smallman
Joined: Sep 24, 2007
|
 Posted:
Jul 10, 2019 - 01:58 |
 
|
https://fumbbl.com/p/match?id=4075943
Another silly +1mil TV gap, which I don't mind so much now that there are better inducements to buy. But its silly to pair 2310 dwarves with my 1260 UW, when I also activated 1410 Renegades. All teams have played lots of games, and neither team has played this dwarf team before, so definite glitch.
Good time to ovehaul the box. The offpeak window keeps growing bigger, and will inevitably grow to 24h if nothing is done. Hourly activation in offpeak is the best answer. |
|
|
razmus
Joined: Jun 23, 2017
|
| Posted:
Jul 10, 2019 - 02:52 |
|
|
ramchop
Joined: Oct 12, 2013
|
|
fidius
Joined: Jun 17, 2011
|
| Posted:
Jul 10, 2019 - 04:07 |
|
THE SCHEDULER WORKS THIS WAY BECAUSE *YOU* WERE ABUSING IT |
|
|
Nelphine
Joined: Apr 01, 2011
|
| Posted:
Jul 10, 2019 - 05:06 |
|
Note: I'm actually trying to take the time to understand the box scheduler, as my underworld have been getting somewhat discouraging recently. This particular example is much worse than any I've seen, but it rang very close to my recent experiences. This is not meant to be an attack on anyone, but rather, an attempt to gain understanding, and where I've seen a possible flaw, hopefully to get some feedback to explain why this is.
I actually didn't understand that match up. It LOOKS like the scheduler only considered two options: Either underworld vs dorf and elven union vs chaos dorf, OR chaos renegade vs high elf. Presumably the high elf is being played by the elven union player or the chaos renegade player.
Why didn't it consider chaos renegade vs dorf and elven union vs chaos dorf?
Below I'll explain why it didn't consider that option, and then explain how i think it could be tweaked to have considered it.
(For purposes of this discussion, I'm going to arbitrarily put the high elf team on the same coach as the elven union team, but I don't think it actually matters if it was on the same coach as the chaos dorfs. It's POSSIBLE it would have mattered for the coach choice based on highest sum, but I believe A still wins that regardless of whether B or C has the high elves.)
Note: the chaos renegade, underworld, and dorf teams have no TV limits
the elven union team can have a 30% difference, the chaos dorf can have 32%, and the high elf team can have 42%.
chaos renegade (A) vs chaos dorf (C) - allowed, presumably highish suitability
chaos renegade (A) vs dorf (D) - allowed, presumably lower suitability due to TV but I dont know how suitability is actually calculated for the racial matching.
underworld (A) vs chaos dorf (C) - allowed, presumably low suitability for racial
underworld (A) vs dorf (D) - allowed, presumably very low due to TV and racial
high elf (B) vs chaos renegade (A) - allowed, presumably good suitability
high elf (B) vs underworld (A) - allowed, presumably not low suitability
high elf (B) vs chaos dorf (C) - allowed, presumably not low suitability
high elf (B) vs dorf (D) - not allowed due to TV
elven union (B) vs chaos renegade (A) - allowed, presumably good suitability
elven union (B) vs underworld (A) - allowed, presumably not low suitability
elven union (B) vs chaos dorf (C) - allowed, presumably not low suitability
elven union (B) vs dorf (D) - not allowed due to TV
chaos dorf (C) vs dorf (D) - not allowed due to TV
First method:
If this picks highest suitability first, then presumably A plays B or A plays C depending on exact racial suitabilities. After this, D has no available options.
If this picks lowest suitability first, then presumably A (underworld) plays D, then B plays C.
If random, then something, but if A doesn't play D, then D has no available options.
Then it picks by coach.
If it picks highest suitability, then either A, B or C is picked (depending on exact racial suitabilities), and A ends up matched in a game. D gets no game.
If it picks lowest suitability, then A or D gets picked; if A picked, then A plays B or C, and D gets no game. If D gets picked, then dorfs play chaos renegades, and B plays C.
If lowest sum, then C gets picked, plays A, B and D get no game.
If highest sum, then B gets picked, plays A, C and D get no game.
Least and most number of opponents have similar results to sum.
Random is random.
based on this, we can see that in the first group, it picked lowest suitability, which ends up with underworld v dorfs, and that leaves B vs C.
In the second group, the result was presumably highest suitability, which was chaos renegades vs high elves. This was a very high suitability match but results in no game for D or the other player, so the total suitability of chosen matches is lower than in the first group.
I disagree with this. Why is lowest suitability an option for the team matchups?
