Discussion
so then they're waiting too often, make waiting _cost_ tactical score, oh wait now they'll do the 'not technically waiting' do-nothing turns.
It's a weights-tweaking arms race that has had a lot of different iterations. surprisingly unintuitive lmao
At one point I had a concept of 'no change in score from last turn is a null turn, which costs score, unless you waited and then cost less' which was pretty good for a bit, b ut the 'from last turn' ended up being very ambiguous and not holding up
today's solution is a special function that queries a turn event for "knowingly significantly impacted the turn" which is essentially 'did you mess with an actor you knew about' but also includes waiting.
never doing any significant impact events gives you a small tactical cost to weed out the do-nothing turns or the 'just barely better than waiting on paper' edge cases of which there are dozens.
And last bit is simply tie-breaking on wait for the best-score query at the end which...yeah
feels good to rip up a ton of ugly plumbing that was trying to wriggle this thing into place and just have a single central place to define if an event constitutes 'impact'
written about it before but its so crucial to not get bogged down in large complicated fixes for possible future exceptions, it's an impossible and maddening task and the larger and more comprehensive you try to make it the worse off you'll be when the inevitable exceptions need to be applied.
minimal work-today tweaks are optimal because the occasional larger refactor will be obvious due to being purpose-driven with immediate need
write code to support the thing you're doing right now today instead of trying to support hypothetical future things you might want
Install bonfire.cafe
Get the full app experience
