Does digital training encourage trigger-happy cowboys?
April 10 2012 at 1:19 PM
Glen the Rotorhead (Login GlentheRotorhead) HyperScale Forums from IP address 188.8.131.52
A timely article in light of one of my recent opinionated rants at work about how simulation training desensitizes some shooters. Fortunately, not a lot of people have actually taken another persons life. And (hopefully!) even fewer have taken one by mistake. Based on personal experience with both, I am upset and offended by some of the radio traffic I hear accompanying war zone videos, especially when it turns out that the people getting killed were non-combatants and/or unarmed. That is not how I like to picture the American Warfighter.
Also unfortunately, theres no UNDO key in real life.
I guess you have to be registered to read it online, so here it is:
Game on: Does digital training encourage trigger-happy cowboys?
Contributor: Andrew Elwell
In an interview with Defence IQ, Dr Bruce Newsome, Lecturer in International Relations at the University of Pennsylvania specializing in military and security capabilities, offered his insight into the rapidly evolving and increasingly important world of digital and simulated training measures apopted by the military.
Here's what Dr Newsome had to say.
First, all of us must agree that no training method is perfect; I say this because many training methods are set up in competition like a zero sum game, where proponents and critics get worked up respectively defending or attacking a particular method as if the choice must be exclusive, but one of our key recommendations is for a blended approach, using different methods where appropriate, where appropriateness depends on the target skill set and the echelon and sometimes the material context (increasingly, the fiscal context). Having said that, one of our motivations for our research was the lack of systematic attention to the efficacy of digital training methods in particular. We werent the first to say it: digital methods have enjoyed a long honeymoon period in which militaries seemed to assume that digital environments were naturally superior, driven by the popularity of video games, particularly in the demographics from which militaries recruit. Consequently, digital training has tended towards more products based on video games, particularly first-person shooters - the most popular genre of military game.
However, little research has addressed the training validity of digital training products for military users; huge sums have been spent on brand new games, with great training ambitions, usually without any evaluation of their training efficacy or any systematic input from instructional designers. Game design and instructional design are largely antagonistic: game designers need to keep games engaging, for which they allow a-realisms, like multiple lives or notional communications. Instructional designs want to train certain skills, but realistic challenges can be frustrating for users. Traditionally, the main perceived value in training games has been immersive engagement, so procurers have not pushed hard for realism that could interfere with engagement. Consequently, many training games have looked little different from entertainment games.
Digital games are engaging; they are also cheap, at least in the sense that they can easily model scopes and scales that would be impractical in the real world scopes such as prospective weapons and scales such as whole cities, even worlds. Digital environments are high adaptable (the modeller can add or take away a building from the simulated battlefield with ease, compared to the burden of removing or adding a building in a real-world environment). Digital environments are replicable: the scenario can be replayed with ease, while a real-world training environment must be cleaned and reset physically and a real-world scenario is rarely perfectly replicable.
In some ways, digital environments can be more realistic: urban environments are notoriously difficult to model in the real world because of their scale and complexity; consider something as apparently innocuous as urban clutter the mass of signposts, rubbish bins, vehicles, rubble, and unidentifiable detritus that litter a real urban battlefield. This clutter is important for soldiers who must navigate around it or use it as cover its also fundamental to soldiers who must train to identify improvised explosive devices, which often are hidden in such clutter. Yet real-world training environments rarely model such clutter. Digital environments can model clutter easily (although that doesnt mean that they are actually testing soldiers abilities to interact with it in any bio-locomotive way).
While there are clear advantages to digital training, many disadvantages are probably overlooked. For a start, digital environments inherently simplify much, particularly human behaviour. Computer-controlled avatars offer at best some rule-based ability to interact with user-controlled avatars, interactions little more complicated than returning fire or taking cover, but these sort of interactions are not normally considered adaptive. Even moderately adaptive behaviour (such as responding to one type of tactic with the most appropriate counter-tactic) are very burdensome to program and compute.
