We have a whole nest of bugs[1] to do with different gold_upkeep_styles,
and the AI / CM handling them more or less well.

These bugs are languishing. I don't feel we (or at least I) can make
progress on them until I understand better what the original motivation
was for having the different styles in the first place.

Unless I've missed something, they all behave exactly the same until you
run out of money, and then they only influence exactly which buildings /
units are sold. Is that the intended difference? How do ruleset authors
decide which model they prefer?

My thoughts on the topic are scattered over the various bugs, but my
most detailed analysis is in the second half of

I guess my main uncertainty is whether the difference in asset-selling
behaviour really is the reason why all this was implemented -- it seems
like a lot of work for that result -- in which case working to make the
AI / automation behave as similarly as possible, despite the rather
different economic model, makes sense; or whether there's something I'm
missing about the behaviour of and rationale for this feature, in which
case tweaking the AI might not be the right solution.

The choice between two styles appeared in RT #40619, r15610, in 2009.
The third style appeared shortly afterwards in gna bug #13582, r15687.
(It looks like Matthias was driving this, but we haven't heard from him
for a while.)

Browsing open tickets, these are the ones I think are entangled
(although I haven't checked them all thoroughly):

  "Governor minimal surplus of -20 not low enough"
  "Governor Prefers Gold to Production"
  "cma: always 1 tax collector"
  "Citizen Governor inaccurate surplus limits"
  "AI fails to account for building expenses with gold_upkeep_style=2"

Freeciv-dev mailing list

Reply via email to