Linux gaming before Steam: the story of Loki Entertainment

Today, Linux gaming can feel surprisingly ordinary. You open Steam, press install, and Proton performs its little translation ritual in the background. A Windows game starts on a Linux machine, the frame rate is decent, and nobody in the room needs to manually edit an XFree86 config file. This is progress. It is also slightly suspicious. But long before Valve made Linux gaming part of a billion-dollar hardware and platform strategy, there was a much stranger, riskier and more fragile idea. What if Linux games could be sold in boxes, on shop shelves, like any other PC title?  That was the bet made by Loki Entertainment, usually known as Loki Software. From 1998 to 2002, the small California company tried to create a commercial Linux games industry almost from scratch. It did not have Steam. It did not have Proton. It did not have a mass desktop Linux audience. A challenging crowd, in other words. Try selling boxed games to people who think compiling their own kernel is a relaxing Sunday activity.

The retail-box era of Linux gaming

The late 1990s were an odd moment for Linux. The operating system had credibility in servers, universities and hacker circles, but the desktop was still rough around the edges. Installing graphics drivers could feel like arguing with a printer that had read Nietzsche. Sound support was inconsistent. Game controllers were not always welcome guests. Hardware acceleration was improving, but it was not exactly plug-and-play paradise.

PC gaming, meanwhile, was dominated by Windows. DirectX had become the practical centre of gravity for developers, publishers and hardware vendors. If you wanted to play the latest big commercial games, you ran Windows. Linux users could enjoy open-source games, emulators, MUDs, roguelikes and various community projects, but the glossy boxed titles belonged to another world.

Loki Entertainment tried to connect those worlds. Its model was simple in theory: license the rights to port existing Windows games, do the engineering work to make them run natively on Linux, package them professionally, sell them as commercial products and prove that Linux users were a real gaming market.

In practice, nothing about it was simple. Every game was a negotiation. Every port involved platform assumptions buried deep in code. Every release had to cope with different Linux distributions, library versions, drivers and hardware combinations. Windows had its own mess, of course, but it was a familiar mess. Linux was a buffet of messes, and the plates were different sizes.

The people who bet early

Loki was founded by Scott Draeker, a former lawyer with a serious interest in software licensing and Linux. The company also became closely associated with Sam Lantinga, one of the key programmers behind SDL, the Simple DirectMedia Layer. Ryan C. Gordon, later known across Linux and Mac gaming circles for a long list of ports, was also part of the broader Loki story.

This matters because Loki’s legacy was never only its catalogue. The company became a training ground and tool factory. The people around it helped build some of the plumbing that later made cross-platform games less painful. Linux gaming was not waiting for a single hero company to arrive. It was a long chain of awkward, clever, stubborn engineering efforts. Loki was one of the early links that did not break quietly.

SDL, OpenAL and the tools behind the shelf

The most durable part of Loki’s impact was not a cardboard box. It was software infrastructure.

SDL became one of the great quiet successes of cross-platform development. It gave programmers a way to handle windows, input, audio and graphics across different systems without writing a completely separate universe for every platform. It was not glamorous in the way a new graphics engine is glamorous, but it solved the kind of boring problems that decide whether a port ships or dies in a folder called final_final_really_final.

OpenAL was another important piece. Games needed more than pixels. They needed spatial audio, effects, positioning and a way to avoid being trapped inside one proprietary platform’s audio stack. Loki’s work around OpenAL helped push the idea that Linux gaming needed serious multimedia APIs, not polite beeps and a prayer.

Loki Setup also deserves attention. Installing software on Linux in that period could be an adventure, and not the good kind with treasure at the end. Loki Setup provided a more consistent installer framework for Linux games, including graphical and console installation options. It was built for a world where users might be running different distributions with different library versions and different ideas about where files should live.

Then there was GtkRadiant, developed with id Software as a level editor connected to the Quake ecosystem. This was another example of Loki’s practical contribution: tools that made Linux useful not just as a place to run games, but as a place to build and modify them.

Why the business was doomed

Loki’s tragedy was not that the idea was foolish. It was that the idea arrived before the market could support it.

The company had to pay licensing fees, employ specialists, support multiple games and deal with all the normal costs of publishing. But the number of Linux desktop gamers willing to buy boxed commercial games was small. Worse, Linux culture at the time had a complicated relationship with paid software. Many users loved free software as a principle, not just as a price tag. Others were willing to pay but had already accepted dual-booting as the practical solution.

