First I would like to say that no matter what one thinks on No Man's Sky, it's virtually undeniable that the game had massively failed to meet even the basics of what it promises. From the 71 average rating on Metacritic (and 4.6 average user rating) to having only 35% of the Steam users giving a positive review, the game has been one big disappointment. I say this as someone who had deeply and sincerely wished for this game to be a sweeping success and the model for future indie games. With less than 4,000 concurrent players still playing on PC, it should be clear that at this point it makes absolutely no sense for Hello Games to dedicate months of development towards making this a good game, and the reason they have been so uncommunicative is likely because they're simply waiting for the vitriol to die out before they move onto the next venture. At this point I believe the game requires a proper post-mortem and a "lesson's learned" to any other similar projects.
No Man's Sky is a game that captured everyone's imagination for one reason: It promised infinite variety to explore. Hello Games recognized a very peculiar aspect that appealed to gamers, that gamers frequently listed huge open world games like Skyrim, Fallout and The Witcher as being their favourite because of the scale of the worlds that allowed free exploration. The developers saw an opportunity: take this free exploration dynamic and expand it to galactic proportions by creating a whole universe to explore. This also specifically appealed to a certain demographic because it included the "sci-fi" element - one would be exploring an entire universe like Neil deGrasse Tyson does in his hit show Cosmos. This sense of wonder towards discovering the novel and unknown was repeatedly iterated by Sean Murray.
So what went wrong? How did the game manage to fail?
The key isn't the small team or any specific feature as some suggest. Minecraft and endless other sandbox indie successes had much smaller teams, yet managed to sustain continual success. No Man's Sky on the other hand fails on the basics of what it means to create a worthwhile sandbox game: Creating a virtual world that feels alive rather than a static set piece.
No Man's Sky chose to focus on one thing above all others: maximizing the NUMBER of permutations, rather than ensuring that there were enough interesting permutations that made sense within the environment, and told an interesting story worth seeking out. This was likely a marketing based decision, making a claim like there are 18 quintillion planets makes it incredibly easy to sell the game to those who don't realize how permutations work (ie. most gamers). If you took 11 categories of an object, and made around 50 variations of each category, you have about 18 quintillion permutations of that object. If the average gamer realized this, then No Man's Sky wouldn't be the pre-order success it was, as they would have realized that the heavily marketed number of 18 quintillion unique planets isn't anywhere near as indicative of variety as they imagined. Similarly this blind objective of maximizing permutations by allowing as many combinations of disparate parts as possible is seen in the design of animals which are supposed to be a major draw. Yet again the animals end up being a disappointment due to having absolutely no evolutionary history or context within their environment. There would have also been the potential of allowing you to capture these animals and sending them to your own zoo, where one could breed them and continue their evolutionary history. But in No Man's Sky, the animals aren't actually part of any environment, they're as if someone played slot machine randomization with animal body parts.
If you look at the vast open world games that do have interesting worlds to explore you notice that they don't try to maximize permutations, but focus on 4 things:
Creating game world that tells a unique localized story through its environment within each area - This is most apparent in a game like the Witcher, where the designers created each landscape feature by hand to tell a history of what that place represents. From the clear signs of erosion among deforested areas to flooded ruins to carefully placed buildings that make sense within the economy of the villages, the game has signs of a history all over the landscape. No matter where you are in the world, the landscape itself tells a story of what that place is and who lives there (monsters and humans alike). Whether it's an abandoned fishing village that has signs that directly tell you what type of monster chased away the villagers, or the depressions within the stone steps of of a busy walkway, or the unique carving within a monestary that tells a story of monk trapped within the dungeon below, or the way the wagon tracks on a road get deeper within a swamp area, the actual landscape has a history. No Man's Sky completely fails to portray ANY sort of history within it's environments, with none of the landscapes reflecting a narrative on how that landscape came to be. From the design of the landscape to the design and placement of the NPC structures, the game needs to tell a story. In No Man's Sky they're randomized Brownian noise. In No Man's Sky there are near identical alien outposts on every single planet, and identical looking monoliths. It didn't have to be that way, there is no reason why the algorithm needed to have outposts on every planet, or why the algorithm couldn't have a plethora of unique temples on certain planets. The localized history that exists in each region of The Witcher is entirely absent.
Creating AI systems complex enough to create emergent gameplay - A series of rather simple NPC affiliations and behaviours can lead to some truly interesting stories, and No Man's Sky completely lacks this. If you look at a game like Skyrim, the reason it was so interesting to explore for so many people was because of the random NPC encounters leading to some unique AI behavior between different groups of NPCs. For example one can be walking around a forest and suddenly see a giant herding a mammoth, who stomps to tell you to back off. But then suddenly a dragon appears and starts fighting with the giant, who sends it into orbit with his club after it lands near his mammoth. Out of nowhere a pack of wolves appears and attacks the giant to protect their den nearby, and a stormclock soldier who is scouting the woods joins by shooting the dragon. Another good example is the recent Deus Ex game, where AI systems can stack on top of each other to create some very interesting emergent gameplay as guards end up confusing each other as to your location. Not only that but your affiliation and reputation carry over to future encounters. No Man's Sky on the other hand has no AI system's that stack affiliation and reputation. Each encounter with a NPC is entirely shallow and has virtually no consequences in the grand scheme of things.
Creating a satisfying gameplay loop that leads to genuine progression by unlocking new areas as a reward for grinding - This is ultimately a game of grinding, which means that there needs to be some genuine novelty provided as a reward for that grinding. One of the biggest disappointments in No Man's Sky is that it completely disregards the obvious gameplay loop that occurs when you make certain planets locked off initially. It's obvious that there is a great gameplay loop that would lock certain planets (Too poisonous, too cold, too much gamma radiation, etc.) without specific upgrades that allow for exploration of that planet. For example, there should be a class of planets that have some sort of weird radiation (let's call it Zeta radiation), that should be completely inaccessible without a Zeta modifier upgrade using specific rare materials to your suit that takes care of this specific radiation. No Man's Sky on the other hand makes ALL planets accessible from the start, making all of the survival aspects dependent on the "life support" stat. This is a very common gameplay loop, it's why Fallout has the high radiation areas that kill you without special anti-Rad medicine or anti-Radiation suits. For example the Glowing Sea in Fallout 4 is one of the most exciting areas to explore specifically because it requires a careful balance of medications or specific equipment in order to explore. If you ask many people, exploring these dangerous areas and carefully keeping tabs on radiation meters and various other survival indicators are some of the most tense and exciting parts of the game.
Focusing on making the NPCs tell coherent story of the game world's history - this is perhaps the biggest reason why the universe of No Man's Sky feels so hollow and empty. The simplistic interactions between the NPC and players lead to the NPCs being nothing more than item dispensers, with no true grand motivations or agency. This is what kills immersion in gaming, when the actual characters don't come across as having the agency to seek their own self motivations within the context of their own society. Instead they randomly hand out upgrades (which you likely already have) for performing simple choose-your-own-adventure puzzles. They are all alone in their outposts and space stations, without any social life or indication of how they fit in the grand schemes of things. This is a huge missed opportunity, because there could have been all sorts of interesting quests that help each faction get an upper hand on the others. Also not having planets with dense populations and cities is a big misstep.
When you combine these four factors, what we see is that ultimately what makes exploration exciting in games isn't the sheer number of object's in a game universe, but the number of unique histories told within the universe. No Man's Sky really didn't understand this at all.
All this said, I do hope we do get another game that uses procedurally generation as a basis for creating a huge world, but that focuses heavily on making an algorithm that creates regional variety and NPC AI complex enough to make each area create emergent gameplay.
Maybe next time...