Fansubbing and Gaming (theory)

[Access]

What do you guys think? Is this too theoretical or useful? Comments?
I try to begin as simple as possible, and then work toward more complex but
'useful' games. I was thinking I just had way too much time on my hands,
but when I re-read it; I've been doing this for seven years and enough of the
conclusions one can draw from it seem applicable.

Fansubbing work in a group can be compared to a simple cooperative game.

Player 1: Choose between "work" or "stall"
Player 2: Choose between "work" or "stall"

If both players stall, nothing gets done, but no one cared in the first place.
+0 points for both.

If both players work, the episode is complete and a release is made. +1 points
for both.

If one member works and the other stalls, the work is wasted and the one who
worked is disappointed. +0 points for the player who stalled and -2 points
for one player who worked.

The game is played several times, and the goal is to end with the highest
positive score possible. The game is purely cooperative, players gain nothing
by lowering the score of the other player.

The analysis of this game is simple. One player should do exactly what the
other one does. Ideally, both players choose to work all the time.

In reality, no one wants to work all the time. So a non-hostile player will
always communicate an intention to stall before doing so. This way, neither
player ever risks losing points.

Only a hostile player who wants the other player to score badly would not
communicate an intention to stall instead of work. If either player is hostile,
eventually the game will degrade to both players stalling repeatedly.

(Refinement #1)

No one wants to work all the time. Quantify this.

1) The game is played a total of 8 times, sequentially.

2) A player can choose work no more than 4 of these 8 plays.

Here, the cooperative gameplay is unchanged. Communication is even more
important than before, since working is limited and failure to communicate leads
to work being wasted.

Now model different types of players.

"lazy" - can choose to work no more than 2 times (of 8).
"normal" - can choose to work no more than 4 times (of 8).
"workaholic" - must choose to work 4 times (of 8).
"super-workaholic" - must choose to work 6 times (of 8).

The cooperative gameplay stays unchanged, but now matchups are important.

Lazy players work best with other lazy players. They can work okay with
normal players, but not optimal from the normal player's perspective. They
are perfectly happy with any type of player.

Normal players work best with normal or workaholic players. They can work with
lazy players, but not optimally. They will be perfectly happy working with
a super-workaholic player, but the super-workaholic player won't.

Workaholic players will be unhappy working with lazy players. They can work
fine with normal players or other workaholic players. They will be perfectly
happy with super-workaholic players, but the super-workaholic player won't.

Super-workaholic players can work fine only with other super-workaholic players.
They will be disappointed with all other players; somewhat disappointed with
normal and workaholic players and very disappointed with lazy ones.

The super-workaholic players are the highest scoring (if optimally placed).
A normal player is always preferable to a lazy player. A workaholic can
be on par with a normal player, but is more picky in matchups.

(Refinement #2)

Back to the original game, incorporate communication and group pressure. Add
a third choice, "communicate".

Each player has the following choices.

work: "I will do it" and do it.
communicate: "I don't have time right now."
stall: "I will do it" and not do it.

player 1: work or communicate or stall.
player 2: work or communicate or stall.

A low-pressure group is one that does not care to make timely releases. In
other words, it will not penalize players for communicating. Here, the game is
virtually unchanged.

If both players work, +1 point for both.

If one player works and the other stalls, the one who works gets -2 points.

All other combinations, both players get +0 points.

Here, for non-hostile players, the game is virtually unchanged. Players have
no reason to choose stall unless they are hostile.

A high-pressure group is one that values timely releases. This makes the game
much more complex. The key is to keep cooperative play universally rewarding.

1) If both players work, a release is made, and both players gain status. Both
get +2 points.

2) If neither player works, no release is made and players lose status. Both
players get -1 points.

If one player works and the other doesn't. This is key. Is the player who
didn't work punished more for communicating or stalling?

3a) Work vs. communicate. The player who worked gains status in the group, +1
point. The player who communicated loses status, -1 point.

4a) Work vs. stall. The player who worked gains status in the group, but is
also disappointed. +0 points. The player who stalled loses major status, -2
points.

