- Additional complexity is annoying
- Reduction of thief income might have unclear ramifications for the domain ecology
- On the other hand, our thieves were generating operating revenues far in excess of that intended by the system's designers. Scaling a cash-generating Fraud hijink to intended levels might solve this problem, with extra cash from stealing and selling cargos of treasure mapping being a bonus for work done
- Agreement that reduction of timescales down to weeks is a nice idea, but punishment remains problematic
- Realization of incompatibility between the mid-level wilderness exploration style of play and thief hijinks being town-centric
- The immediate nominal counter is that magic research is also town-centric, but the counter to that is that wizards can be gathering monster parts and unknown spells while wilderness adventuring, whereas the thief can't really use wilderness adventuring as leverage for hijinks
- Treasure hunting and stealing are both enablers for the wilderness game, but there is nothing in the wilderness that compels a gold-desiring mid-level thief out into it except want of magic items.
- Upon further reflection, Witness' comment on previous post solves this to some degree. The thing in the wilderness for thieves is not being in town while law enforcement are investigating the case. Hate to use the term, but an 'aggro'-like mechanic might serve. Multiple hijinks in rapid succession boost probability of being caught, lying low reduces probability of being caught, and being out of town reduces it even further (but sometimes they figure it out anyway, the wanted posters go up, and the bounty hunters come out...).
Speaking of scripts, been thinking about the Mother of All Scripts again. Some stuff from work has me thinking about ncurses and XML (both sort of groty technologies, but serviceable). In particular, XML seems like it might be somewhat useful for encoding deep structures in text; all the contents of a 24-mile hex, subdivided by 6- and 1.5-mile hex, including NPC and monster stats down to treasure, spells known, and Markov-generated names is not an easy thing to put into a text file in an organized and reloadable fashion (see: the Art of Unix Programming). XML could do it, but it would be verbose about it. I guess that's the tradeoff. The other option is to handle depth via directory structures in the filesystem, and then put hundreds of small text files all over the places throughout that directory tree. They're sort of equivalent; seek a way to handle a gross use case, end up with gross alternatives. As for ncurses, the problem there is one of displaying deep data in a useful way, as well as handling hex-mapping (and zooming in multiple layers of hexes) in a way that doesn't involve actual graphics...
So yeah... cue the A(CKS|X)ML jokes...