influence diversity too.
Also, the argument that the scheduler change wouldn't have an impact because of insufficient numbers activating is flawed in my view. Firstly, examples of where it wouldn't affect the pairings are easy to imagine but then so are examples where it would. We are not saying it would affect every pairing, we wouldn't want it to and it wouldn't need to, to have the desired affect. Secondly, this argument is based on numbers not increasing if the scheduler change was made, this I believe is a false assumption.
I see it working like this:
1. The scheduler change is introduced together with the message 'come and play Box - check out the new scheduler!'
2. Honeymoon period with a small increase in numbers and more coaches trying non-clawpomb races.
3. Non-clawpomb teams find they play clawpomb a bit less than before, this helps a bit with developing TV and affects how coaches perceive the division.
3a. Meanwhile some clawpomb coaches are trying it out for themselves, reducing clawpomb presence further.
4. Non-clawpomb teams begin to thrive a bit better at higher TV, getting enough games against each other to compensate for the tough killer matches.
5. We get a positive feedback loop and things settle down with a slightly different looking diversity graph. We still have a predominance of certain races at the highest TV, but there is more diversity of races than before at a higher TV than before.
6. Coaches feel team building in the box is a viable option with a wider variety of races. (the impossible dream) |
|
|
DukeTyrion
Joined: Feb 18, 2004
|
| Posted:
Oct 29, 2014 - 22:38 |
|
So, are we going to end up with the silly situation where, in this situation;
1800tv Dwarves
1200tv Elf
1800tv Claw
1200tv Claw
The Elves would be matched against the 1800tv Dwarves, rather than the 1200tv Chaos?
I have seen nothing yet about getting rid of the stupid 800tv difference matches. |
|
|
BillBrasky
Joined: Feb 15, 2005
|
| Posted:
Oct 29, 2014 - 22:44 |
|
Please stop responding to this dumb thread :p |
|
|
Mr_Foulscumm
Joined: Mar 05, 2005
|
| Posted:
Oct 29, 2014 - 23:08 |
|
| DukeTyrion wrote: | So, are we going to end up with the silly situation where, in this situation;
1800tv Dwarves
1200tv Elf
1800tv Claw
1200tv Claw
The Elves would be matched against the 1800tv Dwarves, rather than the 1200tv Chaos?
I have seen nothing yet about getting rid of the stupid 800tv difference matches. |
Weren't you the one that suggested a limit at about 700 suitability score? That's something, and surely you saw it! |
_________________
Everybody's favorite coach on FUMBBL |
|
|
|
| |
Blood Bowl 2024 Edit...
Search
