Chapter17: Human OCR > Computer OCR
We continue to have problems in combat because of OCR. On the discovery of the static RAM layout, we're going to go all-in on RSD. I've made Moose an offer so that we can identify the combat area of RAM:
Moose, I have excellent OCR skills as a human and can help. I will take notes with each combat so that I can tell you the count and (active) values for monsters at each round of combat. Deal? ;-)
If you can note, for each round, something like:
Round 1: 5 ORCS (5) Round 2: 5 ORCS (1) Round 3: 1 ORC (1) or whatever Wizardry actually displays,
then we can line those human-read values up against the RAM snapshots and search for fields that track:
total monsters in the group active monsters in parentheses possibly asleep/paralyzed/etc. versus dead group identity/type transitions caused by KATINO, kills, fleeing, and so on
The parenthetical active count is especially valuable because it changes independently of the nominal group size. That gives us a much stronger fingerprint than just looking for kill counts.
You can simply type the observations to me after a run; no special format is required. I’ll correlate them with the corresponding combat rounds in latest.log.
Human OCR + machine RAM diffing. Very period-appropriate Wizardry research team. ;-)
Then things fell apart in all kinds of ways. I realized navigation was acting strange because the Campaign I was running had been running for a number of days, so Moose was skipping areas of the dungeon that were already mapped. That led to a feature idea I had for making a metric that could be assigned to each character, so that we'd know if it was worthwhile to resurrect them if they died.
That design was pretty cool and developed into the concept of Prowess. Each character has a Prowess, and the party as a whole has a Prowess. There's a base Prowess, and an Available Prowess, where the Available one decreases from the base as factors are introduced - like a reduction in HP, or spell points, or if the character dies (Prowess drops to 0). Prowess is also used to determine if the party can go to the next floor of the dungeon.
Anyways, we were working on all this, but now it's getting late, so we're signing off. Here's tonight's summary in 3 sentences from Moose:
Tonight I turned several RAM discoveries into durable Wizardry knowledge: confirmed the packed characteristic layout at $42–45, added attribute decoding to RSD, and introduced a provisional Prowess model for judging character capability. I also fixed maze handling for out-of-bounds movement and door preference, then traced a fresh-campaign regression to a bad assumption that roster A–T selectors were permanently tied to fixed RAM addresses, correcting the Rosetta Stone and selector logic accordingly. Finally, I removed the obsolete stock-character purge, fixed spell-audit selection to use the current displayed roster safely, and squash-merged all Chapter18 work into main.
Thanks Moose. Good summary; don't know if it makes sense to anyone except you and me though :-D