The Military Metaverse: Part 2

Following the September 11 terrorist attacks the US launched haphazard invasions of Afghanistan and Iraq – expecting to receive a liberator’s welcome. Instead indiscriminate bombing campaigns, economic shock therapy (as applied to Iraq) and the heavy-handed tactics of the occupation forces outraged citizens of both countries and set the scene for the decades-long insurgencies that followed. 

Simulation technology would play a part in how the US military responded to this new predicament and, once again, they looked to the entertainment industry for inspiration.   During this period a schism was occurring in the world of action computer games and first-person shooter games were beginning to split off into two distinct genres. The most popular of these took their cue from those early carnival shooting gallery games. Doom and its descendents were designed as escapist power fantasies. Using a keyboard or controller these games invited players to mow down enemies by the dozen while letting them shrug off incoming bullets like the Terminator. Whether they were set in a gothic fortress or a remote space station or some maze-like industrial site the common denominator was that you were a demigod while your opponents were mere mortals. Within this genre the appeal of recreating war – especially the Second World War – was seized on very early on. Games like Medal of Honor and Call of Duty attempted to convey the thrill of leading the charge in a large-scale battle for a righteous cause. 

Within these games battlefield injuries could be alleviated simply with a first aid kit or some other magical potion. Ubisoft’s FarCry series later emphasised the cartoonish nature of this mechanic by having the player’s avatar pull bits of shrapnel out of their arms or cauterise inconvenient bullet wounds with bundles of lit matches. In recent years, games like Modern Warfare have discarded these sorts of contrivances entirely. Instead your character simply recovers from any injuries as long as they remain out of harm’s way for a few seconds. 

While these games were surging in popularity another sub-genre of first-person shooters was quietly amassing a dedicated fan base. These games were marketed as simulations of modern war. Instead of frictionless, ice-skating movement these games attempted to impose some of the limitations and risks of infantry combat. Within this new genre – sometimes called ‘tactical shooters’ or ‘milsims’ – players were obliged to move at a roughly human pace and make use of cover and concealment within the game world. Typically the player’s avatar was able to crouch and crawl and could only carry as much as his real-world counterpart. Moving too fast for too long resulted in a virtual asthma attack – throwing off your aim. Rather than allowing players to blast away with impunity at anything on the screen, milsims forced the player to ration their ammunition, take aim using gun-sights and compensate for factors like recoil and bullet drop.  

In his study Simulating War, researcher Philip Sabin points out that, despite many features aimed at promoting ‘realism’, the end result is often a far cry from what occurs on the battlefield.

“In games ranging from Rome: Total War to the flight simulator Il2 and [modern combat sims], melées, firefights or dogfights end after just a few minutes of frenzied carnage, with all of one side and most of the other side dead. This is not because the weapons modelling is wrong, but because the combatants (even if controlled directly by humans) have nothing like the incentive for caution and self-preservation that they would feel if the blades, bullets and shell splinters were real.”

The treatment of in-game wounds and injuries is perhaps the most noticeable difference between escapist action games and would-be simulations. Rather than magical first aid kits the more self-serious milsims favoured a ‘realistic’ game mechanic whereby battlefield wounds produce tedious restrictions on your avatar’s movement or vision – until a friendly medic heals you a-la Jesus or further damage forces you to repeat the level. In some cases, one unlucky shot is enough to kick you back to the starting menu. Falling into this category were games like Ubisoft’s Ghost Recon series (licensed under the Tom Clancy militainment brand) and Sierra’s SWAT series.

Operation Flashpoint

In an attempt to create the ultimate milsim a game development studio split between Australia and the Czech Republic became the unlikely heir to the Pentagon’s SIMNET. In the late 1990s Czech game developers Marek and Ondřej Španěl started Bohemia Interactive (BI). BI’s inaugural title was a milsim called Operation Flashpoint: Cold War Crisis. Plagued by delays due to technical overhauls the game was finally released in mid 2001. Although the graphics were crude, even for the era, the scope of the game was ambitious in the extreme – allowing players to traverse an open world environment on foot or hop in and out of a range of real-world military vehicles and aircraft. While most other games of the era confined you to a maze of interchangeable corridors, Operation Flashpoint dropped you on an island 12km squared with no fences or invisible walls. 

In many ways Operation Flashpoint was still rough around the edges when it was released. The scenery was very low-resolution and the in-game physics were janky – shot enemies collapsed after a slight delay as if they’d missed their cue while vehicles often twitched and bounced around on the ground like jumping castles in the wind. Yet, despite these quirks, Operation Flashpoint was widely acclaimed for its scale and for its commitment to a somewhat unforgiving form of gameplay.