Digital environments inherently also constrain the users behaviour in unrealistic ways. Game design traditionally has emphasized linear scenarios and levels. While less true of more complex constructive simulations and turn-based games, virtual and real-time simulations tend to be highly linear, stressing fast-twitch responses over force employment. This is worst in first-person shooters the most popular genre. Successful solution of one problem or navigation of one part of the environment will trigger the next. Deliberate constraints (such as blocked alleys, locked gates, unscaleable walls) are built into the digital world in order to force users to follow these linear prescriptions. Users acclimate to linear and reactive behaviours, such as frontal attacks, and are not challenged to find adaptive strategic solutions, such as outflanking.
First-person shooters also constrain the users force employment. Trainees may be denied realistic options or support, forcing the trainee to utilise personal weapons or a limited number of support options in the virtual world even if other options would be preferable and routinely available in the real-world. Most first-person shooters allow the user to control only one avatar, encouraging the user to rely on the avatars resources where teamwork would be more appropriate in real life. Similarly, most games do not allow the user to call up support from armour, artillery, or aerial support; if they do, the command, communication, and coordination challenges are modelled notionally.
Any computer interface encourages counter-productive perceptions, as simple as the narrow focus of a computer screen. Narrowness of focus is a known problem in combat, where the soldier becomes so focused on immediate views that they might neglect peripheral threats. Conventional computer screens do not encourage the soldier to practice attentiveness to peripheries (although virtual reality interfaces help).
Similarly, digital environments do not normally test a soldiers abilities to communicate across distance, over the noise of battle and through the dust and flashes and other distortions. Instead, users communicate reliably by voice with other users sitting in the same room or via some voice-over-internet system.
Finally, digital environments do not exercise anything athletic or bio-locomotive. This is not an issue for physical fitness but for locomotive skills. For instance, in a digital game, a user could command an avatar to open a door, enter a room, reload a weapon, and probably jump up and turn around, all via keyboard strokes, but in the real world such a combination of locomotions, while scanning for enemies and maintaining readiness for lethal force, would take practice.
Some commentators have praised digital games for encouraging willingness to take aggressive action. However, games also encourage rash aggression. Most commercial FPS games do not model weapon effects realistically, at least not in their effects on the users avatar. While enemies may suffer weapon effects realistically, being incapacitated by a hit from a single rifle round or even a nominal punch, the game is usually more forgiving to the users avatar. This helps keep the user playing the game longer, while experiencing frequent exciting dangers. However, unrealistic weapon effects and limited sensory perception encourage the user to practise unrealistic tactical behaviours, emphasizing continual motion, close range, a quick trigger, and lavish expenditures of ammunition. The difficulty of aiming virtual weapons punishes users who keep their avatars stationary and attempt accurate sniping, while rewarding users who keep their avatars in motion, especially since avatars are often capable of unrealistic movements, such as high vertical jumps and sharp changes of direction. Users learn that motion is safer than concealment or stealth and learn aggressive tactics, such as quickly charging enemy avatars rather than seeking a stealthy way to surprise the enemy. Users can add unrealistic tricks, such as throwing grenades to clear the route ahead of hidden enemies, a tactic which in real life would alert the enemy long before the enemy was within grenade-throwing range. Victory goes to the first to react, rather than the most cautious, the best hidden, the most cooperative, or the most tactical. In colloquial terms, most first-person shooters punish the sniper and reward the cowboy.
Were prescribing what we call a blended approach, in which the trainer should match different training methods to different requirements and ideally blend the methods to maximize the advantages across the methods, while minimizing their respective disadvantages.
Blended training is important because it offers an opportunity to combine the best aspects of different methods. For instance, some trainers place trainees in a digital training environment in order to train convoy security, since lethal conflict is much easier to model and replicate digitally than in the real-world. If the digital environment was to produce any friendly casualties, the trainees would be moved to a real-world environment, in which they would respond to a simulated casualty, which offers a more realistic test of practical, manual medical skills than could a digital environment.
Digital training environments are usually less burdensome than real-world training environments - observed primarily as reduced training cost. However, the advantages come with disadvantages and in some cases the disadvantages really might outweigh the advantages. For instance, if soldiers were trained excessively on a simple first-person shooter game, without a reality check in some real-world environment, then those soldiers really would acclimatize to linear problems with linear solutions, simplified weapon effects, undiscriminating cover, unadaptive enemies, etc. Were not saying that the military community is not aware of the balance between advantages and disadvantages, just that the advantages and disadvantages are not routinely identified and assessed, in a manner like a proper risk assessment, before requirement or acquisition.
