[rfk-dev] robotfindskitten should keep stats

[Peter A. Peterson II]

Watching my roommate play robotfindskitten inspired an idea.

He would sit and stare at the screen for about 10 seconds after it first
came up, to try and sense where kitten was. THen he'd go find kitten,
and usually he'd be pretty close. 

Of course, maybe he was no more statistically close than someone
choosing by random.

WHich made me have a thought.

First of all, this is another good reason for rfk to have a .conf file,
so that if a sysadmin doesn't want rfk keeping stats, it wouldn't have
to. But here are my thoughts.

Running rfk -s(n) (or something like that) would print a list of the n
best kitten finders, based on some kind of algorithm averaging time,
distance, number of non-kitten encounters, etc. Or maybe you could
control what criteria it uses, say different switches to display the
fastest kitten finders, regardless of other factors, or the ones with
the least non-kitten encounters, and then have a switch that is like,
the best kitten finder overall.

TO keep things fair, it should only count stats on the standard number
of nki (it's 20, right?).  That way, like nethack, you can print a list
of the record-holding robots, and find out, are you really a better
kitten finder than Seth David Schoen? Does <big>Rick Moen</big> THINK
he's a good kitten finder, but really he's no better than anyone else?

How zen are you *really*, grasshopper?

Thoughts?

Pedro