My personal take on why procedurally generated universe with an insane amount of same-ish real estate and a streamlined routine gameplay is the wrong way to go in designing a space-sandbox game.
Desperations of a space-sandbox fan who’s still playng Master of Orion II
By MARTIN REZNY
(a repost from my blog, click here to check it out)
Short answer:
I haven’t had a gaming rig that could play current games since 1997.
Long answer:
Alright. After reviewing a lot of game footage, reading a bunch of reviews and think pieces about this game both pre- and post-launch, and watching some rants from Youtubers who’s opinions I respect, I don’t think I’d play it even if I could. Which, if you knew me, is saying something. I’m a massive sci-fi fan.
I don’t even care all that much whether it is particularly good. I can get excited about terrible, boring-dumb sci-fi movies or games if they just have one or two cool ideas I’ve never encountered before, or anything done better than anywhere else, or at least unintentionally well. I make it make sense in my mind oftentimes when the authors clearly put in things totally cluelessly.
If you know anything about No Man’s Sky, I posit I can read your mind right know — but Martin, what about those 18 (or however much) QUINTILLION different worlds! Isn’t that a textbook example of a cool, never-before-done sci-fi idea? To which I’d say, it would be, if they were unique. Which they ain’t.
Instead of a game review, which you can get tons of elsewhere, let’s talk about the philosophy of game design in a science fiction universe. For that purpose,No Man’s Sky is an excellent conversation starter, in my opinion. The obvious issues with overhyping this thing sky high (no pun intended) set aside, I think I should start by sharing with you my main frustrations with what TotalBiscuit is talking about in his rant about this game — concept of “desperation genre”.
Desperation genre, in this case space-sandbox, is a genre that has not yet been done well, not even as an aspect or a facet of a game falling mostly into a different genre (such as RPG, survival, or my favorite, turn-based 4X strategy games like Civilization, which are some of the better combos in this case). Sometimes, it is because there is not enough demand (= money), but in other cases, the problems are technological walls, or that it is just too damn hard.
Space, as I have said a number of times already, is hard, and it is certainly true even if you only try to simulate one. One way to overcome the great difficulty of creating a massive universe filled with unique planets is to generate it procedurally, which means that you set some limiting parameters that allow for a game of a certain kind, and then let the script run crazy many iterations, producing all the different variations of parameters within the limitations.
Kinda like snowflakes come to be in so many unique shapes — limited rules for what can happen during the process, minuscule differences in starting conditions, and chaotic random variations over time leading to one of great many unpredictable end results, which is however still totally recognizable as a snowflake. The problem with this approach is that it only creates many versions of what is essentially the exact same thing, and that gets old fast.
This problem is of course not limited to a space-sandbox genre, it can often be found in fantasy games as well, since many fantasy RPGs use procedural generation for dungeons. Don’t get me wrong, it is an important tool, great, when used sensibly, like any other. One of my all time favorite game series,Diablo, has used it to a strong effect, but also probably to its maximum potential. I have played Diablo II and III for at least ten years now, and I’m sorry to report I think I have reached peak-procedurally-generated-dungeon.
As someone who must have run through procedurally generated dungeons literally thousands of times like some lab rat, I can tell you what it is like when you hit the moment that will break this kind of game for you. It is when you realize that you have seen all the limits of the procedural parameters so many times that you have memorized them. In both Diablo sequels, I can now tell, without thinking, the way out of the maze by seeing just a tiny corner of it.
Not perfectly of course, I’m not an alien robot from the future (I think), but enough to get superbored after a while. The amount of game time necessary to reach this point will likely vary from game to game, and from person to person. From what I’ve heard and seen of No Man’s Sky so far, it looks like it is going to be somewhere between leaving the first planet and visiting about half a dozen more. Relatively simple and routine gameplay is not helpful at all.
You see, if you compare Diablo II and III, many long time fans vastly prefer game system of the former, and I think that the procedural generation fatigue is the main reason. If you base your game on procedurally generated places, you actually need to combine it with a gripping story AND a vastly variable gameplay, so that after the exploration and orientation become routine, the players have things to distract them and change the pace, as well as an ability to break the monotony by taking over control away from the randomizer.
Diablo III has made the mistake of holding the player’s hand too much by forcing the skill unlocking progression to be always the same for each class. They have fixed it somewhat by realizing that now they can only increase the replayability of the game by introducing a ton of new unique items that force players to much more frequently change up their skill-build to best utilize those items. But once each player picks up most of the items one too many times and memorizes all that can drop, the boredom will kill the game again.
So to put my main problem with No Man’s Sky very concisely, it’s spaces over experiences. That could work, if the game went all out on spaces and was actually about exploration, but then it would have to contain interesting things to find, or to have vastly varied environments, and not have everything already inhabited by aliens with a standardized commercial infrastructure.
Since, as I understand it, pretty much every planet already has aliens on it so that you can trade with them for gear, there cannot be any sustained immersion to make you believe you are an explorer. Interesting things to find also cannot be procedurally generated, which means that making the universe bigger will only dilute their presence. And finally, having each planet have single climate out of maybe a dozen Earth-like options is very little variety.
