Saturday, March 14, 2015

Random game design thoughts


Computers and humans aren't good at the same things; you can capitalize on that. For instance, let's say I have a game that involves flying a rocket through caves. I could try to make an AI that can also fly a rocket through caves and pit you against it, but that would be a lot of work that may or may not pay off. Instead I can put a bunch of dumb turrets in the caves to give you however much challenge you need.

Positive and negative feedback

Positive feedback is paying dividends on investments. The “rich get richer and the poor get poorer” as Jesus says (Matthew 25 and elsewhere). A lot of basic strategy game design is built around this, where you're accruing advantage over time.

The upside of positive feedback is that it drives things to a conclusion. For instance, in draughts when you get a pawn to the opposite side of the board it promotes to a king, with vastly increased mobility. In a first-person shooter each opponent you kill is one less person who can harm you, so the game gets progressively easier. If you weren't continually presented with new people to kill the game would rapidly lose its challenge.

The downside of positive feedback is that your early choices in a game are the most critical, with successive choices becoming less and less important because they're overwhelmed by the dividends from the early choices. For instance, you might have a critical sequence of things to do to get a Starcraft game started with optimal production. In a stealth game when you get discovered, a guard might call for help from other guards, who then arrive and encircle you making defeat a near-certainty. Game designs have to try to mitigate this.

Negative feedback is resistance to change. It's very useful for balancing games. For instance, you might have a set of items available for sale in a board game, and the ones that aren't bought get a little bit cheaper each turn, until they match someone's idea of their value. Another example is when you acquire territory but your troops become spread thinner and as a result it becomes harder to hold that territory. In my 7DRL the last goblin from one chest will run off to another chest; as a result chests you tackle later tend to have more defenders.


I like to ensure that the player can back away from situations when they get in over their head. A lot of times this means the player can run faster than anyone else, but usually you also need other ways to cool down a situation. This is a form of negative feedback, I suppose.

Designers need to spend a lot of time thinking about what happens when the player fails, and how to keep the game rolling through it. The coin-op approach was to just put up Game Over and let the next person have a turn, but with a bigger game played solo it's good to think about how to not have the game just reset when the player fails. Thief 3, for instance, gave you a jail level to bust out of if you got caught in the open-city part of the game. This served as a re-trainer for the skills you were lacking.

Mowing the lawn

This is what I call solitaire-style or match-three sorts of games, where you're not spending most of your time learning new skills; you're just trying to execute existing skills well. It's a low-intensity sort of fun.

Kill them fast

Every second an AI stays alive in a game is another chance for them to do something dumb that destroys the player's suspension of disbelief.

No comments: