How Ken Silverman’s Build engine shaped Duke Nukem 3D and 90s FPS design

Before first-person shooters became cinematic military campaigns, open-world sandboxes or esports arenas with weapon skins priced like small appliances, they were often cramped, loud, rude and wonderfully strange. In the middle of that strange period stood the Build engine, the technology behind Duke Nukem 3D and several other defining DOS shooters. It was not the cleanest engine, not the most technically pure, and not always the easiest to explain without making hand movements in the air. But it did something many engines never manage. It gave games a personality. Build was the work of Ken Silverman, a young programmer whose engine became the foundation for a whole branch of 90s FPS design. Where Doom made abstract hellish arenas feel fast and readable, Build helped designers make cities that looked lived-in, even if everyone living there had apparently been abducted, mutated or replaced by pigs with shotguns. 

A young programmer and a very useful illusion

The Build engine arrived during a period when game technology was moving at a ridiculous pace. One year, a flat maze with textured walls felt like the future. The next year, players expected windows, lifts, explosions, slopes, moving platforms, mirrors and rooms that looked as if a human architect had at least been threatened with involvement.

Build was technically part of the 2.5D tradition. That phrase is a little ugly, but useful. It means the world looked three-dimensional and behaved convincingly enough for players running around with a shotgun, yet it was not built like the full polygonal 3D engines that would soon dominate. Instead, Build used sectors, walls, sprites and clever rendering tricks to create places that felt more complex than they truly were.

This is where the engine’s character begins. Build was not trying to be a sterile simulator of reality. It was an efficient theatre machine. It gave level designers a box of stage tricks and said, more or less, make the player believe this city exists for the next ten minutes. In Duke Nukem 3D, that was enough. You did not need a fully accurate urban model. You needed a street, a cinema lobby, a rooftop, a bathroom mirror, a lift, some broken glass, a ridiculous alien and a vending machine that looked like it had seen things.

Why Duke Nukem 3D felt like a city, not a maze

Duke Nukem 3D is often remembered for its attitude, jokes and macho parody, but the game’s lasting power also comes from its spaces. Its levels were not just corridors with wallpaper. They had recognisable functions. A cinema looked like a cinema. A bar looked like a bar. A hotel felt like somewhere that should probably fail a safety inspection. The game did not merely ask players to survive. It invited them to poke around.

That sense of place came directly from what Build made practical. Sloped floors and ceilings helped areas feel less flat. Lifts and doors could move in ways that made buildings seem mechanical. Sectors could shift, rotate or change, turning the environment into part of the show. Sprites could become furniture, clutter, switches, enemies, corpses, decorations and visual gags. The result was a world packed with small readable details.

This mattered because 90s shooters were still negotiating what first-person games could be. Doom had proved speed, violence and atmosphere. Quake would soon push real 3D into the centre of the conversation. Build sat between those moments and said that maybe a shooter could be about recognisable places, interactive props and environmental comedy. It was the engine for the period when developers discovered that players loved opening toilets for no good reason. Games have been dealing with the consequences ever since.

Dirty cities, clean design instincts

The phrase dirty cities fits Build because the engine was especially good at urban mess. Its games loved alleys, warehouses, shops, clubs, temples, trains, apartments, laboratories and places where someone had clearly ignored every building regulation. Build’s spaces were rarely elegant, but they were tactile. You could recognise what a room was supposed to be before the shooting started. Often, after the shooting stopped, it looked as if the room had made several poor life choices.

Duke Nukem 3D’s Los Angeles was not realistic in the modern sense. It was a compression of city clichés, action film sets and level-design necessities. Yet it felt fresh because it gave players context. You were not just in Episode 1, Level 2. You were in a cinema. You were not merely searching for a keycard. You were breaking into a location with a purpose. This small shift did a lot of work. It made levels memorable.

That was one of Build’s real gifts. It gave designers enough flexibility to create theme, joke and function in the same space. A room could be combat arena, punchline and navigational landmark. A bathroom could contain a mirror, a secret, a gag and an ambush. The result was a style of FPS design that rewarded nosiness. In a Build game, curiosity was almost a weapon. Sometimes it was also how you found a hidden room full of explosives, which is the most 90s sentence available.

Not magic, just useful compromise

It is easy to overpraise old engines because nostalgia adds bloom lighting to everything. Build deserves admiration, but not mythology. It had limitations. Its handling of vertical space was clever rather than fully free. True room-over-room geometry was not its natural language. Designers had to understand its rules and work around its boundaries. Some effects were smoke and mirrors, though in fairness smoke and mirrors are also a perfectly valid design philosophy if the player is too busy fighting cultists to complain.

That compromise is exactly why Build is interesting. It was not powerful because it removed every constraint. It was powerful because its constraints were productive. They encouraged designers to build dense, legible, surprising spaces. The engine did not ask teams to create giant worlds. It helped them create compact spaces with strong identity.

Modern games often chase scale. Build games often chased density. Every room could have a button, a moving wall, a secret, a sight gag, an ambush or a texture that made the place feel slightly more specific. This is why many Build levels remain easy to remember decades later. You remember the cinema, the train, the carnival, the hotel, the morgue, the temple. You remember places, not just layouts.

