ORIGINAL: swagman
ORIGINAL: CommC
I agree with you 1000%, swagman.
UV is remarkable in that it gives the player nearly complete control over things that matter little, while giving the player little control over the things that matter the most... and leaves those important decisions to an AI (I use this term loosely) that manages to botch things horribly. Friendly AI is just as flawedas enemy AI, making PBEM against another human player equally pointless.
This is the gist of my criticism of UV. Wish I had said it.
Many responses have misinterpreted my complaint to be about the AI when playing a computer opponent. While the computer opponent leaves much to be desired, I am infact talking about the tactical AI which runs battles for both human and computer players. Battles are run by the TG AI commander, however, the nature of the simple AI beast means the TG commander acts in complete ignorance and independently of anything else. For example, if we set React to Enemy ON for a decoy carrier group with no aircraft, would it turn away and retreat when the US carriers pursued or would it charge forward to meet them head on?
Historically, TG commanders were given information on a need to know basis, but he would have knowledge of, hear rumour and be influenced by overall strategic planning, local theatre imperatives and priorities, naval/command doctrinaire shifts, and theatre or higher mission commands. Whether he was informed of these or not, he would receive direction in his briefings and orders. All of these things influence how a real TG commander would act in a given battle and so result in a different decision outcome. However, the AI TG commander only knows how to fight battles one way, qualified by react to enemy and level of TG commander aggression.
In addition, the AI TG commander acts independently of higher command and other TG's. For example, surface fleets set to bombard always cut and run, even though as surface combat groups they would historically be expected to stay to protect transport convoys.
The point about the game being strategic is in reality inaccurate. The game is really simply non-tactical. Strategy was never about micro-management of logistics. I doubt that MacArthur or Nimitz spent 99.9% of their time telling port crews what to load onto which ship and which squadron was to be rested this week. They made plans and issued orders, and others ensured they were carrie dout that the transport fleet were available and loaded on time, just as local TG commanders launched strikes etc. Strategy is about making plans and issuing instructions which the local commanders will hopefully interpret and apply correctly and, in this, UV fails.
Certainly, the game is so complex it would probably take a decade, a Cray and a team of post-doctoral scientists to build an AI to satisfactorily deal with it. However, what I am describing is limited to automating the logistics and creating a basic command template so that the local AI TG commander will not act in isolation of strategic planning but will act more in unison with it.
In a game combat situation there are a limited number of action scenarios, depending on your own TG mission, your strategic mission, other friendly TG's nearby, the distance to them, the presence of enemy TG's, the distance to them and the type and number of enemy TG's.
I expect in UV and probably in WitP, these scenarios have been hard-coded into a set of rules that the AI TG commander carries out with very limited qualification. My suggestion is to allow these rules to be variable through the use of a "matrix of decisions" that reflect these various scenarios, that sets behaviour guidelines in different circumstances, and that includes a cascade effect of influence from the theatre commander, through the operational commanders to the TG commanders.
The developers come back will likely be that they can cover off 90% of cases by hard-coding the rules and the other 10% can be put down to the vagaries of the local commander, which the human player can to some degree remedy by micro-managing the task-group. Micro-managing task-group movements in an attempt to get around development issues is no more part of a strategic game than loading task-groups.
Would a decision matrix do much better? Probably not, but what it may do is more accurately model the decisions that do occur, particularly by including as a driving factor the orders of higher commanders. After all, no man is an island.
swagman, next time choose another nickname. Maybe it could be - Tolstoy.[;)]?