Set on a fictional Soviet archipelago somewhere in the Barents Sea, the game’s single player campaign initially placed you in the combat boots of a U.S. infantryman, before allowing you to roleplay as a tank commander, a Special Forces ‘Operator’ and a helicopter pilot. For the Španěl brothers, who had spent their childhoods under the waning Soviet regime, the game’s alternate history of 1985 – where a plucky bunch of US peacekeepers and local insurgents triumph over the Red Army – was a form of digital catharsis.

In terms of its features and game mechanics Operation Flashpoint was years ahead of its time. Somehow BI had managed to create an engine capable of simulating trucks, armoured vehicles, helicopters, jets, patrol boats and even tractors. Instead of trying to reproduce the atmosphere of a Hollywood war movie Operation Flashpoint gave players a giant sandbox with which to play out different scenarios. Using the game’s built-in level editor players could drop entire squads or platoons onto the topographic map, give them waypoints and watch them awkwardly make their way, in fits and starts, across the landscape. By adjusting a few simple sliders players could also alter the time of day, the weather and the overall visibility. SIMNET was now available to the public.

Right at the outset one of BI’s developers, David Lagettie, recognised the potential of the game to serve as a general-purpose simulator for the armed forces. Most NATO military forces already used expensive and elaborate simulation tools for training recruits in the use of particular vehicles and aircraft but the game engine from Operation Flashpoint offered something unique – the opportunity to rehearse what the military refers to as ‘combined arms’ operations (eg. infantry + artillery or tanks + aircraft). In 2001 Lagettie started an offshoot development studio from his home in Newcastle. Bohemia Interactive Australia (later to become BI Simulations) began working on a more ‘serious’ version of the Operation Flashpoint game.

At the same time DARPA partnered with Raytheon’s BBN Technologies to create their own modified version of Operation Flashpoint in order to train soldiers destined for Iraq.  In May 2003 U.S. president George Bush famously declared ‘mission accomplished’ in a televised address from the deck of the USS Abraham Lincoln but within a few months of this announcement the scattered resistance to the U.S. occupation had begun to coalesce into a nationwide insurgency. Casualties from IED attacks began to mount and, by the end of the year, an average of one U.S. soldier was being killed (with half a dozen others wounded) every day. For coalition troops tasked with conducting patrols and ferrying supplies between the various U.S. bases, better training and situational awareness was urgently needed to stem the rising death toll. The U.S. Army decided that a driving sim which could be run on a laptop or PC and adjusted, on the fly, to produce different scenarios would be part of the solution.

Thanks to Bohemia’s engine, the final product, known as DARWARS Ambush! took only a few months to deliver and cost the U.S. Army a mere $1.3 million USD to develop – a very expensive mod but a very cheap simulator and a drop in the ocean of the U.S. Army’s training and simulation budget. 

Ambush! placed trainee soldiers in a virtual convoy in a Iraqi-esque gameworld and had them travel from point-A to point-B. Most of the scenarios scripted in Ambush! were much more boring versions of the typical Operation Flashpoint missions (which were, themselves, much more boring versions of typical arcade shooters). Participants were required to maintain vigilance for long periods of time while instructors watched and recorded their responses to sudden IED or sniper attacks. The lessons being conveyed were simple; look out for suspicious vehicles, don’t bunch up, don’t get too far apart, keep in contact with one another and be careful about who and what you shoot at.

At the conclusion of each training session the game would provide a summary of the trainee’s performance – including the number of rounds fired (and the resulting ‘hit ratio’) as well as the number of enemies, friendlies and civilians killed or wounded. In a summary of the project the technical lead on Ambush!, David Diller, noted that some features of the underlying game caused problems for the officers facilitating the training program.

“One curious example was that in Operation Flashpoint, a player’s character would turn into a seagull after they died. This would allow them to fly around and see what had happened around them. We initially left this very game-like feature in DARWARS Ambush!. However, as we tested the game, we found it would cause problems on occasion because the players would do recon for their buddies after they were dead and report enemy locations using the VoIP capability.“

Aside from the pace of gameplay the key difference between commercial games and their ‘serious’ sim versions was the addition of formal after-action reviews wherein participants would re-watch their performance with trained instructors, evaluate how they’d responded and determine what lessons could be learned. Between 2004 and 2009, more than four thousand copies of Ambush! were distributed to military bases across Europe, Iraq, and Afghanistan. Ultimately more than thirty-five thousand soldiers from the U.S. Army, Air Force, and Marines were trained using the game.