In the above game, hostile play is not rewarded, hence cooperative play is still
the norm. Only a sadist-indiferent or sadist-masochist player, one who wanted
to hurt others with no or negative regard for himself would choose to stall over
communicate.

Consider the following change to rules 3 and 4.

3b) Work vs. communicate. The player who worked gains status in the group, +1
point. The player who communicated loses major status, -2 points.

4b) Work vs. stall. The player who worked gains status in the group, but is
also disappointed. +0 points. The player who stalled loses status in the group,
-1 points.

Here, the game is non-cooperative. If a player cannot choose to work, the best
choice he can make is stall. This will, however, penalize another player who
chooses to work. The result is disastrous, even non-hostile players will choose
stall over communicate. Eventually the game degrades into a futile exercise
where working offers no benefit and both players choose to stall repeatedly.

Refine the scoring further and you will end up with a three-choice variant of
the famous "prisoner dilemna" problem.

(Refinement #3)

Back to the original game, look at the basis of communication. Communication
involves two things, announcing one's intention (speaking) and observing an
announcement (listening). What will happen when full communication between
players is not allowed?

1) If both players can speak and listen, there is full benefit. Full
cooperation can ensue.

2) If one player can only speak and the other can only listen, there is half
benefit. Limited cooperation can ensue.

3) If both players can speak but not listen, or both players can listen but
not speak, there is no benefit. Players cannot cooperate (unless they break
the rules and cheat).

(Refinement #4)

Multiple players. This game extends well to multiple players, and most of the
concepts established in 2-player play carry over to N-player play.


Combine the concept of multiple players with the concept of player types
explored in Refinement #1.

Here, certain pairings are more optimal than others. For instance, if you have
two lazy players, a normal player, and a workaholic, the most optimal would
be to put the two lazy players together, and the normal player with the
workaholic.

In certain groups, there may be no beneficial pairings. For instance, if
you have group where there are three lazy players and a super-workaholic player,
there is no static pairing that will please the super-workaholic. The best
score the super-workaholic can get when paired with a lazy player is -10.

This begs the question of whether dynamic pairings are allowed. With
dynamic pairings, the super-workaholic player need not be penalized for being
in the same group as three lazy players.

Three lazy players: lazy1, lazy2, lazy3.
Super-workaholic player: Super1

phase 1: Super1 - lazy 1 (both work), lazy2 - lazy3 (both stall)
phase 2: Super1 - lazy 1 (both work), lazy2 - lazy3 (both stall)
phase 3: Super1 - lazy 2 (both work), lazy1 - laxy3 (both stall)
phase 4: Super1 - lazy 2 (both work), lazy1 - lazy3 (both stall)
phase 5: Super1 - lazy 3 (both work), lazy1 - lazy2 (both stall)
phase 6: Super1 - lazy 3 (both work), lazy1 - lazy2 (both stall)
phase 7: doesn't matter (all stall)
phase 8: doesn't matter (all stall)

The above case is optimal, the super-workaholic gets +6 points and each lazy
player gets +2 points. Each player scores his theoretical maximum.


Consider the effect of different levels of communication explored in
refinement #3 when more than two players are involved.

One feature of cooperative play is that if and any player is left out of the
communication, that player suffers. Eventually, they will stop choosing
to work, which will hurt the whole group.

[dj_tjerk]

So fansubbing is like poker?

[Access]

No.

Poker is competitive, played alone (every-man-for-himself), and you win by making the other players lose. If fansubbing was like poker, the group would be hell and nothing would ever get done.

Think of something like baseball, specifically what goes on between a pitcher and a catcher before, during, and immediately after the pitch. Before the pitch, the catcher and pitcher agree what kind of pitch it is going to be, using signals. Then the pitcher pitches it and the catcher is able to catch it. If the pitcher says he's going to do one kind of pitch and then throws another, the catcher might miss it, which is bad for everyone on your team. Only a pitcher who was a total jerk or someone who wanted his own team to lose would lie (to the catcher) about the kind of pitch he was going to throw.

[Zanejin]

This analysis is very interesting. Have you been studying economics or game theory ? It's always fun to read quantifications of social behavior.