Recently we were hit with the sad news that Lode Runner creator Douglas E. Smith had passed away. Lode Runner was a massive success in the early Eighties thanks to its tight arcade gameplay, engaging game mechanics and fast gameplay. It spawned countless sequels and was notable for having an excellent level editor, that ensured you could play it endlessly. We were fortunate enough to speak to Douglas early last year about the creation of his groundbreaking game.
For me, it was really about the excitement of building an algorithm that could be thrown at any level, rather than pure gameplay. I wasn’t a gamer, and nor am I today – that’s the irony of the thing. It was all about the beauty of the algorithm.” James Bratsanos is recalling his early programming days, holed up in high school with a Commodore PET no-one else was using. His surprisingly lenient maths teacher just let him get on with it, and James, having devoured a book on BASIC games, wanted to make something “a little more exotic.”
A friend had returned from an arcade, excited about a game. “He didn’t explain it well, and I took creative liberties and assumed I understood what he meant. So for certain elements, I completely misinterpreted it,” laughs James. But with this vague direction, he devised a game based on you “trying to collect things while monsters were chasing you.” To James, it was obvious to fashion levels from platforms and ladders, because “anti-gravity didn’t seem that big back then,” and it provided options to experiment with gravity and digging holes you could escape through or trap monsters in. But rather than digging directly down, the hero could only dig left or right.
“Part of that was if you’re going to dig a hole, you can decide whether to jump into it or not, but if you could only dig directly below, you’d immediately fall into the hole and could only ever dig straight down,” recalls James. “By digging left or right, there was strategy, and you could plan ahead, digging spaces two squares wide, jumping into that hole and then digging again.”
Aficionados will recognise this concept as Lode Runner, but at this point in 1980 the game was called Suicide, had an ‘@‘ symbol for the hero, and juddered about on the Commodore PET. As James reiterates, he’d had no proper schooling in computers: “Because I had no idea how anything worked, I felt it was logical at the time to build an engine that could interpret a game level and then run a processing loop on the monsters.”
The variety of different levels in Lode Runner is staggering.
In a sense, it’s appropriate that this making-of is already a loop of sorts, because that’s fundamentally what Suicide was. This loop was not only the most important aspect of James’s creation, but it also informed every Lode Runner game that subsequently existed. “The absolute core was no pre-programmed anything. Just suck the level in, get the coordinates of everything – the monsters, gold to collect, the good guy – and run a loop on it. And the cool thing was the loop was constantly changing, depending on what the characters were doing.” James recalls the system scanned for key hits to ascertain changes to the player’s position, and then ran the algorithm for each monster, which would adjust their positions. “Because I was a naïve non-programmer trying to learn, I looked at everything like a human would, and figured you’d try to find the shortest route to your prey,” he explains. An internal scoring engine looked at steps from a monster to the player in each direction, if reaching them was possible, and the lowest score won out. If there was no benefit in moving, the monster would stay put, because it couldn’t get closer to you. “Really, I was doing recursion without knowing what recursion actually was, which was pretty cool,” laughs James. “The beauty of the system was that it didn’t matter how many monsters I had, because the underlying engine worked with them all. And since there was no specific logic tied to any one level, I could make endless levels and apply the logic to them. Later, I realised this was pretty novel for the time.”
The manner in which James’s system worked made the monsters seem to have intelligence – they’d often pause rather than home in on you. “And what made the game really interesting was they ran all this logic to determine if they were going to make this one move, left or right, but you could then jump off a platform and end up falling halfway down the screen. At that point, all bets were off, because your position changed so quickly,” says James, adding that – as you might have guessed – “a lot of the fun for me was applying the logic to different levels and not necessarily playing the game!”
It looks cool, but is tricky to complete.
In 1981, after high-school, James went to the University of Washington. He took a Fortran class, became a teaching assistant and worked in the computer lab: “I became friends with a guy called Tracy Steinbeck, and we started rewriting Suicide for the VAX, and then Doug [Smith] got involved.” The game was renamed Kong, possibly in tribute to the coincidentally similar Donkey Kong, and continued to evolve. “We got bored, thought ‘What else can you do?’, and started adding arrows, bombs, mines and shields,” says James. But the biggest change was still to come, with the game’s shift from spare-time computer-lab project to commercial concern.