Other training experiments utilised ‘off the shelf’ computer games. In 2004 Pandemic Studios negotiated with the Defence Department’s Institute for Creative Technologies to build two variants of their upcoming game – one would be licensed to the military as an infantry training aid while the other would be stripped of the more instructional elements and sold to the general public as a typical action shooter.

Much of the marketing for Pandemic Studio’s Full Spectrum Warrior centred on the claim that the game was ‘used by the U.S. Army’ but, much like Mech War and Marine Doom, its main use was probably to stave off boredom. To judge from reviews at the time this was a task that Full Spectrum Warrior was only partially suited to. 

Nevertheless one particularly patriotic developer at Pandemic Studios resigned in protest over his belief that the training version had been neglected in favour of the commercial product. The developer’s complaints were echoed by Lt. Col. James Riley, who tested Full Spectrum Warrior at the infantry school at Fort Benning. He reported that his soldiers did not learn the intended lessons and were disappointed by the game’s lack of realism. “It’s a neat game” he later told reporters “but I’m not seeing where I can train with it.”

The Rise of Virtual Battlespace

Having glimpsed the potential of a versatile open-world simulated environment the U.S. military immediately began looking for what they called their Game after Ambush. They found it in Lagettie’s Virtual Battlespace (VBS). VBS combined the same basic game engine with a raft of new features and a growing suite of vehicles and equipment, modelled on real military hardware. Now headed by former ADF signals officer Peter Morrison the company has expanded its offering beyond Australia and the U.S. and now supplies versions of VBS to more than 50 (mainly NATO) armed forces.

Much like the Waller Gunnery Trainer the most elaborate implementations for VBS involve projecting 180 degree vistas on giant half-domed screens lit by multiple data projectors. Participants stand in the centre and use mock versions of anti-tank weaponry and laser designators to practise targeting without the expense and inconvenience of hauling everything into the bush. 

Over the last twenty years a great deal of cross-pollination has occurred between BISim and the original Bohemia Interactive studio. Features and engine upgrades undertaken by one company were shared with the other and developers flitted between the two studios. When BI launched ARMA II they nominated a small, somewhat rustic, portion of the Czech Republic in the north of the country to use as the game-world. In their fictional ‘Chernarus’ a winding stretch of the river Elbe was turned into a coastline while the landscape adjacent to the 225 km² battlefield was filled in with ‘procedurally generated’ terrain so that errant fighter jets and helicopters wouldn’t collide with an invisible wall if they travelled too far afield.

Chernarus would also become the setting for the spin-off zombie survival shooter Day-Z which had started out as an ARMA mod created by New Zealand army veteran Dean Hall. In 2016, when the cash-strapped Swedish army purchased a licence for Virtual Battlespace, they requested an ‘off the shelf’ environment to stand in for their own patch of Scandinavia. Through their sister studio, BISim offered them Day-Z’s Chernarus. In a blog post on BISim’s website Julian Saenger – one of the company’s developers – explained that some final adjustments were required to make it fit for purpose.

“The biggest effort was cleaning up Chernarus post-apocalypse; that is, de-zombifying the terrain. [The art and content team] were the guys who came into the zombie apocalypse and removed the bodies, blood stains and broken windows…Because it’s a fictional terrain and our customers use real-world data with our coordinate system, we couldn’t put it in an existing location like the middle of Africa, so we georeferenced the terrain in the Black Sea.”

Given the secrecy surrounding the exact specifications and capabilities of current military equipment one might assume that only military sims like VBS could be counted on to accurately reflect the capabilities of military weapons and equipment. But as some recent scandals surrounding the game War Thunder have revealed – even commercial milsims receive a substantial amount of informal input from military experts.

Created by Hungarian game development studio Gaijin Entertainment, War Thunder started out as an arcade-style combat flight simulator but gradually expanded to include tanks and ships. As well as offering simulated versions of historical military vehicles the game also features machinery currently employed by armed forces around the world. Predictably this has prompted a lot of vigorous debate in the War Thunder online forums over which country has the ‘best’ tank and whether the in-game versions are accurately modelled. Given the international scope of the game, the longstanding national rivalries within Europe and the pedantry typical of the average milsim gamer, War Thunder almost seems designed to provoke off-duty soldiers into revealing military secrets – which is exactly what happens on a fairly regular basis.