The editor mattered as much as the renderer

An engine is only as influential as the work it allows people to make with it. Build’s importance was not just in what it drew on screen, but in how it enabled level authors to think. Its editing tools let creators assemble interactive spaces with a directness that suited the era. The result was a wave of maps that felt handmade in the best sense. You could sense designers testing boundaries, bending tricks and asking what happens if this wall moves, this floor drops, this room explodes or this toilet becomes relevant to combat. History does not record enough toilet-based innovation, frankly.

For Duke Nukem 3D, this meant the technology supported the tone. Duke’s character was loud and crude, but the world around him sold the joke. The levels were full of locations that matched the game’s action-movie parody. The engine made those locations possible on mid-90s PCs without demanding impossible hardware. That balance between ambition and practicality is a major reason Build became so associated with a particular kind of PC shooter.

Beyond Duke: Blood, Shadow Warrior and the Build family

Duke Nukem 3D may be the poster child, but Build’s identity spread through other games as well. Shadow Warrior used it for fast, violent, martial arts-inflected action with more technical showmanship and a different rhythm. Blood pushed the engine towards gothic horror, black comedy and nasty combat encounters. Other Build-powered games explored fantasy, science fiction, rural parody and licensed material with varying levels of success.

That variety shows how flexible the engine could be. It did not impose one exact mood. Instead, it imposed a certain design texture. Build games tended to feel busy, angular, interactive and slightly unhinged. They were often filled with switches, secrets, traps, moving parts and environmental jokes. Even weaker games using the technology inherited some of that restless quality.

The best Build games understood that the engine’s strength was not visual realism alone. It was attitude through architecture. Blood’s haunted spaces, Shadow Warrior’s layered set pieces and Duke’s grubby cities all used the same broad technological vocabulary, but each bent it towards a different personality. That is a sign of a useful engine. It does not just show itself. It lets other games show off.

The smell of DOS-era ambition

Part of Build’s appeal today is that it captures a specific kind of PC development culture. It belongs to the period when small teams, strange tools and obsessive programmers could still redirect whole genres. The gap between bedroom coding and commercial hit was narrowing and widening at the same time. Hardware was improving, players were hungry, and everyone seemed to be inventing the future using caffeine, maths and a user interface that probably frightened at least one intern.

Ken Silverman’s role in this story is unusual because Build carried such a strong individual technical imprint. Many famous engines are associated with teams, studios or corporate pipelines. Build feels closer to a personal machine that other people learned to drive very fast. That does not reduce the importance of the artists, designers and game teams who used it. Duke Nukem 3D became Duke Nukem 3D because 3D Realms built a world, tone and campaign around the tech. But the engine gave them the grammar.

And what grammar it was: sectors, sprites, slopes, tricks, motion, clutter, secrets and urban theatre. Not elegant prose, perhaps. More like graffiti written on a rocket launcher. But effective.

Why Build still matters in retro FPS design

The modern retro revival has made Build feel newly relevant. Players have rediscovered the appeal of fast movement, readable layouts, secrets, strong silhouettes and levels that are designed rather than merely streamed. New games inspired by the 90s often borrow from Doom, Quake and Build at the same time, but Build’s influence is easy to spot. When a shooter gives you a grimy city block, an absurdly interactive room, a tiny environmental joke and a secret hidden behind something suspiciously ordinary, Build’s ghost is probably hanging around near the vending machine.

Its legacy is not only technological. It is a reminder that constraints can produce style. Build could not do everything, so its best games became excellent at doing particular things: dense maps, urban spaces, props, movement, tricks and attitude. In an industry where engines now compete on photorealism, global illumination and how beautifully a puddle reflects a sad man, Build’s achievement looks refreshingly direct. It helped designers make places players wanted to explore and blow up.

Not the future forever, but the right future then

Build did not remain the centre of shooter technology. Fully 3D engines took over, hardware acceleration changed expectations, and the late 90s moved quickly. Quake, Unreal and later engines pushed the medium into spaces Build was never built to dominate. That does not make Build a dead end. It makes it a brilliant bridge.

It connected the speed and clarity of early 90s shooters with the environmental ambition that would define later games. It made players expect more from levels. More interaction, more recognisable spaces, more verticality, more jokes, more secrets, more stuff that did not need to be there but made the world feel designed by mischievous humans. Build was not the final answer. It was the engine that asked better questions.

The engine behind the attitude

The Build engine’s real achievement is that it helped turn first-person shooters into places. Not realistic places, necessarily. Not sensible places. Certainly not places with acceptable plumbing. But places with identity.

Ken Silverman’s technology gave Duke Nukem 3D and its relatives the tools to create a dirty, funny, aggressive, interactive version of the 90s city. It was a world of rooftops, cinemas, alleys, toilets, neon and explosions, held together by clever compromises and design nerve. The result was not just a successful engine. It was a recognisable branch of FPS history.

Build made DOS shooters feel less like abstract mazes and more like trouble you could walk into. That is why it still matters. Not because it was perfect, and not because every game that used it was a classic. It matters because for a few noisy years, one programmer’s engine gave PC shooters a new shape, a new smell and a new attitude. And yes, probably a secret room behind the bookshelf.