Exploration game would need aquatic planets with unique underwater gameplay, toxic hellscapes with unique hazardous environment gameplay, etc. You would also need entirely different alien ecosystems and civilizations with different psychologies and therefore behavior patterns, which is way more important towards fighting boredom than to just have them look different.
As for what I mean by the contrast between spaces and experiences, the difference is that a spaces-heavy game gives you many different locations to go to where you can always do pretty much the same thing, while experiences-heavy game may only give you the same single arena every time, but many different things to do in it. A truly great game will try to do both.
For my (imaginary) money, the games that came the closest to having many different places to be AND many things to do at any of them were Morrowindand Master of Orion II, even though they excelled at it in different ways.Morrowind pushed the design approach of hand-crafted big playground that still keeps all of its details unique to the limit, while having many different dynamically changeable classes to play with, as well as complex systems of custom spell and potion creation. You could even bend rules of that world.
Sure, it would be impossible to balance Morrowind for any kind of multiplayer, but hey, No Man’s Sky is really only a singleplayer game anyway. Allowing the player to become a god when they understand the game mechanics very well is not only not a detriment in sandbox, it can be a lot of fun, especially if the game has scalable difficulty. However fun Skyrim is, it is a little bit too streamlined, and in the end, that’s why it’s bound to get old faster.
However, you cannot have several quintillion hand-crafted planets in any universe. While this approach works for medieval fantasy, which is by its nature very localized in scope and small in scale in comparison to space sci-fi, maybe a different approach is needed for space-sandbox. In my opinion, the very best game of this kind to date is still the Master of Orion II, and game developers should at the very least try to rip off its design features a lot more.
In this game, you always start with one game board, but never the same. It is a map of the galaxy with a limited number of procedurally generated star systems. Not topologically complete planets of course, but that’s key — don’t vary unimportant kinds of details, vary meaningful differences. Here, there are many different classes of planets based on a lot of real science. They come in all the different sizes, with three settings of gravity and of mineral richness.
Each of those different parameteres have meaningful consequences in terms of strategy and overall gameplay, and you can’t even predict how many planets will there be in any given star system, except for little hints that reward understanding of underlying science, and those who pay attention. The end result is also a great number of unique combinations, and here they even matter because they change player’s experience every time they change.
In addition to that, Master of Orion II has great custom race editor that allows for several entirely different overall winning strategies (research, production, combat prowess, diplomacy, espionage, population growth, etc.) that are balanced pretty well between each other. Not to mention its ship design, best thought-through tech tree ever, and tactical spaceship combat that almost all later 4X space games dropped or made suck by making it uncontrollable.
The worst things I have heard about it were that it was ambitious, demanding, not very actiony, and time-consuming, as complex turn based games often are by simply being themselves. Space sandbox doesn’t have to be turn based or a strategy. It can be a flight simulator, or a survival FPS, or anything. But whatever it might be subgenre-wise, it would still be made into a superior game by having meaningfully varied places, aliens, tech, or space fights.
To sum up, No Man’s Sky is largely varying all the wrong details, things that don’t matter to the gameplay, while it repeats the things that need to be varied, to an absurd extreme of a contrast. There is also one more perspective on how more is sometimes not better — generating the whole planet is useless if it’s essentially the same all over. If you must, you should generate more places to go, not more space no one in their right mind has any reason to visit.
I’d say it would make much more sense to create a varied system of places that make sense visiting, like space stations, cities, mines, rainforests or coral reefs, ruins, that sort of thing, instead of making more empty planetscapes. You are already making each planet smaller than it is in real life because a real sized planet of mostly nothing interesting is absurdly impractical and unplayable. A game where one can explore real sized looking planets by going from a logical single location to a logical single location is actually more realistic, anyway.
As a space-explorer, you wouldn’t just randomly walk around planets hoping to stumble onto things, you’d be sciencing the shit out of where the interesting places are from orbit and then go land only at those. All of that can be procedurally generated as well, in great part at least, and I argue it would make for a much more satisfying and immersive gaming experience. You should also be given aims, many different ones to choose from, so that you feel neither aimless at any given point in time, nor not at all in control.
In a first- or a third-person space-sandbox game, you should also be allowed to choose something like a class, ideally dynamically shift between many of them, each of which has specific objectives attached to it. It doesn’t have to be super overt or explicit, but it should be intuitively graspable and tangibly possible, so that even going to the same kind of place won’t ever be exactly the same experience too many times. No Man’s Sky doesn’t do this quite enough.
Whatever doubts you might have, all of this is entirely possible, it was already done rather well in 1996 in Master of Orion II. It doesn’t require much extra computing power, this one falls only into the “it’s hard” category. You simply have to devote some time to thinking through the typologies of available locations, actions that can be performed in them, character classes who perform them, and technologies they can use to do so. Then you can procedurally generate meaningfully variable gaming experiences forever.
The as of yet unsolved problem is that of how to combine the spaces and experiences generation with a well-told story as well, a notably lackluster dimension of Elder Scrolls games, almost non-existent in 4X strategy games. Unfortunately, that is a topic for a standalone article, or a book series, by itself, so I will have to share my thoughts on that some other time. If you think I don’t know what I’m talking about, maybe I don’t — be sure to let me know.
Read more about:
BlogsYou May Also Like