In one recent incident the self-described commander of a French Leclerc Main Battle Tank posted excerpts from his classified gunner’s manual in order to win an argument over the tank’s turret rotation speed. The manual (uploaded with the filename ‘Sekrit Document’) was quickly removed by the forum’s admins but not before prompting a vigorous debate about whether the developers should incorporate classified specifications. That incident followed another leak on the War Thunder forums wherein a player claiming to be a former member of the British Army’s Armoured Trials and Development Unit posted a copy of the manual for the Challenger 2 tank in order to win an argument about the size of the gap between the tank’s turret and hull.

While online arguments over turret dimensions and traverse rates have tripped up soldiers further down the autism spectrum, the temptation to brag about gear has also landed the U.S. military’s uber-jocks in hot water. In 2012, seven members of the U.S. Navy’s elite SEAL Team Six were formally disciplined for revealing classified information while consulting on EA’s Medal of Honor: Warfighter game. This minor scandal represented the best possible PR for a title which proved too on-the-nose for even the most jingoistic gamers. In no uncertain terms PC Gamer called Warfighter a “boring, unoriginal, morally bankrupt, ethically dubious glorification of war”. The virtual weaponry, on the other hand, was presumably spot-on.

When it comes to preparing soldiers for real battlefields the efficacy of simulated environments like VBS remains an open question. A report by Australia’s Defence Science and Technology Organisation in 2013 noted that previous research into the effectiveness of infantry simulations on troop performance had been hampered by methodological limitations and an overall lack of rigour. Few previous studies maintained a control group or established a baseline against which to compare results and most never bothered to examine whether the skills acquired within the simulation actually translated to real-world improvements in performance. The paper notes that:

“As a result of these shortcomings, there are few conclusions that can reliably be drawn from these studies; at best it can be concluded that game-based training does not appear to cause any negative learning effects on dismounted soldiers.”

When it came to basic infantry tactics the DSTO’s own experiment showed that VBS was a poor substitute for real life exercises in the field. DSTO researchers ultimately concluded that the simulated training had no appreciable impact on the skills displayed by their experimental group. While VBS’s proponents would no doubt argue that systems like theirs were designed to complement, rather than substitute, for practical exercises ‘possibly better than nothing’ doesn’t exactly make for a compelling marketing slogan. Despite the DSTO’s ambivalence towards VBS (at least when it came to infantry training) the report granted that it was “possible that such computer games might be effective for training other military tasks, such as mission rehearsal and terrain familiarisation”.

Just like the Marines who’d modified Doom the DSTO recognised that the real value of systems like VBS lay in simulating an actual location – whether that be a single building, a district within a larger city or an entire geographic region. As Diller observed in his chapter of Design and Development of Training Games, whenever Ambush! was demonstrated for field commanders in Iraq their first question was always the same “Can you build my area of operations to train on?”.

Recreating entire suburbs of Baghdad and Mosul was a tall order but the gauntlet had been thrown down and various sim developers had already begun exploring the possibilities. BISim began by recreating various military training grounds in Europe, the U.S. and Australia. With a digital Puckapunyal saved to its servers the ADF could run exercises that combined virtual and real world units – an arrangement known in the acronym-laden military world as Live, Virtual & Constructive (LVC) Simulations.

One of the first, and most comprehensive, large-scale LVC exercises took place in 2002 and ended up resembling the infamous IJN wargaming session attested to by Mitsuo Fuchida. Involving more than 13,000 personnel scattered across the Pacific ocean and more than a dozen simulation locations, the Millenium Challenge was intended to wargame an amphibious invasion of Iran via the Persian Gulf by a massive U.S. Naval armada. The opposing force was commanded by retired Lieutenant General Paul Van Riper – a 64 year old Vietnam veteran who, at one point, had been the Director of Intelligence for the U.S. Marine Corps. 

When the exercise commenced Van Riper unleashed a barrage of missiles from ground-based launchers and disguised commercial ships and aircraft rather than waiting for the U.S. taskforce to make the first move. At the same time he ordered swarms of speedboats to carry out kamikaze attacks on the carrier group – resulting in the immediate ‘sinking’ of 19 U.S. warships.

Stunned military officials suspended the exercise before deciding to restart the whole scenario under more favourable conditions. ‘Wrecked’ U.S. ships were re-floated and the Rules of Engagement were amended to prevent Van Riper’s underhanded tactics. When pressed by Defence Secretary Donald Rumsfeld to explain why his armada needed a second life General Peter Pace replied: “You kill me in the first day and I sit there for the next 13 days doing nothing, or you put me back to life and you get 13 more days’ worth of experiment out of me. Which is a better way to do it?”

