THE REAL AIR COMBAT BEHIND THE GAME

JET3D is an arcade game, tuned for a six-minute match rather than for realism. But nearly every rule that decides who wins a fight in it was learnt the hard way by real pilots over about a century. This page goes through that history in order, and after each part says where the same idea shows up in the game.

The First World War: aim the aeroplane

What made the dogfight possible was a machine gun that could fire forward through the propeller. Synchronisation gear, first fitted in numbers by Fokker in 1915, timed each round to pass between the blades. The gun was now fixed to the aircraft, so to aim it you aimed the whole aircraft. Fights became about getting your nose onto the enemy, and the safest place to be was where their nose could not reach you: behind them.

The German pilot Oswald Boelcke wrote the lessons down in 1916 for the squadron he led. In paraphrase, the Dicta Boelcke say: gain an advantage before you attack, ideally with the sun behind you; once an attack is started, carry it through; fire only at close range, with the target properly in your sights; never take your eyes off the enemy; attack from behind; if someone dives on you, turn to meet them rather than run; and over enemy ground, always know your way home. He also told his pilots to fight in groups.

In JET3D: the guns are fixed forward, so you still aim with the aircraft. Firing close still matters: rounds drop and take time to arrive, the guns are angled to meet at 250 metres, and a kill takes five hits. The red arrows round the crosshair are there so you never lose sight of the enemy. Attacking from behind matters twice over, because the HOUND is a heat-seeker and sees your exhaust best from astern.

The Second World War: pairs, and never turning with a Zero

By the late 1930s the lone hunter was giving way to teams. In the Spanish Civil War German pilots, Werner Mölders among them, settled on a loose pair of leader and wingman, with two pairs flying as a four. The wingman watched the leader's tail so the leader could concentrate on shooting.

The Pacific taught a second lesson. The Japanese Zero could out-turn almost every American fighter it met early in the war, so American pilots were told not to turn with it. They dived in from above, fired, and used the speed from the dive to climb away before it could turn onto them. This is usually called boom and zoom.

In JET3D: the SABRE turns at about 63 degrees a second, more than anything else in the sky. The BOLT is the fastest aircraft and the worst at turning. A BOLT that turns with a SABRE is a Wildcat turning with a Zero; diving in, shooting and leaving still works better. Mutual support is what team battle is built on, and friendly fire is off, so a team-mate can shoot the aircraft on your tail without hitting you.

Korea: jet against jet

The Korean War (1950–1953) brought the first large-scale fighting between jets, mainly the American F-86 Sabre against the Soviet-built MiG-15, in the sky near the Yalu river that pilots called MiG Alley. The MiG generally climbed better and carried heavier cannon. The Sabre handled better at high speed and had a radar-ranging gunsight that worked out the lead for its machine guns.

At jet speeds, a round fired straight at a turning target arrives where the target was. You have to aim ahead of it, by an amount that depends on range, speed and how hard it is turning. Computing gunsights did that sum for the pilot.

In JET3D: the lead circle is a computing gunsight. Point at an enemy within about 1,100 metres and it draws a ring where your rounds will meet it, with time of flight and drop taken from the same ballistics code the server uses. Like the real thing, it assumes the target keeps doing what it is doing now. A pilot who keeps changing their turn still beats it, which is why jinking has always worked.

John Boyd and energy–manoeuvrability

John Boyd, a US Air Force instructor at the Fighter Weapons School at Nellis, set out to make air combat a science rather than a body of experience. His Aerial Attack Study, written around 1960, laid out manoeuvres and counters systematically. Then, in the 1960s, he and the mathematician Thomas Christie developed energy–manoeuvrability theory.

The idea is that a fighter's ability to fight comes down to energy: kinetic energy in its speed and potential energy in its height. Every manoeuvre trades one for the other or spends both against drag. What counts is specific excess power, the rate at which the aircraft can gain energy: thrust minus drag, times speed, divided by weight. Chart it for two aircraft across speeds and heights and you can see where each one beats the other before they ever meet. E-M theory shaped the F-15 and, even more, the F-16.

In JET3D: this is the flight model. Climbing bleeds speed in proportion to how steeply you are pointed, diving builds it, and turning costs speed through induced drag, more so when you are slow. The hard turn on Shift is the most expensive thing you can do. Boost works like fuel: three seconds in a full tank, refilling over about nine. The aircraft were tuned by measuring each airframe's turn rate and speed, which is a simple version of an E-M chart.

Vietnam: the gun comes back

By the early 1960s many planners thought guided missiles had made the dogfight obsolete, and the F-4 Phantom, the most important American fighter of the period, was designed without an internal gun. Over Vietnam it went differently. Rules of engagement often required a visual identification before firing, which threw away the long range and brought the fight close. Early missiles had minimum ranges, tight launch limits and poor reliability in combat. The Air Force fitted gun pods and then an internal cannon on the F-4E.

