In most cases, the military approach works. Wargames worked out, and played out, months before the 2003 invasion of Iraq showed, correctly, that the Iraqis would not be able to withstand the advance of two American divisions on Baghdad. Actually, there were supposed to be three divisions, but the Turks would not allow the U.S. 4th Infantry Division to come in from the north. The British also had a division that took Basra, and then watched over southern Iraq, and Iran, while the American soldiers and marines moved north.
But one thing this gaming did not show was the extent to which the Sunni Arab supporters of Saddam would continue to fight. This possibility was recognized, and the average outcome was some low level terrorism. But wars do involve a lot of chance, and when the Iraqis rolled the dice in the Summer of 2003, they came up with a major effort by the Sunni Arabs to get back control of Iraq. Militarily, this was a bad decision. American military planners knew it, and Sunni Arab Iraqi leaders knew it. But in war, theres always the risk that one side or the other will do something crazy. In the Summer of 03, the Sunni Arabs went nuts. Of course, two years later, they had lost their little suicide ride to glory. But American commanders were caught short by crazy moves their planners could have known about, and examined in detail in early 2003, and made better preparations for. Instead, it was left to the troops to scramble and improvise. Had there been wargaming in advance, of the Sunni Arab terrorism, the troops would have been quickly equipped with police tools (for raids), databases (for collecting and analyzing information) and advance warning that armored trucks and hummers would be needed in large quantities. This could have saved the lives of several hundred American troops.
The American army has long resisted using what if? wargaming aggressively and on a wide scale. Part of this has to do with the fact that letting things happen naturally can lead to a string of unexpected (but statistically possible) events that will, in some cases, ruin a wargaming exercise. The way the army runs wargames, dozens, and often hundreds of officers and troops are involved. These people have to be pulled away from their units, and their regular work (training, and maintaining skills and equipment) to participate in these wargames. The commanders running the wargames want to see how, for example, a large scale operation will play out. They do not like play out worst case situations, which can sometimes develop, as they sometimes do in real life. So average results are often used, and the unexpected events are brushed aside. But sometimes reality later asserts itself when you have moved beyond the wargame, into the real thing.
The American armed forces got as good as they are over the last few decades, partly through their extensive use of wargames. Actually, these things are usually called models, or simulations, and perform more like spreadsheet software in describing and modeling a complex process. Although these simulations look like the wargames civilians can buy (complete with the map on your computer screen, and all sorts of symbols, representing military units, to move around), the military versions lack one very important factor; chance. Military wargames do not, so to speak, like to roll the dice. Actual war is full of risky business, but for planning purposes, professional wargames tend to take the most likely outcome and just plug that in to the simulation.