Following the reset both sides were ordered to follow a predetermined script. U.S. aircraft were declared off-limits and Van Riper’s hidden missile batteries were ordered to be placed in the open so that they could be picked off with impunity. Van Riper sat out the exercise in protest – later calling it a ‘sham intended to prove what they wanted to prove.’

Back in Prague, Bohemia Interactive were also taking the real-world approach for their commercial games. To set the scene for the third iteration of ARMA, the studio decided to use the Greek island of Lemnos as the template for their next fictional flashpoint. The small Aegean island was intended to serve as the site of a near-future confrontation between the U.S. military and an implausibly globe-trotting Iranian army. Demonstrating a bewildering lack of tact, several BI staff travelled to Lemnos to record real scenery to inform their digital game world – including, allegedly, a Greek military airbase. When two developers returned to the island in September 2012 they were arrested by Greek authorities and charged with espionage – prompting a diplomatic stoush between the Czech and Greek governments that persisted for several months until the two men were finally released. Bohemia later grudgingly changed the name of their fictional island to avoid “undesired real-life connotations”.

The Synthetic Training Environment 

In recent years BISim has begun to take the concept of ‘terrain familiarisation’ to its inevitable, totalising conclusion. As part of the U.S. military’s ambitious Synthetic Training Environment (STE) initiative BISim’s developers are working on programs that can recreate 3D versions of any region on the face of the earth. With only a few hours notice, military commanders can nominate a set of coordinates and automatically generate a 3D environment that roughly matches that area. 

Just as advances in lithography and cartography in the early 19th century allowed von Reisswitz to add realism to his kriegsspiel scenarios, the proliferation of high-res satellite and aerial photography will soon allow the U.S. to three-dimensionally wargame their next catastrophic intervention anywhere on the planet. In a promotional video released by the Institute for Creative Technologies in 2019 General Steven Townsend outlined the intent of the STE program.

“STE will immerse soldiers and units in future operational environments that replicate the actual terrain they will operate in before leaving home station. Our soldiers will become virtual veterans of 25 bloodless battles before the first round is ever fired in combat”

If war, as journalist Ambrose Bierce famously quipped, is ‘God’s way of teaching Americans geography’ then BISim’s Synthetic Training Environment represents a stepped change in how the U.S. government prepares for war. It also represents a worrying trend towards relying on satellite and geospatial data to inform soldiers about what they’ll encounter when deployed to foreign countries and ongoing conflicts.

In a demonstration of the latest version of VBS a small rural town in Mozambique called Palma was chosen to showcase the unfortunately-named TerraSim terrain-generation technology. From a godlike perspective somewhere at the edge of space the video zooms down onto the African coastline and the satellite image flickers and resolves as the green patches take on a slightly different hue. TerraSim’s ​​Technical Director, Earl Laamanen, explains that ‘this part of Africa is pre-configured to have a certain type of vegetation…all I have to do is say ‘this area should be forest’ and [TerraSim] takes care of the detail’. 

From a distance the visuals are certainly impressive. BISim’s TerraSim engine generates 3D terrain based on high resolution satellite data – allowing it to infer the position of rivers, forests, roads and pastures. By measuring the relative height differences between objects and their surroundings it can even recreate bridges and overpasses.

As the perspective dives closer to the ground we see individual trees and scrubland start to resolve. Laamanen turns on further settings and small prefab houses with corrugated iron roofs pop into existence where previously there was only a flat, two-dimensional image. Buildings, Laamanen explains, are ‘automatically and randomly assigned a certain type of texture’ (‘texture’, in this context, refers to any image overlayed onto a 3D object). To finish off the digital version of Palma utility poles and streetlights are randomly generated along the sides of roads. To show off the day/night cycle Laamanen fast-forwards to the evening and a gloomy version of Palma slowly fades into view as the engine simulates the adjustment of the human eye to the sudden change in light.

But even after buildings and roads have been conjured into existence the net effect is somewhat sterile and empty. The shortcomings of the engine are revealed when it comes to rendering anything smaller than a house. To supplement the satellite data TerraSim can incorporate drone-collected 3D measurements – using a whole host of scanners and sensors that fall under the umbrella of ‘photogrammetry’. But as Peter Morssion, BISim’s CEO, explains.

