InfiniteMonkey
Posts: 355
Joined: 9/16/2016 Status: offline
|
One of the ideas for the new editor that I have been kicking around in my head has been a feature I have dubbed "Scenario from Save". The idea is simply to take a saved game file and turn it into a Scenario. I originally envisioned it as just a way to turn an interesting game situation into a Scenario. As I've pondered it, I can see some uses for it I did not originally envision. Disclaimer: Before I go too much further, I want to acknowledge and have acknowledged that the ideas listed below are a kludge. Ideally, the AI would be modified as the game progressed and there would be more slots for scripts available and more script features (like an actual scripting language), stop loss settings, dynamic sizing, wildcards for LCU/air groups, etc. I'm simply trying to think of ways to work with the few tools we get to address some situations that cause WitP:AE players angst. Anyway, some thoughts on how it might be used: 1) It can serve to accelerate testing of AI scripts. Take PaxMondo's situation as he refers to in this thread: http://www.matrixgames.com/forums/tm.asp?m=4331665&mpage=3 The problem is that each iteration of his AI script he needs to play the game to the point in time the script activates to test it. He has no way of knowing if his scripts work as desired until he gets the game to the point the script would trigger. To test late/mid game scripts takes playing the game to that point. So, how might you use "Scenario from Save" to help with that? a. Pax creates his scenario file and identifies some guinea pigs (hereafter referred to as GP's) to test it for him. Each of those GP's (General Patton almost HAS to be a tester, right?) play the game for a period of time - say 3 months. At the end of that period, they each send Pax the save. Each of those GP's might be asked to take a different tack in that phase of the game. Brave sir robin, Palembang defense, etc. to simulate different player approaches against the AI. b. Pax takes the received files and evaluates his script performance and determines if tweaks are needed, adds new scripts, etc. If he determines that new scripts are needed, he adds them to his existing AI file. c. The AI file gets embedded into the save file when the first turn of game is run, so it is locked in. Here is where "Scenario from Save" comes in. Pax loads the turn file into two copies of Tracker - one for Allies, one for Japan. Then in the scenario editor (SE), he selects "Scenario from Save" and gets prompted for 4 things: folder location of the two tracker instances, game date to create on, and a scenario slot to save the new scenario to. SE then pulls state from both tracker DB's and creates a new scenario from the state in tracker and saves the files using the scenario number requested. d. Pax copies and renames his new scenario file to match the new scenario number. e. Pax can then zip and send the complete set of scenario files to his GP. The whole process begins again from step one, but with from a new, later start date. f. His scenario file can evolve with the state of the game without constantly restarting on 12/7/41. He can even go back to "save points" in the game. If a GP has a save file from before a bad/catastrophic scripts began, he can modify it and try again. g. The above DOES come with costs. There is state information that cannot be saved (pilot skills are not in tracker/scenario files - only exp), or should not be saved (e.g. KIA's clutter up pilot DB. Only 20k pilots can be in scenario files, but 70k in save files). However, it could greatly improve AI testing (and therefore make it easier to test AI scripts). h. This works in part because the AI refers to bases, LCU's, air groups, HQ's, etc. by their "slot id" which could be preserved by "Scenario from Save". 2) Eventually, another use for this might be possible - to improve AI performance in AI games. By extracting state from a game and rewriting the AI scripts to handle the current state of the game, the AI could be made more reactive to game situations. An expert system AI written to create AI scripts could dynamically analyze the game state and generate AI/new scenario files to better fight the war and provide more variety to the AI play. An AI file only allows for 1999 "scripts" to cover the entire war. If there were 1999 available to fight a 3 month period of the war and they were "written" based upon the state of the war as pulled from tracker, it might be possible to greatly improve the quality of the AI as an opponent. 3) Finally, if 2) occurs, then it becomes possible to address the "Disappearing JFB" phenomenon. If a JFB decides to depart from a game and provides the password to his opponent, 1 & 2 above could be used to turn a PBEM into a AI game. While the opportunity "To crush your enemies. See them driven before you. And to hear the lamentations of their women." would be mitigated - at least a game could be played to a finish by the Allied player against a marginally competent opponent.
|