|
swagman -> Convince me WITP will be worth playing (4/22/2004 12:06:51 AM)
|
I have played both Gary Grigsby?s Pacific War and Uncommon Valor. I am not familiar with this thread, but as WitP carries on the tradition of these other games, I am expecting the underlying flaws to persist. I believe the principal cause is that while GG is brilliant technically in terms of game algorithms, the AI is particularly lacking.
Anyone who plays Steel Panthers will recognize the poverty of the AI in these games, where the programming effort has gone into the technically detail. Great scenario designers such as WBW have made up for this by being smart in how they made their battles and used features like reinforcements. Nevertheless, even they almost always have to resort to the only equalizer that worked, which was to give the AI unrealistic numerical superiority to make up for the number of units a competent human player would blow away due to the stupidity of the AI.
The opposing type of game is what I call the AI moderated style, in which the AI plays most aspects of the game except the strategic/tactical decisions that are left up to the player. These games are usually less technically detailed, but more naturally balanced, since the game AI is able to execute the strategic decisions of the player. This is something UV does not do.
I believe the poverty of the AI in UV will most likely be begat to WitP. In UV I have seen the AI move its task groups back and forth between two non-base hexes in the course of a single turn, something the human player cannot do. Thus, the AI operates fully for the computer-controlled player, while it does not for the human players moves, whose turns are really computer moderated. There is no facility for defining for the AI the strategic intention of the player. It will thus treat a carrier escort mission the same whether the protected TG is a strike carrier group or a transport mission. For example, the bombard rules have been defined so bombard TG?s can strike down the Slot to Lunga and retreat. As a result, when they are sent to bombard Noumea they retreat right by US bases that bomb the ships to scrap, despite there being an entire ocean for them to sail safely away. The only way to deal with some of these issues is for human intervention to cheat the rules in order to trick the computer moderator into doing something vaguely sensible, instead of the AI doing something sensible because it has been sensibly programmed.
A game such as UV can have up to 610 turns, taking some 200 hours to play. Of those, maybe only ten turns really matter. 3 or 4 carrier battles, a few surface engagements and a couple of important invasions. The rest is micromanagement, be it of logistics/supply or in cycling squadrons to minimize fatigue and maximize experience/morale. Some players obviously enjoy it, but for a large potential market it makes for one DULL game that, when combined with a poverty of AI, makes for one DULL and FRUSTRATING GAME.
The first thing that needs to be done is to provide an option that effectively and efficaciously simplifies the game logistics, while allowing the control freaks out there to still micro-manage everything themselves. At the moment, a human player has to micro-manage because the AI does it so poorly or not at all. This could be achieved by such things as:
· A squadron rotation matrix: for example allow one or more squadrons of the same type (e.g. level bomber) at the same base to be assigned to a rotation group. The player is allowed to set the default fatigue level at which a squadron is rested, the minimum/maximum number of squadrons to be on roster in any turn, etc. It could even go to the level of the number/percentage of pilot available to operating planes with a fatigue level below X, meaning a squadron with enough pilot could still fly missions even though its average fatigue exceeded the defined rest threshold, because it still had sufficient pilots bellow the threshold to fly the majority of its available planes;
· I don?t know about the Pacific, but in Europe both fighters and bombers from multiple bases were co-ordinated into a single mission. A matrix could be developed to facilitate the same thing from nearby bases, by defining which a/c types from which bases can operate together on the one mission. This could then extend to squadron rotation across multiple bases.
· Allowing multiple way-points for task group?s routes. Rather than simply setting a destination hex for a TG, allow multiple way-points along the route so the player can choose a preferred route.
· Allowing multiple route options for task groups. Currently, computer controlled convoys follow the same route every time. A human opponent who knows this can place his subs along this route and devastate such convoys. A human player must therefore micro-manage these convoys to avoid the expected route. Multiple route options Thus, a continuous supply convoy could have several defined routes between base A and B, and follow them in order or randomly.
· Allow divergent routes for task groups. This option would allow a convoy or TG to diverge by an optional number of hexes from the direct route between to bases. Thus a routine convoy that supplies the most needy of many bases could take a route that diverged 2 or 3 or 4 hexes from the most direct route or the selected way-points.
