How game developers can predict the future with correct forecasting techniques
/
Our Production Director, Peter Willington, recently published a piece with Games Developer around forecasting and predictions when making video games. Good forecasting is hugely important to what we do, and our values, as good forecasting allows us to make great games without video game crunch.
Below we have an intro about predictions and forecasting from Peter as well as a link to where you can read the full article from him on Games Developer. If you’re interested in joining the games industry, how Auroch Digital works as a studio, or understanding how games are made, give it a read!
Stories about prediction ☔
Once, I bought a ticket for The National Lottery. You picked six numbers from a list that went up to 59. Then, on the night of the draw, after a quick prediction on who would win by Mystic Meg, a button was pushed beginning “the drawing process” from a giant machine filled with numbered balls, which would then randomly pick six regular balls and a “Bonus Ball”. If three or more of your numbers were drawn, you won a prize; the more numbers you got right, the higher the pay out. That day, I predicted the numbers of three balls, and I won a tenner.
This was, for all intents and purposes, a game of chance. Sure, it might theoretically be possible to predict which numbers would be drawn - either you believe it’s possible to have perfect knowledge of the physics at work and the state of all the atoms in the universe at the time of the draw, or you believe a TV astrologist from Accrington really was gifted the ability to read the future - but let’s just assume for now that neither of these things is possible. Winners of the lottery are able to predict which numbers will win, but their “method” is sheer luck.
Another time in my life, I went to Bristol city centre, where Auroch is based. I took a raincoat, because the weather forecast app I used claimed there was an 80% chance of rain. The app I use is generally right, so I predicted I would need a raincoat. It rained that day, so I got that prediction right too.
Prediction is not one thing 🤞
See, we often think of prediction as being a series of definitives: your prediction was either right or wrong; an event can either be predicted or it can’t; the value of types of predictions are worthwhile or worthless; you use one method of predicting or you use another; prediction methods work or they don’t. But as the examples above demonstrate, prediction actually sits on a series of scales.
For the lottery, there was almost 0% certainty I could predict which numbers would be chosen by Guinevere, but I still won, and my prediction model (guessing randomly) was the same used by millions of others who didn’t win. Not only that, but there are degrees of success; my “method” won me £10, but it could have won me £10 million.
My weather app defined an amount of certainty for rain. It didn’t say “it will rain” or “it won’t rain”, instead it effectively said “80% of the times we have this data, it rains”. And even then I added to this prediction an appropriate weighting to the app itself’s ability to predict weather accurately.
Prediction then, is complex and messy.
Predictions and video games 🎮
Making video games, or any commercial art, is also complex and messy. During development things go wrong, a turn of events up-ends everything, assumptions are challenged, consumer tastes change. This isn’t a complicated environment we’re working in, it’s a complex one.
But this doesn’t mean that we’re playing a game of chance when it comes to predicting many of the things that we care about when making video games. In fact, I would argue that we can get reasonably good at prediction within the medium, and I think forecasting is one of them.