Re: Giving an application a window icon in a sensible way

Bent C Dalager wrote:
As for game theory, if that is what you are trying to apply here then
it seems to me you are entirely ignoring the shadow of the future in
your analysis, and you may be stuck in a trivial model where you think
there are only two distinct players in the game.

Actually, my model is fairly sophisticated. The model I use for verbal
mudslinging (wherever it might arise) is as a game where there is a one
to one correspondence between participants and people, and each
participant has a score. This score is never positive, and the only
actions available are to insult someone (lowering their score by an
amount that depends on the insult) or to defend (raising one's own
score). In fact, it's even more complex, since each participant has an
associated log of insults received, each with a weighting (the sum
gives the participant's score) and a logical model of the semantic
content of the insult of some kind. The score is worse for insults that
make nastier claims ("idiot" is not as bad as "murderer") but less in
magnitude for insults that are unbelievable (e.g. "he screwed his
great-grandmother's distant monkey ancestors -- every last one of
them", which is difficult to believe as it requires a) an awful lot of
screwing and b) a time machine). Actual evidence strengthens an insult.
A defense has a similar structure and involves evidence opposing an
insulting claim, or the pointing out of a logical fallacy whereby an
insult's conclusion is not properly inferred from the premises. A
simple model would break down a given insulting message into individual
insulting claims and a purported chain of logic from premises. More
independent chains from independent premises strengthens it; ludicrous
premises weaken or nullify it; and the severity of the accusation in
the conclusion is also significant. A rebuttal either breaks a chain
(by finding a flaw in the inferences made) or attacks a premise (e.g.
with countervailing evidence), and when successful, nullifies the
insulting conclusion. In the simple case of one insult with two
independent supporting arguments and one rebuttal that exposes a flaw
in one of those arguments, a score of minus 2N occurs but is increased
to just minus N, pending successful rebuttal of the other argument (or
the appearance of new insults).

The game is obviously negative-sum, so whoever starts an insult fest is
clearly an idiot. ;)
Once started, though, the goals of the participants vary. Some seek to
exit with a zero score and others seek to cause a target to exit with a
negative score. The best case scenario is a zeros all around, because
the targets' rebuttals are successful and the attackers give up.

It can be complexified, for example in trivial ways like permitting
someone to defend a third party; or considering a rebuttal's effect to
begin decaying if excessive time passes between insult and
corresponding rebuttal. As considered in practise, I assume that a
rebuttal that is a direct followup to an insulting usenet post is more
effective than a rebuttal located elsewhere, since a reader of the
insult is most likely to also read the rebuttal in the former case.

Unfortunately, there is clearly no strategy that guarantees a
zero-score exit. The closest is to monitor for, and rebut, insults as
they come, and to remain vigilant until a substantial amount of time
has elapsed without a new one, suggesting that all of your attackers
have given up on holding your score down and moved on to other targets
or quit the game entirely.

Perhaps a simpler analogy is to a bunch of people in a swimming pool,
in which a person can hold another's head underwater, a person can keep
their head above water (requiring constant effort if under attack, and
as a simplifying assumption none otherwise), and one's score is one's
blood oxygen level (which drops when held under, rises when you are
afloat, and, in the real world, becomes dangerously low after four
minutes of being held under). Clearly in this model the target not
defending is suicide, but the attack is still futile (given the
assumption that the target can indefinitely survive through continuing
defensive action; in a real pool of water, with people of varying
strength, that obviously doesn't work) unless the defender is really,
really stupid.

Of course, the first serious complication (in both models) is adding in
a rule by which *an attacker is allowed to try to trick a target into
not defending*...

Actually, there are some variation scorings that may apply to real-life
situations. In one, the attackers' goal is to waste targets' time, or
else it's to either waste targets' time or make an insult stick. In the
former, the defender should ignore the attacker instead of defend, but
this only applies to a real life situation in which a) there's no third
party present for any of the insults and b) the attacker gains somehow
by the defender's waste of time, despite the attacker spending a
comparable amount of time. In the latter, the attacker always wins,
since the defender is either insulted or wastes time; to "win" all you
have to do is make an insult(!).

Timewaster? You bet, but then again, this *is* Usenet...

.