“Photogrammetry is good at 500 feet but at ground level everything looks like melted wax. We’re working on fixing that”

In the meantime the TerraSim engine comes with a menu of dollhouse-like objects that can be manually positioned within the virtual world – piles of rubble, burnt-out vehicles, 55-gallon drums, empty market stalls, and portable generators can all be mixed and matched to produce the right third-world mise-en-scène. 

Employing automated systems to reconstruct real-world locations is both the main challenge and the main attraction of BISim’s take on the Synthetic Training Environment. Manually recreating an accurate 1-to-1 three-dimensional version of a real location would be incredibly time consuming and finicky whereas combining a GIS data with satellite images and letting the computer fill in the blanks is relatively easy (although designing the program to do this is quite an achievement). 

But the devil is in the details and the decision to trade fidelity for speed has ensured that lots of real-world considerations get lost in translation. The end result is more like a vague memory of a place. Fences and walls are missing or in the wrong place, two small trees have somehow become one larger one and a children’s playground is reduced to an empty field. Similarly, all the garbage-in/garbage-out problems typical of machine learning systems suddenly assume much higher stakes when the task is to turn someone else’s neighbourhood into a firing range. It’s easy to imagine what could go wrong. An algorithm trained on suburban backyards in North America is liable to mistake a water pump for a letterbox or turn a minaret into a phone tower. 

The near-enough-is-good-enough approach to simulating real locations also risks misleading its intended users. When the U.S. Marine Corps considered using Doom to rehearse a hostage rescue at a U.S embassy they counted on being able to recreate the exact floorplan of a specific building. The assumed benefit lay in being able to conduct a virtual walk-through to get a sense of the space, determine lines of sight and know, in advance, what they could expect to find behind each door. But had they been offered a version of the simulation in which the layout of their target building was only 85% accurate with the other 15% being ‘procedurally generated’ based on ‘typical’ building layouts they might have had second thoughts.

But the real problem with the Simulated Training Environment is that it’s a tool designed for a form of warfare that no major power can afford to wage. Ever since Hiroshima and Nagaskai were razed by nuclear weapons in 1945 the prospect of war between modern, nuclear-armed states has been too terrifying for most political leaders to even contemplate. And yet that is exactly what many NATO militaries spend their time doing – imagining and simulating ‘conventional’ wars of manoeuvre and artillery with vaguely defined peer opponents. Virtual Battlespace is ideally suited to this sort of wargaming. At its heart VBS is simply a much more sophisticated version of SIMNET which, even for its time, represented a nostalgic way of looking at war.

As military theorist Martin Van Creveld argued in his book The Transformation of War, the development of nuclear weapons by several nation states in the second half of the 20th century has had the unintended consequence of making their conventional forces somewhat redundant. Because no nation can afford to provoke a nuclear exchange any government that wants to apply military pressure to a nuclear-armed adversary is forced to do so indirectly – by sponsoring rebel groups, carrying out covert operations and by waging ‘economic warfare’ through tariffs and sanctions. Throughout the Cold War dozens of these proxy wars raged across Africa, South America, the Middle East and South-East Asia. Most notably the Soviet Union funnelled weapons, aircraft and equipment to the North Vietnamese Army to fight French and U.S. forces in Indochina while the U.S. armed and trained Afghan militias to contest the Soviet occupation of their country. As Van Creveld summarised in his chapter on ‘contemporary war’:

“Thus the effect of nuclear weapons, unforeseen and perhaps unforeseeable, has been to push conventional war into the nooks and crannies of the international system”

Admittedly the war currently raging in Ukraine looks like a fairly large ‘nook’ but the general trend that Creveld identified still holds true. Over the last 20 years western militaries have mainly been tasked with ‘peacekeeping’ operations, training foreign allies, carrying out disaster relief work and patrolling maritime trade routes.

The current fixation on recreating specific places also appears misguided in light of modern military capabilities. When Prussian officers were attempting to perfect battlefield manoeuvres in their kriegsspiel the emphasis on geography made sense. They were preparing to fight conventional armies, much like their own, in set-piece battles. Terrain was a genuine obstacle and many of the lookup tables painstakingly compiled by Reisswitz dealt with the time required for armies to navigate forests, fields and marshland and traverse obstacles like rivers and mountains. But, for the member-states of NATO, the era of agonising over daily march rates and baggage train capacities ended definitively sometime in the 1960s to an overture of Creedence’s Fortunate Son.

