[MUD-Dev] TECH: Trusting Network Clients

[Sasha Hart]

[Raph Koster]

> A lot has to do with a sense of consequences.

Positive *and* negative, as you point out implicitly in referring to
reciprocal altruism. The incentives which originally drive the
behavior are easily as important as the lack of punishments after
the fact, and can even make attempts at punishment ineffective or
temporary.  The prisoner's dilemma wouldn't be a dilemma if
defecting didn't have a point. And in many cases, cheating would be
a lot less attractive if it didn't make you so powerful, hated, and
attended-to.

>>  Is there a way of creating a game (in the holistic sense) such
>>  that an extreme majority (90%) of players are fully inclined
>>  towards fair play, despite opportunities to cheat?

> Only by policing.

You might just mean by "policing" - "having consequences," which I'd
certainly agree with. Although some people would also emphasize
imitation of others, stuff like that. But reciprocal altruism, just
to reuse an example, doesn't constitute policing in the
ticket-or-jail sense of real police.  Refusing to buy drinks,
babysit children, or do business is not punishment in how it works
or in the behavior it elicits, so strictly speaking I'd disagree
with this very strongly.

> Based on various studies, the most successful policing appears to
> be capricious, unfair, harshly punitive, and near-random.

These are what I understand to be the basics of effective
punishment, out of lab research on punishment (which may be a
crucial difference from the stuff you are citing):

  1. Magnitude - Big punishments suppress more

  2. Delay - Fast punishments suppress more

  3. Stimulus control - If players can recognize conditions in which
  punishment won't occur, behavior won't be suppressed by punishment
  in those conditions

  4. Get rid of the incentives driving the behavior, and you make it
  more likely that punishment will work, and that when it goes away
  that the problem behavior won't recur

  5. The same punishment applied at the end of a sequence of
  escalating punishments is much less effective than that punishment
  presented first thing. If you identify problem behavior, don't
  give the player the chance to habituate and become a hard case.
  Then you are wasting your time punishing them, and probably also
  making them angrier than they would have been.

  6. Consistency matters a lot

  7. How much players like you probably has a lot to do with how
  they take your more nominal punishments. This recommends against
  being excessively unfair, or continuously (and ineffectively)
  applying harsh punishments when they won't work, etc.

There is probably some kind of interesting difference between the
studies we are looking at - it would be useful to understand what
that is. The basis for the above claims is predominantly
experimental research, both animals and people, tightly controlled
circumstances, no actual "police" or "rules" per se, etc.

Some other interesting research has indicated that punishment brings
up all kinds of behavior even as it suppresses other kinds -
aggression, efforts to circumvent enforcement, shifts to other kinds
of misbehavior, etc. It's also a no-brainer that lots of harsh
punishment can really turn one off on a relationship, and that
people who hate your game are a little more likely to try to bring
it down, for example. Thus, it really is worthwhile to make the
punishment that is used as efficient as possible.

Sasha

_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
https://www.kanga.nu/lists/listinfo/mud-dev