In most cases, cost always has driven acquisition of digital training environments over real-world environments. Training environments are cheaper in a digital than the real world. Digital environments can easily represent weapons (such as tanks and aircraft) and weapon effects (such as indirect fire) that are fiscally challenging to include routinely in live exercises. Digital environments can be networked, allowing trainees to train together remotely, reducing the costs of moving units to training facilities. Thus, digital training saves money.
Financial savings are particularly pronounced at the highest echelons, such as the divisional level, where live exercises involve many individuals and units, or the heaviest echelons, such as the heavy manoeuvre brigade, where live exercises involve large, expensive weapon systems. For instance, battalion training on the Close Combat Tactical Trainer (CCTT), which includes virtual simulators of the US Armys heaviest weapons, is 25.6 times cheaper than equivalent field training. For small numbers of trainees, lower echelons, and combat intensive training scenarios, commercial-off-the-shelf (COTS) products offer obvious advantages. For instance, the McKenna MOUT Site at Fort Benning, Georgia, costs many millions to operate annually (the site does offer experimentation as well as training services). The Manoeuvre Captains Course at the Infantry School (also at Fort Benning) has used a commercial-off-the-shelf FPS (Delta Force: Black Hawk Down) to help trainees practice their tactical decision-making and procedures such as Troop Leading Procedures and Action on Contact. The game costs the school about $20 per copy.
In austere times and with defence inflation still running about twice as large as general inflation, we naturally expect more drive towards cheaper digital training environments.
Yet we should remember some interesting past cases of acquisition that did not prove cost effective. Also, we can offer some fairly obvious advice for procurers: the cost must be justified against the training efficacy, not just against the cost of other training options. Procurers should use proper risk assessment and should encourage assessment of the likely training utility by instructional designers or should insist on instructional designers within the design team.
A comparison of Full Spectrum Warrior (FSW) to Americas Army, both digital first-person urban combat training games originally developed for the US Army, illustrates the dangers of promoting game design over instructional design. FSW proved an expensive and under-utilized training product. Although originally conceived as a military training game, the US Army Infantry School, which had helped in its development, concluded that FSW was not realistic enough to serve as a training tool. The US Army paid $4.44 million towards development, while the developers agreed to invest $2.5 million. Since the developers went on to make a lot of money from the commercial version, some commentators concluded that the US Army was short-changed.
Meanwhile, the US Army adopted training variants of Americas Army, which had been developed primarily as an entertainment game for recruiting purposes. The game was designed also to teach users about the US Army and some of the knowledge or skills, such as combat first-aid, that recruits would be taught during real-world basic and advanced training. Americas Army was a cheaper product to develop than FSW and is usually distributed to consumers free. Its development was led by the Naval Postgraduate School, with the Army an influential partner. Development of its training applications has been led by instructional designers.
Even free procurement may not be pursued if the operating costs are prohibitive, especially if the product appears weak next to more specialist training products. For instance, the British Army once was offered Americas Army for free, but British procurers ruled that its support costs and inflexibility did not justify acquisition.
Finally, lets not forget how cheap real-world training can be, albeit without some of the advantages of virtual worlds. In the real world, the simplest and lowest cost option is to experientially practice a set of desired behaviours. These sorts of experiences are often termed battle drills, walkthroughs, or dry-runs. These experiences are cheap and accessible, especially if they imagine or notionally represent the fighting environment. Nominal floor plans can be marked out with white mine tape, for instance, in which case the training environment is known as a glass house, a term suggesting one of the main advantages - trainers can see trainees at all times. (A frame house is delineated by see-through wire fences.) Battle drills are particularly valuable at the beginning of training, when more advanced methods would be too risky for under-trained troops, or for practicing risky behaviours or skills that require rigorous trainer attention, such as muzzle discipline or tactical positioning.
This sort of training is usually one-sided; if an enemy is simulated, the enemy is usually used as a trigger a scripted event used to trigger the desired trainee response. For instance, trainees may be tasked to practice a react-to-fire drill in response to simulated enemy fire. Since the simulated enemy is not really competing with the trainees, this sort of training is best described as a one-sided drill and not a multi-sided game. Thence, a digital training environment could be used to provide that multi-sided experience, or a higher-echelon or combined arms experience, without losing the training benefits of the real-world experience again, what we are prescribing is blended training.