· Searching needs to be improved, so that search aircraft have increased flights along player determined directions of threat. For example, you don?t want search planes at Noumea flying north. You want them flying east if you have two bases already searching North.
In most PBEM games, I have seen my TG?s do the most stupid and frustrating things. For example, air groups ignoring other carrier groups at 4 hexes distance into order to sink a damaged destroyers/cruisers in single ship detached task groups at 8 hexes distance. This allowed my carriers to be sweetly turned to scrap before launching a single strike against the opposing carriers. Such acts are not historical, not consistent with good strategy or good tactics. It simply reflects flawed AI.
Certainly, as proven at the Battle of Samar, task group?s commanders can prove unreliable, with Hague running off after the Japanese decoy carriers and the Japanese Northern force withdrawing at the very time the US forces were vulnerable.
If the game is to be properly responsive at a tactical level, the AI should reflect the general tendency for task group commanders to follow the strategic planning of their commander, and for some task group commanders to act with independence. The possibility for independent action is not properly modeled by a simple aggression rating. At the moment, the TG commanders simply follow the game mechanics of moving from A to B etc, with some variables such as competency rating. There needs to be a method for defining the strategic intent of the player, and to integrate the behavior of the task groups other than a simple ?follow?. If an invasion is being mounted, you don?t want task-groups all acting independently, which is currently the case. Historically, the rule was they acted on instruction and advice of the local overall commander, which in reality is the player. Sometimes, such as at Samar, the TG or TF commander would make his own interpretation of those instructions and advice, but rate
This could possibly be achieved using a matrix of command, which recreates several levels of command. At each level, the commander has a number of decision options which the player can pre-define and rated, for example react to enemy decision could be set as yes, no, conditional on enemy type or refer to commander. A rating priority from one to 10 would define the importance of the decision. If two decisions with different rating occurred in the same turn, the more highly rated decision would take precedence. Choosing to obey the priority rating and the decision requirement would then have an outcome check against the TG commander?s inspiration, independence and aggression. This would also involve a check against accuracy of available intelligence, being the number of enemy sightings, the type of ships sighted, the number of enemy search aircraft spotted etc. It would also involve a check based on available communications and operational silence requirements to determine if a commander could contact the next level of command and, if failed, the TG commander would have to make his own decision based on the matrix and checks. This means some conditionality would be required in the command decision matrix, such if decision circumstance A then decision Y, if circumstance B then decision Z. etc.
If a TG decision outcome was to refer to commander (either because that was the set outcome for that decision or the TG commander check resulted in that outcome), then the TF decision calculations would be used; and so on to the Operational Commander at that top. At this level, there would have to be a complex and detailed matrix of command decisions, covering the normal range of action scenarios in the game currently left to the local commander to decide. This would represent the players overall strategic plan and requirements. The Operational Commanders decision outcome would then be passed back down the chain; with associated commander independence and communications checks along the way.
Ideally, a request for Command Instruction to the higher commander would occur at one of two levels. If communication was open, then the Decision would be passed on up to the higher commander, who would make his decision using the decision matrix and his independence/aggression check and (if communication was open) refer it back. If communication were not open, then the request for a Command Instruction would assume that the TG or local commander was ?aware? through communication of the Operational Commanders intentions. This would be replicated by the TG or local commander using the Operational Commanders decision matrix, subject to an accuracy check, but using the TG or local commanders own independence/aggression/etc rating adjustment.
Of course, there are penalties for referring a decision to a commander. For example, the more removed a commander is from the action or TG commander, the greater the fog of war and lack of Intel, so the more likely these would impact the Operational Commander and lead to overly cautious or aggressive decisions. Also, there would also be a time delay variable, meaning an enemy could attack or escape in the time taken for a command decision request to be processed.
The responsiveness of TG?s to each other also needs to be improved. Sometimes you don?t want all your task groups set to follow to end up in the same hex, at other times you do. Hence, there needs to be additional classes of action other than simple follow, including the number of hex separation permitted. Sometimes you may want a bombard TG two hexes in front of your transport, with a cover carrier TG two hexes behind, and a surface combat escort TG in the same hex. In addition, sometimes TG?s are set to sail together separate because the following TG is slower than the lead or some other reason. In such circumstances, the lead TG should slow so that it stays with its companion, rather than both ending up in separate hexes.
|
|
|
|
