Guru Selection

[Vandaley]

In some games the selection of the Guru's suffers from something called "selection bias". Any game where there is a 2 player format will inherently over select players who participate solely in 2 player games. This is because the selection criteria, as I understand it, is on games played in any format. Two player games take a lot less time to complete than 3 or 4 player, thus those players have the ability to complete more games. Realtime 2 player games are more likely to run as well.

This is playing out in the gurus voting to change arena settings in some games to 2 player, games that feel objectively worse. Puerto Rico, for example, is not 2 player, this game's official rules are written for 3-5 players. Likewise Happy City is a game that is recommended for 3+ and it has gone to 2 players.

Is there a way to include other players in Guru's? Maybe give more weight to arena matches completed?

Sam

[ufm]

Here is the list of games with player count change.

Increased (out of 2p now): Uncle Chestnut's Table Gype, King of the Pitch, Expedition: Northwest Passage, Diams
Increased (was already multiplayer): Go Nuts for Donuts, No Thanks!, Automobiles, Iwari, For Sale, Homesteaders
Decreased (previously 3p+ but now 2p): Dragonwood, Luxor, Libertalia, CuBirds, Balloon Pop!, Via Magica, Just Desserts!, Chinese Checkers, Kami, Cacao, Beyond the Sun, Trekking the World, Troyes, Puerto Rico, Fluxx, Florenza: Card Game, Happy City, Dice Hospital
Decreased (still multiplayer): Not Alone, Marco Polo, Saboteur, Incan Gold, In the Year of the Dragon, Hugo

Since 7 Wonders and Stone Age have adopted rotating format, I excluded them.

Games with higher player count setting tend to have significantly less games.
That leads to much rarer matchmaking (with longer waiting time obviously), and players usually don't like it.
Most of games went 2p have small player pools and 4 or less active leagues.
For example: viewtopic.php?f=320&t=20532
I also have experienced painful matchmakings in Letter Tycoon (which had 4 leagues but had to wait literally hours to get a real-time match).
Multiplayer setting just for the sake of multiplayer? No thanks.

A different point of view for Puerto Rico: 2 player mode was initially a houserule, but it was so popular and went to get an official variant!
Besides, some strategy games are better with 2p to be competitive (especially with current Arena matchmaking system).

So... it is not really 'biased'. Even there are some games with increased player count.
If you want to keep multiplayer mode try pursuading other gurus into it or proposing rotating formats like 7W or Stone Age.

[Vandaley]

With respect, you have not addressed my comment about the gurus themselves suffering from selection bias. Two player players will be overrepresented in the group of gurus.

[ufm]

Emergence of 2 player gurus means 2 player mode of that game is good enough to be played frequently in BGA.
Players must play games to be selected as gurus.

For example, Agricola, Splendor, and many other Arena multiplayer games are often played as multiplayer in normal mode too.
2 player game of 6 nimmt is pretty much nonexistent.

Some Puerto Rico players may be displeased at this phenomenon, but as I mentioned adopting rotating format is also possible.
Happy City, I don't buy '3+ player recommended'. Most games are played as 2 player setting even in normal mode.
The recommendation on the package is usually suited to real life game groups than generic online platforms like BGA.

In short, it is natural to have a lot of 2 player gurus... when 2 player mode of the game in question is enjoyable.

[Vandaley]

This has nothing to do with Puerto Rico, I regret bringing it up. My comment is about how gurus are selected, they are selected by completed games. Two player games finish faster, thus those who prefer 2 player games will be overrepresented in the selection. This is called "selection bias".

For example, lets say that there is a game that can be played as 2 player, or 3 player.

Now lets say there are 100 people who play the most, they play the same number of hours of this game on BGA per week. Of these 100 people, 25 play mostly 2 player, 50 play mostly 3-player, and 25 play both about the same. I would estimate that a 3 player game, in most games, takes about twice as long to complete as a 2 player game (in turn-based at least). For simplicity in calculation, lets say that is 1 hour vs 2 hours.

This means that for every 100 hours of playing, the 25 people who play only 2 player will complete 100 games, the 50 players who play mostly 3 player will complete 50 games, and the other 25 will complete 66 games. Thus someone who plays only 2 player can spend half the amount of time on BGA but get the same number of completed games. This makes them more likely to log more games and more likely to be selected. Thus, even though only 25% of the players play mostly 2-player, they will make up much more than 25% of the gurus if say the top 20 are selected.




Now that I have typed that out, I guess my suggestion for this suggestion forum is to weight games. Make the competition of a game with more players worth more when counting how much a player plays.

[ufm]

Real time games, probably, or not for games with adjusted time limit or simultaneous actions.

Turn-based, not so much. Players can open and/or join multiple games at the same time.
Playing time itself may vary more, but the number of games played depends less on it.

After all, why should multiplayer games be worth more than 2 player games in a competitive format?
'Players can keep each other in check', nope, kingmaking and uneven matchmaking happen quite a lot.

[Vandaley]

I stand by my post, Gurus will be overrepresented by those who play 2-player more.

[ufm]

How would you distinguish overselection from sheer preference? On what basis would you weight games?
Without any figures your argument is just an assumption. 2 player guru domination didn't happen in all games and it won't.
You didn't consider adjusted time limit, simultaneous actions, multiple tables, and total time investment. Players can't play the game alone.
You assume turn-based time limit in your scenario, but delays in turn-based games are mostly from waits between turns, not game play itself.
Actually it is more appropriate for real time games (and still not accurate).
And again, multiplayer is not essentially the better format for competitive platforms. No reason to value multiplayer formats higher.

[Vandaley]

I have a ph.d. in statistics, this is a common type of bias that the rules of gurus fits perfectly. It might be true that 2 player is preferred and 70% of players want it, still in this case they would be over selected and much more than 70% would be gurus. The argument that it might just be preference is immaterial to whether selection bias is occurring or not.

[ufm]

With all due respect, an argument from authority is unacceptable. Please bring the actual figures, not a hypothesis or an estimate.
Even your own post on Happy City player group suggests the effect of player count on individual preferences.
A proper and universal standard is required to calculate and rule out so-called bias from preference.
If nobody provides one, it is impossible to do so on every game (more than 200) in BGA.
Or you seek a special treatment for a chosen few games?