Archon: Chess With Fights

The short description of Archon has been the same for forty years and it is doing the game a quiet disservice. Chess with fights. Two armies on a chequered board; move a piece onto an occupied square; the screen drops into an arena and the two of them shoot at each other until one is dead. Everybody who has heard of it has heard that, and it is accurate, and it explains almost nothing about why the thing is still interesting.

Free Fall Associates — Jon Freeman, Anne Westfall and Paul Reiche III — built it for the Atari 800 in 1983, and Electronic Arts published it in their opening run alongside M.U.L.E. and Pinball Construction Set, in the album sleeves, with the designers photographed like a band. The C64 version arrived the following year. Freeman had already co-founded the company that became Epyx; Westfall did the programming, which in 1983 made her one of vanishingly few women credited as the engineer on a major release.

This one landed before I had a machine, so I have no war stories about it. Good. The war stories tend to be about who was better on the joystick, and that is the least interesting thing here.

The board is on a clock

Here is the mechanic that everyone mentions and almost nobody follows through on. The squares of the board are not fixed colours. They cycle — light, through a run of intermediate greys, to dark, and back — on their own schedule, tracked by an indicator at the edge of the screen. A Light piece fighting on a light square enters combat stronger. A Dark piece on that same square enters weaker. Wait long enough and the square changes its mind.

Think about what that does to strategy.

In chess, the board is inert. The value of a square derives entirely from the pieces around it, so evaluation is a pure function of position, and if nothing moves, nothing changes. Archon takes that assumption and puts a metronome under it. A square that is a death sentence for your Golem right now is a decent place for him in four turns’ time. The five power points scattered across the grid sit outside the cycle and mend whoever is standing on them, which turns them into fixed anchors in a board that is otherwise breathing.

So time becomes a positional resource. You can wait as a move — shuffle something irrelevant on the far side of the board while the shade under the square you actually want rolls towards your colour. Your opponent can see it coming and has to decide whether to spend a piece to contest the square early, when the arithmetic is against them, or concede it. That is a tension chess simply does not have and cannot have.

It is also, mechanically, the same problem Populous would set five years later from the opposite end — a landscape that changes value under both players while they are trying to plan on it. Free Fall got there first, on a grid, with a colour cycle.

A capture is only a proposal

The second thing the “chess with fights” line hides is what happens to calculation.

Chess is deterministic in its captures. Rook takes bishop; the bishop is gone; the only uncertainty is whether you read the position correctly. Every strong player’s skill is built on that guarantee, because it lets you calculate lines forward with confidence.

Archon removes the guarantee. Moving onto an occupied square does not capture anything. It opens a negotiation, conducted in an arena, with real-time movement and projectiles and obstacles that shift position while you fight. Your Archer attacking their Troll is a favourable matchup, and favourable matchups are lost regularly by anyone with ordinary hands.

The strategic consequence is enormous and underrated. Because captures are probabilistic, you cannot play Archon by calculating lines. You have to play it by managing risk across a portfolio — which trades you can afford to lose, which pieces you can spend, how much of the board you can hold if the next three fights go badly. It is closer to how you play a wargame with dice than how you play chess. Free Fall built a strategy game whose strategic layer is fundamentally about expected value under uncertainty, and then made the uncertainty something you personally have to resolve with a joystick, which means the variance is partly your own competence.

The armies deepen this by refusing symmetry. Light and Dark do not mirror each other. The Phoenix immolates everything near it, the Banshee’s scream damages without a projectile, the Basilisk turns things to stone, and the Shapeshifter becomes a copy of whatever it fights — a piece whose entire identity is “this matchup is always even”. The Wizard and Sorceress each carry seven spells, one use apiece: teleport, heal, revive a dead piece, exchange two, imprison an enemy, summon an elemental, shift the light cycle itself. Seven irreversible decisions per game, and the last one — reaching in and moving the clock — is the game admitting that its own central system is a lever.

