Computer games helping to train commanding officers.

The U.S. Army's Command and General Staff College, at Fort Leavenworth, Kan., has trained 600 students using computer simulations designed for video-game enthusiasts.

Such methods would have been unthinkable in the old days of paper-map exercises. Even with modern Army simulations such as Corps Battle Simulation (CBS) and Joint Conflict and Tactical Simulation (JCATS), hundreds of instructors and computer operators would have been needed to give meaningful training to a handful of students.

This year, 600 of the 1,600 majors taking brigade staff training at the college played TACOPSCAV, a PC-based hobbyist war game designed by a former Marine intelligence officer. The civilian version sells for about $25.

The comparison with the military's huge, expensive simulations is staggering. A CBS exercise to train division staff might require 400 or 500 participants, according to Jeff Leser, head of the college's Digital Leader Development Center. The problem is that only a few of the participants will actually be running the simulated division, while the rest will be busy controlling the division's subordinate brigades and battalions. That's because the software has minimal artificial intelligence. Participants must use mouse and keyboard commands to laboriously input instructions for every maneuver unit, friend or foe. Thus, putting together a multi-division exercise would be an administrative nightmare.

As for JCATS, at a June brigade staff exercise observed by National Defense, there were 40 students, split into two brigade staffs, on the second floor of a college building. On the floor below, sitting before banks of terminals, were seven instructors, 26 support staff (mostly retired military officers hired for the exercise) and 10 students drafted to control subordinate battalions. The brigade staff students telephoned their orders to the control room, where operators executed them on their computer screens. Add up the numbers, and you have more than 40 staff running a simulation for 40 students.

Contrast this with TACOPSCAV across the hall. There were 15 sections of about 40 students each. Supporting these 600 students were about 100 instructors and staff.

The TACOPSCAV students were broken down into four-person teams, with one sitting at the computer issuing instructions, while the others assumed the roles of brigade staff. Each team was cycled through a series of four-hour scenarios. Set in an overall context of U.S. intervention in a Trans-Caucasian conflict, there were seven scenarios, each with six possible variations to keep the students guessing.

The TACOPSCAV artificial intelligence allows players to issue general instructions and "allow the computer to do the rest. Like most entertainment games, TACOPSCAV can be played against the computer or versus human opponents versus linked computers. It also comes with an easy-to-use scenario editor, so instructors can quickly create a meeting engagement or some other training scenario.

From an instructor's standpoint, small simulations like TACOPSCAV also have the advantage of brevity, said Lt. Col. Blaise Cornell-d'Echert, a tactics instructor and exercise coordinator for the CGSC's advanced operations war fighting course.

In previous years, students would plan large staff exercises in a classroom, and then input them into JCATS for execution. "They would be able to spend one day executing the plan, and maybe there would be enough time to run it a second time," Cornell-d'Echert said. But running a scenario once or twice only teaches a brand-new major so much. In contrast, the shorter prep time and scenario length of TACOPSCAV allows repetition. "Because those scenarios are shorter, you can do one, have an after-action report, and do it again with a modified version of the scenario."

When Cornell-d'Echert was a young officer being trained in staff work, exercises were map-based "and you just accepted what the instructor said." He was sold on the training value of computer war games when he began teaching at the college four years ago. One of his classes was assigned to devise an Anzio-type landing. They didn't believe Cornell-d'Echert's assessment that their plan would fail--until the instructor used a commercial war game called Decisive Action to demonstrate the probable outcome.

Budding staff officers need to see the consequences of their decisions. "Our students are novices," Cornell-d'Echert said. "They've just been promoted to major, and they never had a responsibility for planning and supervising battalion and brigade-level operations."

Some in the defense simulations community are concerned that simple commercial war games won't have the fidelity of more elaborate simulations with classified databases. But Cornell-d'Echert doesn't see this as an impediment; games that train brigade staffs don't have to include the precise accuracy of the M-1 tank's fire control system or the exact frontal armor thickness of a T-80 tank. "As field-grade officers, they are going to direct by instructions, instead of direct physical control over units."

What's important is that they grasp concepts of command, such as the importance of synchronizing fires. That only comes through repetition of scenarios that give students a chance to learn from and correct their mistakes. "We're looking for understanding concepts and providing them with a model of what synchronization of effects looks like," said Cornell-d'Echert. "Even if we're not validating the exact weapons characteristics, a qualitative correlation is good enough."

Cornell-d'Echert said this year's exercise was more focused on decision-making. TACOPSCAV enabled more students to participate in command activities. Some of the scenarios pitted them against the TACOPSCAV artificial intelligence, while others were player versus player.

Surprisingly, Cornell-d'Echert found that students learned more from fighting a computer opponent than fighting each other. Though the artificial intelligence in computer games tends to be rigid and vulnerable to exploitation of "gamey" tactics, the students were actually more uncertain about how the artificial intelligence would behave than their human opponents, who could be expected to use the common doctrine they were taught. The result was that human opponents mirrored each other. "When you're fighting someone you've been with for the past year, you know what he's going to do," Cornell-d'Echert said.

Beyond ease of logistics, using simpler computer games signals a shift in Army thinking, according to Leser. Bigger simulations like CBS required so much preparation that the scenarios became tedious. But the response from the field was that the college was turning out "good planners, but they needed to do better in execution." Simple computer games enable students to practice battlefield maneuver skills.

CBS and JCATS do have the advantage of being able to interact with the command and control equipment that staff officers would use in the field. In the exercise at the school, the JCATS students were able to use the equipment they would use in real command vehicles. But Leser said that commercial games are being modified to do the same thing.

Big, highly accurate simulations such as CBS are valuable for in-depth training, Leser emphasized. But they have to be planned well in advance so scarce computer terminals can be booked. Using a PC-based game gives instructors the flexibility to conduct training at any time.