Fun! A good implementation of "engineer a better score trap" style play, I had a good time.
There's no end-game, so when I pushed the SCORE indicator into the "unidentified unicode entry" region (where I get boxes instead of text, like trying to shove Japanese or Arabic script into notepad) I called it a day. Adding an end-state or goal could improve things a bit?
Great job capping variables and adjusting things so that the system doesn't crash when the numbers get really big! I'd consider taking the cost-cap for upgrades into account if you implement an end-game - scaling up became super easy once the income gets a log + over the cap for upping a point multiplier.