Map Selection: I never get certain maps

[.Memories.]

It's a bell curve distribution for coins, at least. It's a little different for multiple variables, such as Outwitters' maps, though. Check this out if you want to delve into it further: Probability Distributions

[TheGoldenGriffin]

It doesn't take into account if the coin lands on its side lol. That looks nothing like the probability we learned in algebra

---

Whats the weird E symbol? It has numbers on all four of its sides

[.Memories.]

(05-16-2013 07:51 AM)TheGoldenGriffin Wrote:  It doesn't take into account if the coin lands on its side lol. That looks nothing like the probability we learned in algebra

---

Whats the weird E symbol? It has numbers on all four of its sides

That's a capital sigma. It means "sum of" in mathematics. You'll see it in Calculus and higher (and sometimes lower, depending on the class).

The first one you see on that site is summing 'kP(k)' from k = 0 to 4.

[TheGreatErenan]

Yes, that's correct.

Sometimes I wonder if the likelihood of getting a certain map as P1 depends in any way upon your existing record on that map going back a certain number of games. Or something along those lines.

I guess my thinking is that if the matchmaking system is designed to guide your record towards 50/50, then map selection could theoretically be involved in that.

[Szei]

This is one of the reasons I advocated replacing three maps instead of keeping them all for ranked play. You won't play some of the maps very often even if it is just down to pure randomness since we have so many. On a side note, in actuality it's not quite random because some maps are thrown out by P1 if P1 doesn't like the map.

[Szei]

I don't think there needs to be any "transparency" because I'm pretty sure it is already transparent. It's likely just random. Perhaps people are over thinking it.

As for awarding points differently based on which race you beat, I would rather them just try to balance the maps instead. Throw out the maps where one team has a significant advantage and replace it with a new map. Also there should be an option to veto some of the maps (3 or 4 probably) so that if you feel your team is significantly disadvantaged on one of those maps then you can avoid it. The veto system is used in Starcraft so Alex should be familiar with it.

But, I think we won't be seeing any major new systems or overhauls until an Outwitters 2 is made (if one is made).

[TheGreatErenan]

(05-16-2013 08:46 AM)OutwittersFan422 Wrote:  Is there any confirmation it does that? Based on all numbers that seems to make sense but Im not sure. It certainly seems to do that in terms of skill placement but Im trying to think. By far and away my best map in terms of win % is Long Nine and I seem to have gotten a disproportionally high number of games on LN in the past month as you can see by my map breakdown so i dont know

Personally I wish they would simply make the whole system completely transparent. And make it so that you cycle through games in perfectly equal distribution. Make it so that you could not choose race when playing a league game and for each individual map there are the following possibilities:

Take Long Nine for example:

S1 vs S
S1 vs F
S1 vs A
S1 vs V
S2 vs S
S2 vs F
S2 vs A
S2 vs V
the # denoting the turn and then repeat that for each race. making every league player cycle through that in whatever order doesnt matter. So for each map there are 32 possibilities. So you dont repeat something until game # 353 (32 possibilities on 11 different maps = 352)

Then internally keep general records of each matchup and base skill rating on that. So if I beat someone on Reaper with Adorables and the internal stats say that Adorables statistically outperform all other races on that map then I get fewer points than say if I did so with whatever the weaker races are on that map.

I think this would be 100% objective and really show who is the best. In all likelihood it will be many of if not all of the same people but it would present a much more tangible system and be much more accessible to the average player.

I don't think there is any confirmation on that at all. It was pure speculation on my part. I just wanted to know if OML ever officially stated that it was completely random.

I agree that more transparency would be nice, and a robust system is certainly desirable. However, I imagine it could potentially cause problems without having been tested like crazy. Anyway, OML is unlikely to change it at this point since they are working hard on TTL2 now. The system we have now is almost certain to stay in place.

[[PETA] Cor13:4]

That's true, about P1 having the additional advantage of picking the map inflating ratings, but I think it's not that big a deal.

Eventually they will hit a plateau where even picking the map will not win them the games and they'll be P2 sometimes also.

[TheGreatAnt]

I am not sure whatever "random" means in an algorithm, but it makes more sense in my mind if each map had a number in sequence (IE, 1-9) and then whenever someone starts a game, they get a number, then the next game started gets the next number in the sequence until it repeats, so every new game gets the next map.

[.Memories.]

(05-16-2013 10:10 AM)TheGreatAnt Wrote:  I am not sure whatever "random" means in an algorithm, but it makes more sense in my mind if each map had a number in sequence (IE, 1-9) and then whenever someone starts a game, they get a number, then the next game started gets the next number in the sequence until it repeats, so every new game gets the next map.

For the maps to indeed be "random", they would be labeled with a number (1-12), then a random number generator would pick a number "randomly" from 1-12 when each game is started. The map that matches that number would be played.

As any Comp Sci person will understand, random is in quotes here because nothing is truly random to a computer.