Starting with the invasion of Vietnam the U.S. and its Western allies have strived to establish ‘air superiority’ over any region they decide to turn into a battlefield. In 1982 the British air force demonstrated its reach by flying Vulcan bombers more than twelve thousand kilometers from Ascension Island to bomb an airbase on the Falkland Islands – each mission required almost a dozen aerial tankers and a relay of fourteen separate air-to-air refuels. In the Persian Gulf War the U.S. Air Force flew thousands of sorties which succeeded in destroying or scattering the entire Iraqi Air Force before they deployed any ground troops. Likewise NATO’s interventions in Bosnia and Libya have been predicated on their ability to designate ‘no fly zones’ and carry out bombing raids with impunity.

This emphasis on achieving and maintaining air superiority has fundamentally transformed how high-tech militaries approach war. Decades of investment in helicopters, aircraft carriers, aerial tankers and transport planes have helped ensure that geography no longer imposes hard limits on military ambitions. 

Nowadays small groups of soldiers can be flown hundreds of kilometres and dropped off at precise GPS coordinates in a matter of hours. The return journey is usually just as efficient. U.S. soldiers wounded in Iraq were routinely triaged, stabilised at trauma centres in U.S bases before being whisked away to military hospitals in Germany in under six hours. The world has shrunk yet war planners insist on squeezing greater and greater detail from their maps.

Needless to say the military organisations that can afford to purchase licences for systems like VBS and bankroll research and design into virtual environments aren’t suffering from a lack of geographic knowledge. Neither are they threatened by the prospect of all-out war with a ‘peer’ opponent. Instead they’re increasingly faced with the challenges posed by unconventional warfare carried out by terrorists, insurgents and paramilitary criminal networks. 

As one retired US Army officer pointed out in a surprisingly poetic article for the US Army War College the military has always been a grudging participant in what they refer to as ‘military operations other than war’. And because these operations have traditionally been treated as a side show, the training, equipment and doctrine required to make them successful has typically been neglected.

“We frequently will be unprepared for [peacekeeping missions], partly because of the sudden force of circumstance but also because our military is determined to be unprepared for missions it does not want, as if the lack of preparedness might prevent our going. We are like children who refuse to get dressed for school.”

These ‘low-intensity’ conflicts are socially, rather than geographically, complex and they demand that soldiers demonstrate a certain level of discretion and tact rather than relying on overwhelming firepower. Simulations might help armies train for these missions but, so far, very few of these ‘serious games’ have attempted to simulate civilian behaviour or tried to represent the communities they belong to. In the early 2000s the ADF produced games designed to familiarise troops with cultural moores in East Timor and Afghanistan but these can be more readily compared to the sort of low-fi ‘eLearning’ modules that plague the corporate world.

The decision to use Palma as a test-case for the STE only emphasises the literal and figurative emptiness of their simulated world. Seventy five thousand people live in the actual district of Palma (or did up until the most recent outbreak of violence) which has been at the epicentre of an anti-government insurgency that extends into neighbouring Tanzania. In 2021 militants from a group Ahlu Sunna Wal Jammah carried out a raid on the town in an attempt to disrupt construction of a Liquified Natural Gas plant operated by the French energy giant Total. 

To make room for the plant and the adjoining company-town, thousands of local inhabitants were evicted from their land with only token compensation. Many were forced to relocate inland away from the fishing industry which represents the mainstay of economic activity in the region. The wider province of Cabo Delgado remains Mozambique’s poorest region despite its natural resources. Thanks to long-running government discrimination towards the country’s muslim minority, unemployment in the region is high, particularly among the youth, and civil services are either patchy or non-existent. Rather than attempting to address the concerns of local residents Total employed South African mercenaries to provide security for their staff and operations.

Most of the militants recruited into Ahlu Sunna Wal Jammah belong to the marginalised Mwani ethnic group but others come from a range of nearby countries and a number of other ethnic and language groups. Most of those who join are illiterate and, despite claimed associations with international Islamic extremist groups like ISIS and Al-Shabab, few members appear to have received any real religious instruction. Instead religious convictions appear to have been used as justification for a spree of looting, murder and arson aimed at undermining government control and disrupting resource extraction by foreign powers. Whatever the group’s political goals, many fighters appear to have been drawn to the group simply by the promise of stolen goods or some sort of monthly wage.

But in the digital version of Palma there is no community. There are no tribal affiliations and sectarian grievances, no coercive foreign business interests or corrupt local power brokers, no commerce or agriculture to be disrupted, no black markets or narcotics trade, no refugees or health clinics or groups of displaced families or unemployed young men. In short there is nothing in BISim’s virtual Palma that would help outsiders make sense of the wider conflict. Instead foreign soldiers deployed to the region will simply be given a hallucinatory version of Google Street-View that includes a virtual gun. 