The arena is a second game with its own grammar

Spend an evening with it and you notice the arena has depth the board never advertises. Each piece brings its own movement speed, projectile speed, projectile shape, rate of fire and hit points, and the interactions between those numbers are where the matchup chart actually lives.

The Knight and the Goblin are the pawns, and their fights are grim little melees decided by who commits first. The Archer and the Valkyrie fire fast and die fast, so they win by kiting and lose the instant they get cornered against the shifting obstacles. The Golem and the Troll are slow enough that a competent ranged piece should beat them every time, which is precisely why an experienced player will trade a Golem into a bad square deliberately: the threat of that trade shapes where the opponent is willing to stand.

The Dragon and the Djinni are the heavy artillery, and the game handles them the way good design handles a superweapon — they are strong enough to be worth building a plan around and slow enough that losing one for nothing costs you the game. Neither side gets a piece that simply wins.

What makes this hold together is that the arena rewards knowledge over reflex up to a fairly high skill ceiling. Knowing that a particular projectile passes through an obstacle, or that a given piece cannot fire while turning, is worth more than a fast hand for the first fifty hours. Beyond that, the reflex takes over, which is the problem below.

The obstacles are the underrated part. They move on their own timer, so the arena is a second system with its own clock running alongside the board’s colour cycle. A fight that would be routine on open ground becomes a scramble when a barrier drifts between you and the thing you are shooting. Free Fall put a random element into the resolution layer, kept it visible, and made it symmetrical — which is exactly how you build variance a player will accept.

Where it fights itself

Archon has a real problem and it is worth naming plainly, because the fan case for the game tends to skate past it.

The action layer can eat the strategy layer whole. A player who is significantly better in the arena wins games they have no business winning. They can attack into bad matchups, on the wrong colour square, against superior pieces, and simply out-shoot the situation. When that happens the board stops being a strategy game and becomes a queue of duels, and every clever thing about the light cycle evaporates because nobody has any reason to wait for a square when they can just take it.

Against the computer this shows up as a difficulty curve with a hole in it. The AI plays the board reasonably and the arena adequately, so once your hands are good the strategic layer becomes decoration. Two evenly matched humans get the real game. Two mismatched humans get an exhibition.

Free Fall knew, I think. The obstacle field in the arena shifts around, which caps pure reflex play; the strength differential from the square shade is large enough to be felt. Those are dampers on the problem, and they are not enough. Any hybrid that resolves a strategic decision through an action test carries this tax, and it is why the design has been admired for four decades and copied almost never. Speedball 2 sits on the other side of the same line — a game with a genuine team-management layer that everyone plays as a brawl, because the brawl is where the hands are.

The bloodline

Archon II: Adept followed in 1984 and went further out — elemental zones, a more abstract board, less of the chess grammar that made the first one legible. It is the more ambitious design and the less loved one, which is a pattern.

The real inheritance is elsewhere. Paul Reiche III went on to co-create Star Control, and if you have played it you have already spotted the architecture: a strategic layer where you commit forces, resolved by dropping into a real-time duel between two asymmetric ships with wildly different toolkits, where the better pilot beats the better position. That is Archon, in space, by one of the people who built Archon. The lineage is documented in the credits rather than inferred.

Beyond that, the descendants are mostly people rediscovering the idea by accident. Every tactics game that resolves an attack with a minigame, every board-and-brawl hybrid, every strategy title where a favourable matchup is a suggestion, is walking ground Free Fall surveyed in 1983 with 48K and a colour cycle.

Where to play it

The C64 and Atari 800 versions are both preserved and both play well; the Atari original is marginally cleaner. There is a licensed NES conversion which is worth knowing exists and not much else.

The requirement is the same as M.U.L.E.’s: another person, in the room, of roughly your ability. Get that and Archon stops being a curiosity from EA’s first year and becomes what it was built to be — a strategy game where you can be outplayed and still have a chance, and where the board itself is running a clock neither of you controls.

Forty years of “chess with fights” has sold it as a novelty. It is an argument about what certainty does to strategy, and it still wins.