I came into the hobby through eurogames, attracted by their machine-like elegance and their general aversion to Random Elements. In fact, the eurogames that I disliked tended to be the ones that used dice or swingy card decks. When I started designing games, I kept these random elements to a minimum, eschewing them when I could and taking great pains to mitigate them when they were necessary.
Sometimes eurogamers that move into wargames express shock at the frequent use of dice to resolve combat, but that wasn't the case for me. Rolling dice on a CRT didn't seem like a Random Element, but like a Probability Element, one that could be mitigated by marshalling the right forces against the weak points of the enemy line. I might not know exactly what was going to happen in a combat rolled on the three-to-one column - defender retreat? defender eliminated? - but I knew it wasn't going to end up with attacker casualties like I would at a two-to-one. The vagaries of the die roll were carefully bounded by the columns of the CRT, like a circle of salt around these six-sided demons.
When I started designing wargames, I went heavy on the salt myself, tightly constricting the range of results. My first game to use a chit-pull mechanism, Blood in the Fog, gave each player their own cup and limited the number of alternating activations each turn. This created some variability and unpredictability without really creating the same kind of chaos or randomness that you usually get when a game calls for what Brian Train has called a Dunnigan Ceramaceous Randomizer.
But over time, I began to wonder if my games were too restrained - if I had drawn that circle of salt too tightly. One of the problems some folks had with Optimates et Populares is that, other than some slight variance at set-up and one action that depends on a die roll (which may or may not ever get used), it's essentially a combinatorial game, completely determined by player actions. Which doesn't sound like a bad thing necessarily, but it also doesn't particularly sound like the end of the Roman Republic. It's too stately and controlled; where's the blood and the terror?
I'm still proud of that game, but when I was working on its sorta sequel This Guilty Land, I leaned hard in the other direction - using the swingy deck mechanics to embody a tumultuous and contentious political period. It didn't just create a fuller sense of narrative, but a greater number of narratives: the story of a game that starts with a four-four split is going to be different than one with a four-six or a four-eight. It might be in the course of a game in which Oppression is dominating the proceedings that one new card is needed for Oppression, but four new cards for Justice come out, cards that might shift things decisively in their favor. The primary skill needed to play the game well isn't so much being able to adapt to these changing circumstances, but being able to adjust the flow of useable cards from and back into the deck, a process over which the players have agency but imperfect control.
And the me who started playing and designing games almost ten years ago would've had a conniption fit reading that, because for that me control was paramount, and all of this sounds like Random Elements rearing its ugly head. Somehow, the circle of salt has been disrupted, and the demons are loosed. What do you mean, I might draw twelve cards on this turn and the other side only two? That sounds nuts! And it is nuts. It's unlikely to happen, but it's possible, and that combination creates something that a more constrained game experience can't.
Earlier this year, I played the game Sea Evil, and I'm beginning to wonder if it wasn't the single most consequential game that I've played all year. It was so completely unlike anything else I have ever played. It did not fit any of my notions of game balance or "good" game design. I'm not really sure if it is "good" in any conventional sense of the term, but it's not really interested in being "good"; it's interested only in being its own thing, and succeeding on its own terms. That game is set up so that one new demon, witch, sea creature, or what-have-you emerges from the murky depths each turn. At the end of turn one, my fellow whalers and I had things well in hand. Except at the beginning of turn two we rolled a random event that had all twenty-plus abominations popping on the map at once. It was unlikely, but possible; the game created room for something that was utterly insane and utterly memorable.
Playing that game codified something for me, something I was vaguely aware of when I was working on This Guilty Land: randomness isn't always a thing to be mitigated and guarded against, but a thing to be embraced, a very useful tool in the designer's kit that creates a space for unique and compelling experiences.
Of course, this kind of thing comes with the territory for a game like Sea Evil, which might end with the players arm-wrestling the Master of Atlantis, but is far less appropriate for a serious historical model. Or is it? If the point of the game is to model history as it happened in the past, this led to that led to that, an inevitable chain of events, then yeah, it's a poor fit. But if you want to model history as it is lived, in the present, then uncertainty and variability - that special alchemy of unlikely but possible - becomes essential.
Both approaches are valid, and both can be concerned with modeling the same factors - the same tensions and reasons and problems. But the first approach privileges explaining what happened and perhaps why, drawing a tight protective circle around the historical result, ensuring its faithful and accurate recreation. The second emphasizes the feel of the thing, the messiness of history, the tyranny of coincidence and the vagaries of fate. It requires players to make the best choices they can without any certainty of result.
George Washington was shot at a number of times, but the bullets always seemed to just miss him by inches. In any serious, scientific, and statistical model, this result would be an aberration rather than a certainty; George Washington absolutely would've been shot dead and then, we assume, the Continental Army would have dissolved and capitulated. The way things actually happened was the result of an unlikely but possible outcome - a freak roll on the random events table if ever I saw one.Centering that, and understanding that, is just as valuable I think as illustrating and recreating an inexorable chain of events, and for me as both a player and a designer, it's a heck of a lot more interesting.