In light of the long wars in Iraq and Afghanistan this insistence on training for conventional ‘manoeuvre warfare’ appears, at best, dangerously misguided. At worst it risks encouraging foreign powers to intervene in conflicts they don’t understand and will only inflame. There’s no doubt that the ability to rehearse missions in virtual environments makes intervention much more tempting – giving commanders a false sense of familiarity with a given region while ignoring the people who live there. When assurances are made that soldiers earmarked for deployment overseas are ‘virtual veterans of 25 bloodless battles’ it will be easy for the public to assume that this knowledge extends to the social and political dynamics at their destination.

In many respects the insurgency in Mozambique is typical of the sorts of armed conflicts likely to arise over the coming decades. In Palma a corrupt government working hand in hand with multinational corporations and private military contractors has dedicated itself to exploiting the poor and the marginalised – which has, in turn, led to unrest and fed religious and ideological extremism. The U.S. has both contributed to, and fought against, this phenomenon in dozens of countries over the last seventy years but their continued fascination with training systems like VBS reveals that planners still approach these conflicts like a collection of small conventional skirmishes that can be fought and won with the right tactics and just a little more sensor data.

In reality this sort of unconventional warfare requires soldiers to conduct what military theorist David Kilcullen has referred to as ‘armed social work’. The challenge for foreign peacekeepers isn’t negotiating rivers and mountain passes, sighting artillery or capturing strategic points. Rather their challenge is distinguishing between militants and the general population and determining who supports who and why. 

These conflicts are culturally, rather than geographically, complex and they demand the sort of soft skills that cannot be conveyed via first-person shooter. As one member of General Stanley McChrystal’s assessment team admitted – the counterinsurgency campaign in Iraq required “a level of local knowledge that I don’t have about my own hometown”. If U.S. commanders find themselves booting up a laptop to work out where best to deploy an infantry battalion it’ll be because they’ve somehow managed to make a bad situation much much worse.

Given the disastrous foreign interventions in Iraq and Afghanistan it’s hard to escape the conclusion that the millions of dollars currently being spent on developing a global military metaverse might better be spent elsewhere. If the U.S. and its allies like Australia insist on policing the globe then they might be better off purchasing a few thousand subscriptions to DuoLingo so that their officers might be able to converse directly with the people living in the shadow of U.S. drones.

Looking at the latest simulation technologies on offer it seems that the question which Jon Peterson posed in regards to the Prussian kriegsspiel – whether early wargames were instructive or simply entertaining – still remains unanswered. The very fact that military institutions like the U.S. Army don’t seem interested in testing the efficacy of these systems is telling. After all, the original SIMNET was designed to allow soldiers to replay and rehearse a battle against a helpless enemy and the Millenium Challenge exercise – after an embarrassing false start – forced participants to adhere to a script of U.S. triumph. Likewise VBS and the global Synthetic Training Environment seem designed to offer military planners yet another wishful mode of warfare – a version of the earth cleansed of people, history and politics. 

References
Phillip Sabin (2012) – Simulating War
Alex Vines (2021) – Responding to Mozambique’s Islamic Insurgency
James Kosur (2021) – Classified Leclerc Tank Documents Were Posted To Win An Argument On A Gaming Forum
Steven Katz (2021) – Don’t let the Afghanistan generals deflect, displace, or dissemble
David Kilcullen (2006) – Fundamentals of Company-level Counterinsurgency
Ralph Peters (1999) – Heavy Peace
Martin van Creveld (1991) – The Transformation of War
Susannah Whitney et. al (2013) – A review of the effectiveness of game-based training for dismounted soldiers
Graham Smith (2012) – Medal of Honor: Warfighter Review
Julian Saenger (2016) – ‘De-zombifying’ DayZ’s Chernarus Terrain For Military Training
Bill Adair (2005) – Did the Army get out-gamed?
Jennifer McArdle, Caitlin Dohrman (2022) – The Full Potential of a Military Metaverse
S.C. Stuart (2019) – How Simulation Games Prepare the Military for More Than Just Combat
BISim (2022) – VBS Tech Conference 3.0, Day 1 – Major Initiatives in VBS4
Philip Dyer (2023) – Arma’s Seagulls of Doom
Talib S. Hussain, et al (2015) – Design and Development of Training Games
Martin Van Crevald (2010) – Why War Games Don’t Work