Digital training environments are useful for training certain types of skills: visual processing, attention management, fast-twitch responses, fine motor skills, and some higher-order competencies.
Digital training environments (in their current form) are inferior to real-world methods for training the bio-locomotive and non-deliberative adaptive skills that are so important in combat at the lowest echelons (individual, team, and squad at least). Combat skills - urban combat skills in particular - at the lowest echelon are best trained in the real world.
Digital environments are best for training higher-echelon, mounted, extra-urban (because of the longer shooting ranges), and discrete non-adaptive or contextual (such as shoot/dont shoot) fighting skills. Real-world training environments are best for training lower-echelon, dismounted, urban, and adaptive or contextual (such as shoot/take cover) fighting skills, as well as non-fighting (non-kinetic) and higher-order competencies, such as managing coalition partners and stakeholders during counterinsurgencies and stabilization operations.
Real-world or live environments are best for training the automatic or naturalistic urban fighting skills required at lower echelons (individual, team, and squad levels, at least, and perhaps the platoon level). The best available training method for training such skills would use a real-world training environment, a real-world OPFOR, and simunitions-type equipment. (One-sided target shooting with lethal rounds should be practiced too). Bio-locomotive skills (such as weapon-handling and motile skills) are not practiced in most digital environments, although some postures can be practiced in virtual-reality environments. Moreover, real-world OPFOR behaviour is more flexible than automated behaviour and is thus better for training adaptive decision-making.
Real-world environments should be encouraged for training dismounted combat skills at lower echelons. Except at higher echelons (we suggest company command or above) dismounted skills are too physical and contextual to warrant virtual training and they also depend heavily on adaptive enemies if the training is to be positive. The sort of dismounted skills that are useful to practice virtually are easy to practice in real-world environments anyway. Digital simulators sometimes disallow human-controlled dismounted avatars. Real-world dismounted urban combat training should use simunitions. For non-urban environments, which involve combat at longer ranges, trainers should use a Multiple Integrated Laser Engagement System (MILES; Tactical Engagement Simulator System, TESS), while noting that such a system does not model ballistics or cover realistically. Digital environments should be encouraged for training mounted combat skills. Digital environments are better for mounted training for two main reasons. First, training within simulated mounted stations is much cheaper than real-world mounted training. Real-world mounted training of the heavy manoeuvre brigade is the Armys most expensive type of training. Second, mounted skills are less bio-locomotive than dismounted skills.
Digital environments should be encouraged for training commanders (at least of company-level and perhaps as high as battalion level) in force employment skills, since many of their decision-making problems are more deliberative and less bio-locomotive than the individual fighting skills required from their followers. (Higher echelons are probably not practical for training in FPS environments; turn-based strategy games are a better choice). While good for training force employment skills, digital trainers do not inform the leader about their individual physical performance or their real-world communication challenges or even of their leadership skills (meaning the inter-personal or emotional leadership skills considered outside of command skills). Blended training or complementary psychological training or assessment could offer solutions here.
FPS environments are appropriate for training non-kinetic skills, such as negotiating and adaptive thinking skills, at the lowest-echelon (individual and team), even though (kinetic) combat skills at this echelon are best trained in real-world environments.
FPS environments are useful for training discrete non-adaptive or contextual automatic cognitive skills at the individual level, such as the shoot/dont shoot cognitive switch, even though they are not as useful for training dismounted skills at the same level. This is because digital training can easily isolate discrete cognitive tasks in a replicable way and without the need for bio-locomotive skills (beyond pressing a button) or an adaptive OPFOR.
More deliberative digital games, especially turn-based strategy games, should be encouraged for training higher-echelon (at least brigade and higher) command and force employment skills. Turn-based strategy games are useful also for higher-order competencies, such as the political, economic, and sociological skills useful in stability operations. Real-time strategy games are probably not useful for military training, except perhaps for training force employment skills at higher echelons with realistic time delays for the communication and execution of orders, since they are neither as deliberative as turn-based strategy games nor as naturalistic as first-person shooters.