(This may explain why my underworld have faced dwarves and chaos dwarves in 3 of the last 4 matchings I've found a match in, although one of those games got cancelled by an admin) Starting the entire list with the worst possible match up seems like a bad method, unless it was used relatively rarely. More specifically, if one coach randomly happens to be involved in a team matchup based on this pick several times for this option (and is playing a team with particularly bad racial matchups such as stunty teams when a dorf is also in the match up), that coach will probably be the lowest suitability several times, and that ends up feeling really bad for that coach and discourages them from playing that team in the box again. Note that this will happen to this coach even if they are activating many teams (and therefore presumably not trying to abuse the box) because stunty/dorf teams presumably have the worst racial suitability modifier, so that same stunty team will keep getting picked each time the low suitability team matchup is chosen by the scheduler.
In the case of choosing by coach, lowest suitability still makes some sense; that particular match up isn't likely to be chosen, so by choosing that coach first, you can then immediately take the highest suitability match available to that coach. In this example, if the low suitability option had been used, then chaos renegades plays dorfs, and B plays C.
At the very least a follow-up step, between 3 and 4, which says 're-examine each coach paired in step 3. select highest suitability between those coaches.' would avoid this kind of thing. Step 3 would then select the highest number of coach matches (or close to it) that have a reasonably high suitability; and then step 3.5 would refine that to make sure if any better match ups exist between the coaches that have already been paired (and these matchups are already calculated in step 2), those team match ups would get chosen instead. |
|
|
ArrestedDevelopment
Joined: Sep 14, 2015
|
| Posted:
Jul 10, 2019 - 05:24 |
|
Because the object of the box scheduler is to create matches, not to create fair ones.
There's also a negative malus applied to both the Pelves & Chorfs facing the 16+ game teams (not a very strong one for the PElves, but it's there). And the HE & PE are absolutely the same coach.
I've said it a million times, I'll say it again: The NA timezone is the exact settings you'd feed the box if you wanted it to fail.
1. A low active population
2. Members of that population who have very old teams and only activate those old teams
3. Members of the population with differing approaches to the game in regards to teambuilding.
4. A not-unreasonable percentage of the low population is newish coaches or coaches playing newer teams.
Ergo, in every single "successful" draw there is usually at least one high tv monster, a low tv team of varying age, and then the last two vary - sometimes there's another high tv team, some times there's two mid-tv teams, and sometimes there's two low tv teams.
In this situation, someone is always going to get fed to the high tv team. The only way to avoid it is not to activate or to activate a new team, or one that is within the %tv range protection to as to suitably be unable to draw that high tv team (assuming you know what it's going to be).
Occasionally, said high tv team will be the biggest mismatch, and won't get drawn - but at that point there's a combination of points 2&3&4 that may lead to a very unsatisfactory game for someone unfortunate enough to be drawn against a team built with the mindset of staying smaller. |
_________________
 |
|
Nelphine
Joined: Apr 01, 2011
|
| Posted:
Jul 10, 2019 - 05:29 |
|
Edit: I believe this IS too complex, as the piece in step 3 would need to calculate lots of match ups; for this example it needs to calculate 3 (=3*1) sets of match ups; but if there were 6 people, then it would calculate 15 matchups (=5*3*1) and if there were 8, then it would be 105 (=7*5*3*1) matchups, which is increasing fast. So feel free to ignore the rest of this post, although, i wonder how far it could go (as in, could it do 10 people fast enough for Christer's needs? what about 8? etc) and if there was a point where it was ok, then only use this method when there were that few people or less.
Another question:
We've already calculated all possible match-ups in step 2.
Add a step 2.5.
A) List all coach Pairs.
B) Identify highest suitability for all coach pairs.
Is this a step that isn't reasonable to do given time constraints? I know calculating the highest total suitability out of all possible allowed team pairs isn't reasonable, but I'm not sure if this lesser version is reasonable.
In this case, it identifies the following pairs: A+B; A+C; A+D; B+C; B+D; C+D
and then identifies the highest suitabilities:
A+B: high elf vs chaos renegade;
A+C: chaos renegade vs chaos dorf
A+D: chaos renegade vs dorf
B+C: high elf vs chaos dorf
B+D: none
C+D: none
Then in step 3, there would be a 3rd set of criteria: Matches based on highest combination of highest suitabilities.
In this case we would get chaos renegade vs dorf and high elf vs chaos dorf.
(Recalling my assumption about which coach high elf was on, and since elven union vs chaos dorf was based on lowest suitability then high elf vs chaos dorf would be higher suitability)
However, I'm worried that due to the fact that there are a bunch of 'no valid options' I'm actually underestimating the time required and this may actually be too complex. |
Last edited by Nelphine on Jul 10, 2019; edited 2 times in total |
|
Nelphine
Joined: Apr 01, 2011
|
| Posted:
Jul 10, 2019 - 05:32 |
|
| ArrestedDevelopment wrote: | Because the object of the box scheduler is to create matches, not to create fair ones.
