Failure and Learning

[This Article was Originally Published on AGamesDesignBlog.com]

There is a belief that a good game never punishes players and only rewards them.

It's a belief that often causes debate amongst Designers because the phrase is usually reinterpreted to mean "The player must never fail." and this couldn't be more wrong. Failure does not equal Punishment and in fact Failure is an intrinsic feature of all games and forms of play.

One of the biggest problems is that failure is thought of as being a negative thing: "If you fail the mission then it's Game Over". This is only a correlative relationship however in that Punishment (which is almost always negative) often follows Failure (which can be positive). "Cum hoc ergo propter hoc" as somebody who understands Latin might say (I don't I just stole it from Wikipedia to look clever).

Failure is a Positive experience when it is possible for us to learn from it. This may sound a bit like a line from a self-help book but it's something that is worth emphasising.

When most gamers buy a new game the first thing they do not do is read the manual. Instead they will load the game up and immediately jump in and start trying things out to see what happens. When something works they add it to their mental model of how the game works and start applying that knowledge to future interactions. The same is true when something doesn't work. With every interaction the Player is refining their mental model and making it more accurate.

Valve found that Play-testers for Team Fortress 2 weren't as upset when they died if they were shown who killed them (and the surrounding context) and told how they had improved. By providing Players with this feedback on why they just died they were helping the Players learn from their mistakes and reinforcing their own progress. They had managed to turn dying from something that was relatively punishing into something that was beneficial. By dying you were learning.

Take a racing game for example. You step into a car you've never driven before and you start out by driving it in a similar way to the last car you drove. Using this base experience you start to develop a mental model of how to drive this new car (and I should point out that it is very rare that the player is actually aware of this information). Both success and failure help you refine this model. You start to take corners faster and faster and with each success confidence in your model grows. Similarly if you spin out or hit a wall the failure helps establish the limits of that model. Failure because of over-confidence in your own mental model is a positive gain. The failure is your own and fits within your model. Both Successes and Failures that contradict your mental model result in dissonance and confusion.

Key to this learning process is that once you have failed you are able to test out your new model quickly. If you cannot affirm what you have just learnt then it is very likely that the experience will be forgotten and the failure repeated later.  Respawning in TF2 is often very swift and you can nail that corner on the next lap.

This somewhat matches something Clint Hocking said at GDC

The fact that failure didn't mean the game ended or that you had lost a lot of time preparing and were now unable to continue meant that Players are much happier with trying something new and potentially failing. The short turn-around on being able to integrate changes into their model meant that failure was a Positive force which helped the player learn.

Failure or a sudden change in circumstance is often a trigger for improvisation. When your plan fails you have to call on that mental model of the game again and respond.  Part of the feedback loop that makes games so entertaining. A perfectly executed plan has it's own rewards but success in the face of failure can be even more rewarding. Being able to adapt quickly and effectively requires many more skills than just solving a static puzzle.

reset

Take Reset by roBurky.

A game that lasts about 3 minutes and never punishes the player for Failure. A game that is very challenging despite not actually having any formal win or lose conditions. You will always reach the end of the game but along the way you will have been subject to both success and failure. At no point does this make the game any less enjoyable as either an experience or a game.

It's worth noting that I ended up playing the game 5 times just now before remembering to take a screenshot.

Here every time you hit an asteroid or a missile it is a failure. The ships controls alter as you take damage with it favouring either the left or the right which forces you to adapt your model as you play. Each new element is introduced carefully allowing you to experiment with them in relative safety letting you build a model of how the element works and how best to deal with it. Then as all of the elements combine together you are forced to improvise and adapt using what you have learnt.

This game is a perfect example of how Failure isn't always a negative experience.

You don't need to be able to lose for a game to be enjoyable or challenging. You just need to be able to fail.