Digital environments should be encouraged for mission rehearsal exercises, for practicing drills and standard operating procedures and working relationships, and for experimenting with force employment. The replicability and editability of digital scenarios make virtual environments particularly appropriate for practicing procedures, especially highly predictable, deliberative procedures for which adaptable behaviour may be unnecessary.
Long hyped technologies are now offering maturity that would be of use to the trainer. Take virtual reality: I noted earlier that virtual reality interfaces encourage the user to check their peripheries, whereas a normal computer screen encourages narrowness of focus.
Perambulation technologies, where the user interacts with the digital simulation by moving about a flat interface on the floor, would exercise pedestrian bio-locomotive skills but probably not cost-effectively today. Most militaries have long prioritized the immersiveness meaning story-based scenarios with high physical (usually graphical) fidelity - of its digital training environments. Traditionally, new entertainment and game technologies, such as perambulation, have been evaluated in terms of their effect on immersiveness, rather than on training, but perambulation should be evaluated primarily in terms of its effect on a soldiers bio-locomotive skills, such as negotiating urban clutter when entering a room while remaining ready to fight.
Procurers today should demand more from old technologies, for instance, interfaces that more realistically model real weapon interfaces and controls; perhaps procurers should be looking for technologies that simply plug into the trainees real weapons so that they can use their weapons the same way in a digital environment as they would in the real world.
Wed like to see more attention to what we call combat fidelity, particularly realistic modelling of OPFOR, so that trainees face genuinely adaptive enemies inside digital environments this is receiving plenty of attention in some quarters, but the fundamental need is not appreciated universally. Computers probably never will model human behaviour realistically enough to supplant real human OPFOR, so human-controlled OPFORs must persist, but automated OPFORs could save a lot of money and enhance training replicability and adaptability.
Wed like to see more nuanced attention to what the gaming industry calls physical fidelity essentially the accurate visual representation of the real world. Physical fidelity is important if it has training impact. For instance, urban clutter should be modelled in as much as its complexity as possible because soldiers need to learn how to fight through urban clutter, navigate around it, understand how it could be weaponized, and understand how different materials offer different levels of cover and resistance to enemy fire. Ironically, the quest for graphical fidelity and immersiveness through graphical fidelity has neglected these sorts of needs. Some game theorists imply that immersiveness, at least as suspension of disbelief, is not necessary to games, since games are played within special environments anyway, which players recognize as separate to the real world. The freedom of open digital environments has been shown to distract trainees from their instructional tasks. In retrospect, while a good story and physical fidelity are certainly engaging, they may interfere with positive training by, for instance, priming soldiers with inaccurate cues.
In some cases, physical fidelity is not necessary. Take the simple problem of training soldiers to shoot real enemies but not shoot illegitimate targets. Since the cognitive task is discrete and instinctive, physical fidelity can be low, even abstract. For instance, legitimate or illegitimate targets could be represented by differently coloured or shaped digital graphics these abstractions are used on real-world ranges. The sort of physical fidelity used in digital games is abstract too, however immersive they may feel for instance, a digital enemy might be modelled on the screen as an avatar that moves, waves a weapon, curses, etc., but the modelling always includes simplified graphical cues (some commentators have noted that many enemies, across many games, are modelled with red scarves and swarthy features) to which the user will become primed, when we want them to become cued to the behaviours that indicate when an ambiguous actor transitions from non-threatening to threatening.
In summary, we urge designers and procurers to aim for combat fidelity rather than graphical fidelity. War gamers have complained about the trend, during the digital age, away from the serious modelling of combat (combat fidelity) towards graphical fidelity. Were not saying that graphical fidelity isnt important, but its time to leverage that technology towards modelling things that really test the soldiers skills rather than just make them feel immersed and engaged.
Bruce Newsome is now Lecturer in International Relations at the University of Pennsylvania, specializing in military and security capabilities. These answers are based on research carried out by Bruce Newsome and Matthew W. Lewis, while they were working at the RAND Corporation, an independent policy research institute in Santa Monica, California; the research was sponsored by RAND as an independent research project; their views do not necessarily represent the views of RAND, the US Army, the British Army, or any other organization that might have been mentioned here.