No Man's Sky is having a grand old time in the public eye as of late, what with being featured on multiple show stages in the past year as well is being the cover story in the latest issue of Game Informer. Part of Game Informer's coverage of NMS is a [url=http://www.gameinformer.com/b/features/archive/2014/12/05/take-a-30-minute-behind-the-scenes-tour-of-no-mans-sky.aspx]half-hour demo from Hello Games' Sean Murray that shows off some of the tech behind the game.[/url] A large part of the commentary focuses on the NMS's use of procedural generation to create its content, and I thought it might be interesting to get some community discussion about that. So what is procedural generation? Well, to put it very simply, it's a way of creating content by using algorithms rather than by having a designer or artist create it by hand. An easy way to understand it: I can define a simple math function of f(x)=5+x. For any value of x that I input, I will always get an output that is unique to that input. Procedural generation algorithms work in a similar way, except that they use far, far more complex equations with many variables. This allows for huge amounts of varied content that can be created through an equation rather than spending lots of time and resources individually creating each piece of content. Consider No Man's Sky. Every piece of content in that game (according to Sean Murray) has sort of "sliders" like you might see in a character creation tool- terrain, rocks, trees, creatures, ships, and more. Procedural generation adjusts the components of these objects to create all kinds of variations on the same basic piece of content. A question that was posed at about 6:00 of the video I linked above (by Game Informer's Ben Hanson, I believe) to Sean Murray is "Why haven't other developers done this before you?" Personally, I think that question is a very interesting one to discuss, as it can reveal some of the process behind designers and games. I think a lot of developers have a very clear idea of what exactly what they want players to see as they experience the game, certain images in their mind that they must show the player, and traditional content creation is the best way to make sure the player sees this complex thing that they are imagining. On a similar note, procedural generation can sometimes only go so far. If you want to keep your world aesthetically pleasing, you might have to limit what your algorithms can do, which could possibly lead to very similar or replicated content after a while. Sometimes it's probably for the sake of gameplay- if you're playing a shooter, it's very important that combat spaces are designed to be fun and challenging, and that means taking direct control over the content. Perhaps it comes down to pure hardware power. Sean Murray describes procedural generation as a "next-gen" idea, noting that traditionally in the video game industry, "more power" means "make it look prettier." Deciding instead to focus the power to a fully realized world around you requires much more powerful hardware than we've had in the past. Of course, we've had things like Minecraft that have done that, but not to the extent that No Man's Sky does it, and certainly not as attractively in visual terms. So, I turn the discussion to you folks. Why do you think procedural generation hasn't been used to this extent before? Would you like to see more of it? What games and ideas do you think would benefit most from using procedural generation?