There's also a negative malus applied to both the Pelves & Chorfs facing the 16+ game teams (not a very strong one for the PElves, but it's there). And the HE & PE are absolutely the same coach.
I've said it a million times, I'll say it again: The NA timezone is the exact settings you'd feed the box if you wanted it to fail.
1. A low active population
2. Members of that population who have very old teams and only activate those old teams
3. Members of the population with differing approaches to the game in regards to teambuilding.
4. A not-unreasonable percentage of the low population is newish coaches or coaches playing newer teams.
Ergo, in every single "successful" draw there is usually at least one high tv monster, a low tv team of varying age, and then the last two vary - sometimes there's another high tv team, some times there's two mid-tv teams, and sometimes there's two low tv teams.
In this situation, someone is always going to get fed to the high tv team. The only way to avoid it is not to activate or to activate a new team, or one that is within the %tv range protection to as to suitably be unable to draw that high tv team (assuming you know what it's going to be).
Occasionally, said high tv team will be the biggest mismatch, and won't get drawn - but at that point there's a combination of points 2&3&4 that may lead to a very unsatisfactory game for someone unfortunate enough to be drawn against a team built with the mindset of staying smaller. |
ArrestedDevelopment, that doesn't respond to my proposed step 3.5. I agree that getting matches is the primary requirement, but my step 3.5 still avoids the absolute worst matches.
Additionally, it also doesn't explain why starting with the absolute worst match is actually a good thing. Starting with the coach with the worst match up, and then picking their best match up, does ensure more matches by not running into that bad matchup - I'm not sure how starting with the worst match up actually encourages more matches. (Yes, I actually want to understand how it does, as I do believe it does or it wouldn't be an option.) |
|
|
morehouse
Joined: Sep 29, 2005
|
| Posted:
Jul 10, 2019 - 05:37 |
|
"Scumbag Min-Max draws entirely fair match up when trying to pick on noobs" Wonderful headline, keep um coming. |
|
|
Antithesisoftime
Joined: Aug 20, 2014
|
| Posted:
Jul 10, 2019 - 06:07 |
|
Looks like a filthy minmax coach trying to game the system got hit by the system working as intended, then comes to the forums to whine about it because the system worked against him this time.
Doesn't really seem like a glitch at all |
|
|
Amon242
Joined: Aug 02, 2003
|
| Posted:
Jul 10, 2019 - 06:09 |
|
Ho Ho Ho - thanks for the chuckle |
|
|
MenonaLoco
Joined: Jan 05, 2016
|
| Posted:
Jul 10, 2019 - 06:23 |
|
Imho the box should have less matchmaking criteria, less "consideration", so more activations will actually end in matches and not activation fails. |
|
|
ArrestedDevelopment
Joined: Sep 14, 2015
|
| Posted:
Jul 10, 2019 - 07:10 |
|
Btw Nelphine, I went back and looked at the two box draws for your UW:
https://fumbbl.com/p/blackbox?op=scheduler&r=2019-07-08+00%3A45 (dorfs)
https://fumbbl.com/p/blackbox?op=scheduler&r=2019-07-01+12%3A45 (cdorfs)
On both occasions it appears the box went with the highest suitability scores.
Incidentally:
| Quote: | | This is a bit tricky to describe with a formula. Essentially, each race vs race matchup is tracked by the site for different TV levels and a ranking (calculated similarly to the win probability above) is tracked. This factor is applied to calculate the true win probability |
None of us really know what this is, nor do we have the exact data for it. But given years of minmaxing and cpomb, it's not entirely unlikely that some races actually have matchup biases that you wouldn't really expect.
eg. UW vs Dorfs. You'd never really want it, but in crp with access to a wiz, 2-3 cpombers+, babes, apoths, cards, this matchup would end up being a lot tighter than you'd think sometimes. |
_________________
|
|
MrCushtie
Joined: Aug 10, 2018
|
| Posted:
Jul 10, 2019 - 07:17 |
|
| Antithesisoftime wrote: | Looks like a filthy minmax coach trying to game the system got hit by the system working as intended, then comes to the forums to whine about it because the system worked against him this time.
|
If 1620 TV Underworld is minmaxed, what isn't?
Then again, with the benefit of inexperience I can't see much of a difference between minmaxing and rational team building. By reputation, playing any of smallman's teams is awful because they're the epitome of minmaxing, but when the dread day happened and I finally got drawn against him, I didn't find it a horrible experience at all. Maybe everyone just needs to build a team of goblins stacked with Dirty Player and activate with that. |
_________________
 |
|
ArrestedDevelopment
Joined: Sep 14, 2015
|
| Posted:
Jul 10, 2019 - 07:22 |
|
1260tv UW mrcushtie  the team in question in the OP's case have never broached the heady heights of 1500tv. |
_________________
|
|
|
|
| |
Gnomes are trash
Search
