Written by Iain Angus, development manager at Total War developer Creative Assembly
Decision-making is a phrase you hear a lot and will regularly see on job descriptions for senior roles in the games industry. Despite this, it isn’t often explored in any serious depth, which is strange as it will determine the success or failure of any project.
In this article we’ll look at what decision-making really means, and how we can make better decisions.
I’ve experienced the disaster we all hope to avoid, having worked on a cancelled project (Fable Legends) that led to a studio closure (Lionhead). Upon reflection of this experience, what I learned is that there was not one single wrong turn that had been made. Instead, there were several separate decisions that were not fatal on their own, but when combined they add up.
When developing a game, you start at the very beginning with nothing but a vague idea. This idea is turned into a product (or a disaster) by a series of decisions, big and small. Every decision, from which engine to use or which publisher to go with to the color of the main character’s hat, contributes to the success of the product. Even a small improvement in the quality of decision making on your team will compound during the course of a project to make a huge difference in the final quality of your game.
One thing is immediately obvious, if you make no decisions at all, or if you make them very slowly, your project goes nowhere and that is a fast track to disaster. Part of the role of the producer is to make both prompt and high-quality decisions and to help the people around them to do the same.
So, decisions are important, we just need to make sure we all make good decisions all the time, right? Unfortunately, it isn’t as easy as that, because decisions are made by:
People
“When dealing with people, remember you are not dealing with creatures of logic, but with creations of emotion, creatures bristling with prejudice and motivated by pride and vanity” - Dale Carnegie
If you think this doesn’t apply to you I’m sorry to tell you that it does. People are emotional and irrational and that affects our decision-making. Let’s look at a well-known example:
"Part of the role of the producer is to make both prompt and high-quality decisions and to help the people around them to do the same."
A study of parole judges found that at the start of a day and immediately after lunch a prisoner had nearly a 70 percent chance of being granted parole. At the end of a day, when the judges were tired and hungry, the chance was close to 0 percent after all other factors had been accounted for. If a decision as simple as yes or no, as important as whether someone should stay in prison and made by people who have been trained specifically to make this type of decision can go wrong, then what hope do we have with the complex and creative decisions that happen on a game team?
Intrigued by this I did some research and came across the work of Antonio Damasio who researched the emotional component of decision making. Upon studying a group of people who had brain damage which meant they did not feel emotions, he concluded that they had difficulty in making any decisions. These individuals were able to understand the better option out of a range of choices, but they didn’t have the drive or motivation to commit to it; they would always sit on the fence. Damasio’s conclusion was that emotions are a crucial part of how we make decisions but, as we’ve seen with the example of the parole judges, emotions can affect our decisions in unpredictable ways.
Automation
The natural conclusion at this point would be to remove the emotional human element from the decision-making process. This is done by systemizing, data-driving and automating your processes as much as possible.
Checklists, A/B testing, standard procedures and KPIs with actions that are applied automatically when the value goes beyond a threshold are common examples of this. They take the emotion out of a decision by separating the creation of the decision-making criteria from the measurement against that criteria.
For example, a game I worked on had a game mode that divided opinion. Some people loved it, others hated it and we were considering cutting it altogether. To help make this decision we ran a focus test that would score the game mode out of 5. The result came back: 3.7, a good score but not a great one. It was inconclusive and didn’t help with the decision. What we should have done was to set the score threshold needed to keep the game mode before we knew the score. Doing so afterwards was impossible.
Automating and data-driving decisions can be very effective but beware of two common drawbacks:
The paradox of automation: if your automated system has an error this error will be replicated over and over. The more efficient your system the faster the error gets replicated. The paradox is that the more efficient your system the more crucial the contribution of the human operators, who are the ones who spot the errors.
The human bias called “What you see is all there is”. We are biased to treat data that we have as a lot more important than things we don’t know and to make decisions based on only the things we know. To combat this, you need to frequently ask yourself “what information am I missing?” whenever you make important decisions. In addition to this, when collecting data from your game define in advance what you will use it for. If you don’t, the temptation will be to use it for something even if it is not useful or relevant.
Despite the drawbacks, automating decisions is a powerful and effective tool. But game development is a creative enterprise and we haven’t (yet) found a way of automating creativity. At some point you need to allow emotional, irrational and unpredictable people to make decisions and there are some techniques to help with this.
Choices
It may seem obvious, but boiling decisions down into a finite number of choices is a good first step, then you just need a method of selecting one of these choices. It is however, worth listing out every possible option, even if it seems ridiculous or impractical at first. After some investigation it might turn out that one of these options is the best choice.
If it is difficult to enumerate all your choices, but in any given situation you can either:
Continue what you are doing
Stop what you are doing
Do something else
For example, if your schedule says that your project is going to overrun its next milestone you can either continue as you are and just be late, stop the project altogether or make some changes. Potential changes could be reducing scope, adding resource or asking to push the milestone date back.
Now you have a list of choices you are ready to choose one of them.
Choosing
There are two approaches to making a choice:
Optimizing - evaluating all your options and choosing the best
Satisficing – choosing the first option that meets your criteria
Optimizing finds the best option possible but isn’t practical when there is an overwhelming number of options, Satisficing is the best approach in this situation. For example, when choosing a game engine for your project there might only be a handful of possible options and Optimizing is the correct approach. However, if you have one vacancy and receive thousands of applications, you might consider saving time by using a Satisficing approach and hiring the first candidate with the right skills and experience.
Optimizing-type decisions are prone to being
No tags.
