Searching for design

This Problem Changes Your Perspective On Game Dev looks at game design as though it’s a search algorithm touching on:

• Speed vs. accuracy trade-off
• Local minimum
• Infinite search space
• Wrong reward function
• Noisy measurements
• Exploration costs
• Multiple captains

Jonas Tyroller talks quickly and makes an argument for discovery rather than one off design. While I’m interested in game design are there lessons to be learned beyond that?

Is game dev different?

Jonas is suggesting this for games not for, say, spreadsheets. For games, in particular smaller indie games, the goal is to make a game, any game, that appeals to gamers. The restrictions on what the game should be are likely to be:

• Personnel – What programmers, artists and similar are in the team?
• Time and money – How long can the team support itself?
• Platform – What can the desktop, laptop or console do?

That might be it. If you’re trying to make a living then as long as you attract enough users, keep them long enough, and deliver the product then you can do it. The genre, themes, graphics, control and whatever could all be flexible. A big search space means a lot of opportunity to find something different. That means spending time, early on, searching is valuable.

I suspect this becomes less true the bigger the game and company is. If you’re making a sequel then you probably need to deliver something similar to the earlier games but “better”. That probably means bigger in some way. If you’re dealing with a publisher or a licensed intellectual property then the deal may come along with conditions. Maybe it has to be a 3D shooter and the final plot has to be signed off externally. That would greatly restrict the potential search space and make this approach less useful.

Often the inovative game ideas are come out of the indie market. The mainstream games market tends to just make small variations on what has come before. Frequently picking up the ideas found by indie games.

How open is your space?

Looking at design as a search algorithm seems more relevant for more open design spaces. A spreadsheet is a known design space. New or updated entries are likely to be similar to those that have come before. Prototypes probably won’t find much new. However if your product is a new idea or an old idea taken on to a new platform maybe there is room for innovation.

Moving to a mobile platform probably was an instance of this but may not be any more. There are still opportunities out there smart TVs or speakers are still relatively new. Simply going with your first design for one of these could be missing something much better. As for new product ideas I can’t really tell you what those are. Whatever they are some experimentation is likely beneficial.

Multiple captains

I won’t go over all the details of the video. If you think your project has room to benefit from a bit exploration then go watch it. I will highlight one section.

Jonas had some useful things to say about the issues with multiple captains in your project. If one person is searching for a design that’s easy. They try a few prototypes and pick the one that’s best. If there’s more than one person that can be hard. The time invested in a prototype and familiarity can’t help but produce some bias. If discussion doesn’t produce an obvious winner it can be useful for the captains swap prototypes. This get give each captain experience in and investment with another option. With a better view of both sides discussions can have less bias and, hopefully, be more productive.