Pat Kerr Revisits Grand Theft Auto’s Early Driving Physics in Playable Demo

Before Grand Theft Auto became synonymous with enormous cities, criminal escapades and ever-more elaborate simulations, it needed cars that were simply enjoyable to steer. Pat Kerr, the programmer behind a crucial early driving solution for the first game, has now revisited those ideas in an independent playable demo built in JavaScript.

The project is not an official Grand Theft Auto release, nor is it affiliated with Rockstar, Take-Two or the GTA brand. Kerr describes it as a slightly prettier, “remastered” presentation of the underlying driving concept: a compact chance to see and play with the kind of vehicle logic that helped define the original game’s feel.

That distinction matters. Calling the demo a remaster is a useful shorthand for its visualized revisit of old technology, but players should not mistake it for a remake of the full game. Its appeal is much more specific: it is an interactive technical artifact, focused on how a small 2D car accelerates, rotates, grips and slides.

A weekend solution that became GTA’s vehicle foundation

Kerr says he wrote the initial 2D physics simulation during a weekend in late August 1996. That work later became the basis of the vehicle system in the game eventually released as Grand Theft Auto.

Former Rockstar technical director Obbe Vermeij has offered a vivid account of why that contribution stands out. In Vermeij’s recollection, the early version of GTA’s driving was not fun. Kerr, who was skilled with physics and was not then part of the team, supplied driving code that changed that outcome. Vermeij says the new demo feels remarkably close to GTA1, with the space-bar handbrake especially bringing back the character of the original.

It is an unusually direct window into a design truth that can be easy to overlook in modern open-world games: visual scale and simulation detail are not the only things that make driving memorable. A vehicle model has to communicate momentum clearly enough that a player can predict it, push it and recover from it. In a top-down game, where the player is reading a car’s movement from above rather than through a dashboard or chase camera, that clarity is particularly important.

What “rigid body dynamics” means here

Kerr identifies a simple classical 2D rigid body dynamics simulation as the core of the system. Put plainly, the simulation treats the car as a body with position, movement and rotation. Instead of only moving a point around a map, it accounts for the fact that a vehicle has orientation and can spin when forces act away from its center.

The key word is torque. Torque is the turning effect created when a force is applied at a distance from an object’s center. Push at the center of a body and it tends to move straight; apply force off-center and it also wants to rotate. In driving terms, the forces acting through tyres can affect both where the car goes and which way it faces. That is why a fast turn, a loss of grip or a handbrake maneuver can produce a slide rather than a perfectly neat change in direction.

Kerr contrasts this with a more basic approach often summarized by the equation F = ma: force equals mass times acceleration. That principle remains fundamental physics, but a basic point-based implementation can struggle to represent rotation elegantly. Developers may then add separate shortcuts to make an object turn. Kerr’s approach incorporated rigid-body rotation, including torque, so turning behavior could be handled more coherently by the underlying simulation.

That does not mean the model was attempting to reproduce every part of real automotive engineering. Kerr is candid that the vehicle layer sitting on top of the rigid-body core used a simple approximation of tyre behavior. He calls it technically incorrect and says it was effectively a semi-educated guess. Yet it was good enough for its purpose.

“Good enough” is not a dismissal in game physics. It is often the target. A game vehicle does not need to be a laboratory-grade recreation to feel responsive, consistent and expressive. In fact, a more detailed simulation can be less approachable if players cannot understand why a car reacted a certain way. The meaningful result is whether steering inputs, speed, grip and rotation form a readable system that rewards practice.

The demo makes its invisible rules visible

One of the most interesting elements of Kerr’s JavaScript recreation is that it is not content to leave the math hidden beneath the car sprite. It uses visual aids to show what the model is doing. A movement arm attached to the vehicle helps illustrate the physical relationship between turns and acceleration, while a red trail shows tyre grip during cornering and handbrake actions.

Those overlays make the demo useful beyond nostalgia. They give curious players a way to connect a familiar sensation—“the car is beginning to swing out”—to a visible representation of the motion and tyre behavior producing it. The handbrake becomes more than a button that makes a car slide; it becomes a way to alter grip and provoke a change in direction.