“During summer, I’d drag my young nephew to campus and he played the game while I worked on it,” remembers Doug. “He asked why I couldn’t stick it on a floppy, take it home and put it on the Apple II. I had to explain that the VAX had no floppy drive and the game was in the wrong language, but he kept bugging me, so one weekend I rewrote it for the Apple II.” Interestingly, Doug didn’t own an Apple II nor even care for the machine: “I had no fondness for it, but I did have a natural knack for understanding computers and resolving problems with them.” Fortunately, his then-roommate owned one and often asked for Doug’s assistance, which drew him into understanding the machine and provided the means to craft the last Lode Runner prototype, Miner.
Luring enemies to their deaths is always fun.
Recollections differ regarding this period of history – Doug remembers James lost interest, but James says: “I can see how Doug would interpret it that way, but I didn’t know assembly and wasn’t in a position to quit school to work on the Apple II. But Doug was gonna invest everything in writing the 6502 version before he ever had anyone to peddle it to.” Subsequently, everything progressed rapidly. Newspapers from the time report Doug paid James $1,500 for the rights to his work, finished Miner and submitted it to Brøderbund. “I got my rejection letter, saying ‘Thank you for submitting your game, but it does not fit within our product line’,” says Doug, who then borrowed $800, bought a colour monitor and joystick, and set about improving the game to make it saleable.
Graphics and movement were overhauled, the latter shifting from the character-block movement the game used since its PET incarnation to smoother pixel-level motion. Controls were honed, enabling commands to be better cued, like in Pac-Man. “That was important—there was a margin of error and you could time it to execute when you got to the appropriate point,” explains Doug. Despite a perfectionist streak, he wanted everything to be fluid versus overly precise, but details also mattered. While initially keen on animation assistance for the hero and enemies, no-one understood Doug’s desire that humanoid characters not be ‘treadmill’: “I wanted to make sure when they put a foot down, it stayed fixed to that part of the ground.” He subsequently created the graphics himself.
Further changes impacted the game, including a pre-level pause, awaiting player input before the enemy guards first moved. “Lode Runner was always intended to be a kind of strategy game, and I thought it would be good if you could pause, figure out a strategy and execute it,” he says. But although Doug was the game’s absolute director, some aspects were out of his hands. Having re-sent the improved game to various publishers, Brøderbund reconsidered, and Doug signed a contract, promising 150 levels. Lacking time to create them, they were farmed out to friends and neighbours; hundreds were supplied to the publisher on disk, with a utility enabling levels to be eliminated and ordered to suit.
Collecting gold sounds easy, but the levels could be very challenging.
When asked about the quirky guard AI and its surprising sophistication for the day, Doug laughs and suggests we’re “maybe giving too much credit,” because that too wasn’t entirely his doing. “Back then, a lot of it was the luck of the process, realising there were quirks of the logic you could utilise.” During testing, he realised you could ‘force’ enemy guards to collect gold on your behalf, and so “use them as a bridge to get to otherwise unobtainable behaviours”. This opened up avenues for advanced level design, and, in a sense, followed James’s lead, emphasising a system over hard-coded level-specific knowledge and seeing what happened. Doug even admits his penchant for optimisation almost ruined the game: “In all honesty, the logic ended up being kind of a fluke, even though it was integral to some puzzle-solving aspects. I did at one point refine the logic to be perfect, so guards would always find the shortest path to you, but that lost a lot of the quirky character. The game became too perfect. I spent a month on that and had to rip it all out and revert to an earlier version!”
A rarity at the time, Brøderbund intentionally left the level editor in the shipping product, resulting in countless custom levels being created and submitted to the publisher. “We got so many we did a sequel, Championship Lode Runner, using the toughest levels,” says Doug. “We removed the editor, so people couldn’t peek at the levels, and there are some I spent weeks trying to solve! People had definitely taken advantage of and understood the quirkiness of the AI, and created levels that took a dedicated person to get through.”