Retail was also unforgiving. Getting a box onto shelves did not guarantee sales. A Linux version of a Windows game was often released later, aimed at a smaller audience, and sometimes priced in a way that reflected the cost of porting rather than the discount-bin expectations of PC shoppers. For the average publisher, Linux looked like a rounding error with a beard.

There was also the support burden. A Windows game had to survive Windows. A Linux game had to survive Linux, the distribution, the kernel, the graphics driver, the sound system, the C library and the customer’s personal belief that nothing should ever be installed outside their preferred directory structure. Somewhere, a support engineer probably stared at a bug report and aged three years.

Native ports before compatibility layers

The modern Linux gaming story is dominated by compatibility. Proton, Wine, DXVK and related technologies allow many Windows games to run on Linux without a bespoke native port. This changed the economics. Developers no longer have to produce a separate Linux version for every game. Users no longer have to wait for a specialist porting house. Valve can improve the layer, and thousands of games benefit at once.

Loki lived in the opposite world. Its work was native porting, game by game. That meant real access to code, direct engineering, platform-specific fixes and a proper Linux executable. When it worked, it felt clean. The game belonged on the platform. It was not pretending to be somewhere else.

But native porting also scaled badly. Every title was its own mountain. The more games you wanted, the more mountains you had to climb. Proton turned Linux gaming into a road network. Loki was still hiring mountaineers.

The games that made Linux feel legitimate

For Linux users of the time, seeing familiar commercial titles arrive mattered emotionally and practically. Civilization: Call to Power was not a tiny experimental project. Railroad Tycoon II and Heroes of Might and Magic III were recognisable PC strategy names. Quake III Arena was a technical showcase. Descent 3 and Heavy Gear II appealed to players who wanted Linux to do more than run terminals and web servers.

The boxes themselves mattered. A boxed Linux game on a shelf said that Linux was not just a hacker workstation. It could be a home computer. It could have entertainment software. It could participate in the same consumer culture as Windows, even if the shelf space was tiny and located somewhere between networking books and a lonely copy of Corel Linux.

There is a danger of overstating this. Loki did not make Linux gaming mainstream. It did not transform the market. Most PC gamers barely noticed. But for the people who cared, those releases were proof of life.

After Loki

Loki filed for bankruptcy in 2001 and closed in 2002. The retail-box dream had run out of cash. The Linux gaming industry did not vanish, but it became more scattered. Linux Game Publishing continued part of the mission. Independent developers, open-source communities and individual porters kept pushing. id Software remained unusually friendly to Linux for a time. Later came Humble Bundle, Steam for Linux, Steam Machines, Proton and Steam Deck.

The irony is that Loki’s dream eventually became partly true, but not in Loki’s form. Linux gaming did not conquer retail shelves. The shelves disappeared. Digital distribution changed the business. Compatibility layers changed the technical model. Handheld PC gaming gave Linux a mainstream device through Steam Deck, but most users never had to think about Linux at all. That may be the most Linux outcome imaginable: success arrives when the operating system becomes invisible.

Why Loki still matters in the Steam Deck age

Steam Deck users do not need to know Loki’s name to benefit from the world it helped imagine. The current Linux gaming ecosystem stands on many layers: open graphics drivers, Vulkan, Wine, Proton, SDL, community testing, developer goodwill, commercial pressure and plenty of work by people whose names never appear on a store page.

Loki belongs near the beginning of that chain. It was not Valve before Valve in size, money or influence. That comparison is too grand. But it was Valve before Valve in one narrower, important sense: it treated Linux gaming as a market worth building, not as a joke or a weekend curiosity.

The company tried to make Linux a retail game shelf and failed. But in failing, it left behind tools, experience and a proof that the platform could do more than enthusiasts were used to expecting from it. Loki’s boxes are now collector curiosities, its installers are antiques, and some of its binaries need more coaxing than an old scanner on a modern laptop. Yet the idea behind them still feels familiar.

Linux gaming did not begin with Proton. It did not begin with Steam Deck. Before all that, there was a small company trying to sell proper commercial games to people running Linux on beige PCs. It was brave, technically impressive, commercially fragile and occasionally absurd. In other words, it was very Linux.