For anyone learning how driving systems are designed, that is a valuable framing device. Games routinely present the final sensation without showing the layer of rules underneath. Here, the presentation encourages players to inspect the relationship between input and outcome. Turn while carrying speed, and the car’s movement does not instantly align with the direction it faces. Use the handbrake, and the shift in behaviour is rendered in a way that can be observed as well as felt.

Why the handbrake remains such a strong memory

Vermeij’s specific mention of the space-bar handbrake gets at why early GTA driving has endured in players’ memories. A handbrake input is a simple, legible tool: it offers a deliberate way to make a tight corner, initiate a slide or rapidly reorient a vehicle. In a top-down format, that kind of maneuver can be especially satisfying because the player can immediately see the car’s angle changing relative to its path.

The demo’s tyre trail helps unpack that sensation. A visible trace turns the normally invisible question of traction into something concrete. Grip is the tyre’s ability to hold the road and guide the car in the intended direction. Reducing that grip through a handbrake action makes it easier for the vehicle to rotate or drift. The effect may be simplified, but the resulting behavior gives the player a manipulable language of speed, angle and recovery.

That readability is arguably the lasting lesson of Kerr’s work. The original GTA did not require a dense vehicle-customization system or a realistic cabin view to make driving matter. It needed cars that responded in a recognizable, playful way—cars capable of getting into trouble quickly and, just as importantly, allowing a skilled player to extract them from it.

An early design compromise, preserved rather than hidden

Kerr’s description of his tyre model is also a useful corrective to the assumption that landmark game systems must begin as perfect technical breakthroughs. The vehicle simulation was built from a sturdy rigid-body basis, then supplemented with an approximate model that Kerr knew was not physically accurate. The result became influential not because every detail mirrored real life, but because the combination created believable-enough handling.

That is the practical difference between simulation and simulation-inspired game feel. A simulation can borrow real physical concepts—mass, force, rotation, torque and traction—without being an exhaustive reproduction of reality. A vehicle system then becomes a design decision as much as an engineering one: which behavior should be stable, which should be exaggerated, and which should be easy for players to learn?

Kerr’s demo is valuable precisely because it does not rewrite that history into a claim of flawless realism. It instead preserves the experimental character of the original solution. The code was a response to a particular problem: GTA’s driving needed to become fun. Its eventual success shows how a constrained approximation can become part of a game’s identity.

From a 2D experiment to modern GTA expectations

The contrast with the series’ current ambitions is substantial. Grand Theft Auto VI has been associated with highly detailed, simulated vehicles moving through a fictional Florida setting, a far more elaborate proposition than an early 2D vehicle model. But the lineage is still easy to see at the level of player interaction. Cars remain central to how the series creates movement, chaos, escape routes and opportunities for improvisation.

Recent discussion of GTA VI’s driving has characterized its goal as combining the strongest aspects of GTA IV with the accessibility of GTA V. That is a broad modern target, but Kerr’s demo offers a useful reminder that accessibility and personality were already central concerns at the beginning. The oldest version of the formula needed players to understand a turn, enjoy a skid and trust that the handbrake would produce a satisfying result.

There is also a wider lesson for contemporary projects, including smaller games and technical experiments. A compact prototype can reveal more about an interactive system than pages of design description. Kerr’s recreation isolates a foundational idea, makes it playable and adds diagnostic visualizations. It is a modest scope with a clear purpose—an approach that can be as instructive as it is nostalgic. Developers and players interested in smaller, distinctive game concepts may also want to look at Humblets’ handcrafted fantasy pitch, another example of a project whose identity rests on clearly communicated mechanics and presentation.

Why this small demo matters

For longtime GTA players, the immediate attraction is the chance to revisit a recognizable handling style—particularly the handbrake-heavy feel Vermeij singles out. For technically minded players, the value lies in the annotations around it: motion, turning and tyre behavior become things to inspect rather than merely accept.

Most of all, the demo gives credit to an often invisible part of game-making. Graphics, maps and characters are usually easier to point to, but vehicle handling is a layer of decisions that players register through their hands. Kerr’s recreated system demonstrates how much personality can emerge from a modest 2D rigid-body model, an imperfect tyre approximation and a carefully useful handbrake.

Thirty years on from the late-summer weekend when the original simulation was written, the project is less a claim that old technology is superior than a clear demonstration of why it worked. The code solved a fundamental problem, and its behavior still communicates a particular kind of momentum, risk and control—the ingredients that made GTA’s earliest cars worth driving in the first place.