Complexity vs. Randomness

This is a followup to this post: Core Cycle. If you haven’t read that yet, read it first.

—-

One of the most important lessons I learned working on the Sims was this: If you can’t communicate it to the player, it might as well be random.

As a new designer, I was excited about developing the most complex, deep system I could imagine. For The Sims 2 on consoles, I’d created a food creation system that had pages of stats that controlled all manner of possible results.

After working on designing the system for a week, I finally showed the lead designer the spec for the system. It had everything! Endless depth! A huge variety of possible results! Little subtle details could affect the outcome in radical ways!

“Simpler.” That was basically the advice I got for how to make the system better.

I was annoyed. The system was flexible, elegant, and deep! Beautifully designed! I mocked the whole thing up in Excel, so that using a simple drop-down list, players could choose from a variety of ingredients and cook them in a variety of ways to achieve a huge variety of results. I’d build the whole system, and show them that it worked!

Huge failure. I mocked up the thing, had one of our in-house testers mess around with it for an hour or two, and the result was clear as day. He had no idea how changing the inputs changed the result. I knew – I knew that a Meat with a Protein stat of +40 instead of +10 would change the result, but there wasn’t any reasonable way to communicate that to anyone – particularly when there were five other stats also changing the results simultaneously.

In the end, the solution was to radically simplify the system. Fewer stats on each ingredient. Each powerup was controlled by a single stat, and triggered at the same threshold for every item. It wasn’t as complex, but it was a thousand times more understandable.

Now, a player would (for the most part) understand that the goal was to pick ingredients that had specific high stats, combine them, and you could get something that would produce a specific power-up effect. Ingredients’ stats were clearly communicated to the player, and the power-up effects had their own visual indicators, and were clearly labeled on the end results. From my perspective, it was almost painfully simple.

The key, though, was that people understood it – they knew how to make repeatable, predictable results, and were willing to “explore” the system because every possible combination produced a useful result.

Players want to be able to understand what is going on. If you get shot in a game, you don’t want to die because you’d contracted pneumonia two weeks before. You want to die because you got shot. This means that the next time, if you don’t want to die, you can avoid being shot – there is a clear causal relationship between the action and the response.

What players want in most cases, is control, or the ability to learn. They want their actions to have consequence, and their decisions to be based on “perfect” information. If X happens, Y will happen. You can make the system relatively complex, as long as players feel that it’s ultimately predictable – like they *could* understand it given the right circumstances, even if they don’t understand it yet.

There is a threshold, however, where a system is either so complex that it’s completely unpredictable, or where a system’s inner workings are communicated so badly to the player that the result seems random – in which case, no causal relationship can be formed, and the result can’t help a player form new decisions. The end result is both frustrating and useless. Not good.

As a designer, you have to go back to the basic razor: “A game is when a user is presented with a compelling choice that allows him or her to make an informed decision that has consequence.”

Randomness means that any decision point associated with that randomness can’t be informed – because it’s… well, random. This makes the player’s involvement with the game essentially meaningless. No information matters. No choices matter, and there’s no consequence that you have any control over.

Again, it comes down to three things – consistency, communication, and consequence. To form causal relationships, the game’s systems need to be consistent. Not “predictable,” where the player understands them so completely that they can predict the result with 100% accuracy – that’s boring. But they have to establish a set of rules, and then communicate them clearly to the player.

And every step of that is helped when your system is simpler.

Next: The Four Stages of Grief.