The Navy changed its training. A 1968 review of its air combat results led to the Navy Fighter Weapons School, Top Gun, set up at Miramar in 1969 to teach dogfighting against instructors flying like the enemy. The Air Force followed with Red Flag at Nellis in 1975, so that pilots flew their first "combat" missions in training. The lesson has held ever since: missiles are decisive, but only alongside pilots who can still fight close.

In JET3D: the HOUND was designed with that lesson in mind. It needs a solid lock, about a second of holding the target in a narrow cone out to 2.6 kilometres, and without a lock it will not launch. It is deliberately slow, topping out at about 520 km/h, close to an unboosted jet, so a pilot who sees it coming can out-fly it. Gun ammunition is unlimited. Bots will not fire a HOUND from inside about 700 metres, where the target would have no time to react.

Heat, radar, flares and chaff

Air-to-air missiles mostly use one of two kinds of guidance. Heat-seekers home on the target's heat, above all its exhaust. The best known is the American AIM-9 Sidewinder, first used in combat in 1958 by Taiwanese pilots over the Taiwan Strait. Early seekers could only see a hot tailpipe and had to be fired from behind. The all-aspect AIM-9L, which could be fired from any angle, made its name in the Falklands in 1982. Radar-guided missiles such as the AIM-7 Sparrow, and later AMRAAM, home on radar energy reflected off the target, and reach much further.

Each has its own decoy. Flares burn hotter than an engine and give a heat-seeker something brighter to chase. Chaff is clouds of thin metal strips that confuse radar; the RAF first used it in 1943, codenamed Window. Most guided missiles steer by proportional navigation. Rather than pointing at the target, the missile turns in proportion to how fast its line of sight to the target is rotating. If that line stops rotating, the two are on a collision course.

In JET3D: the HOUND is a heat-seeker using proportional navigation, with a limit on how hard it can turn so that a hard break can beat it. Aspect matters. From behind it sees your exhaust and steers at full strength. Head-on it keeps only about a third of its turning, and holding your nose on it while decoupled, inside 900 metres, can break its lock for good. A flare takes the seeker if it lands within 140 metres of the missile and closer to the missile than you are, and you get one set every seven seconds. Cloud blocks a lock in both directions. There is no radar missile and no chaff. See the weapons page for every number.

The Cobra, and flying past the stall

At the 1989 Paris Air Show the Soviet test pilot Viktor Pugachev pulled a Sukhoi Su-27 up past the vertical in level flight, held it briefly, then dropped back to level with almost no change in height. Pugachev's Cobra was done without thrust vectoring, as a demonstration of an airframe that stayed controllable far past the stall. Post-stall manoeuvring became a field of its own. Thrust vectoring, steering the engine exhaust to move the nose when the wings have stopped working, appeared on later Sukhois and on the F-22. Its value in combat is still argued over. Pointing the nose somewhere unexpected can win a shot, but it throws away most of your speed, and a slow fighter is an easy target for anyone else nearby.

In JET3D: there used to be a Cobra on F, a fixed manoeuvre on a timer that took the controls away while it ran. It was replaced by decoupled flight. Hold F and the engine idles, the flight path stops following the nose, and you keep your controls, made sharper on purpose; the explanation in the code is thrust vectoring. The cost is the real one. Drag climbs, you come out slow, and there is a long cooldown before you can do it again.

Modern basic fighter manoeuvres

Fighter pilots today learn close combat as basic fighter manoeuvres (BFM), taught as offensive, defensive and neutral. The vocabulary is common to most air forces:

• Lift vector. The aircraft turns where its lift points. Roll the lift where you want to go, then pull.
• Pursuit curves. Lead pursuit points ahead of the target for a gun shot, pure points at it, and lag points behind it to avoid an overshoot.
• Yo-yos. Climb out of the turn to swap speed for height and avoid overshooting (high), or drop inside it to gain speed and close the gap (low).
• Guns defence. Keep changing your plane of motion so the attacker's sight never settles.

In JET3D: a jet turns by banking and pulling, and the rudder does very little, which is the first lesson of the flying guide. Yo-yos work because gravity and drag are modelled. On guns defence, the game agreed with the textbooks when the bots were tested. A steady break turn got them shot down faster, because a steady turn is exactly what a lead predictor tracks best. The corkscrew they use now keeps the best of them alive for around twenty seconds with an ace behind them. The tactics page applies all of this in detail.

What the game leaves out

A lot, deliberately: radar, long-range fighting, fuel apart from boost. What it keeps is what makes a dogfight a contest of decisions: a fixed gun, energy you spend and must win back, a guided missile that can be beaten, and the old rule that if you lose sight of the enemy you usually lose the fight. For how the game itself is built, see how it works.

Pilot's library · How to fly · About · What changed · Privacy · Terms