I was thinking about games where you build stuff. There's a lot of games where you build stuff.
One of the core elements of a Building Stuff game is the limits that get put on your ability to build stuff. As an easy example, if you look at a video of a castle in Minecraft, you might go "oh, I guess that looks rather nice".
Then you find out he did it in survival mode on hard and your opinion changes to "holy shit that's insane".
Sometimes these limits are on the materials you have available to you, as in our Minecraft example. You have X resources of type Y today. Then the game tends to actually be more about obtaining more resources - the building stuff section is usually pretty flat in such situations. This is often extended to be a more complex gating system - you can't even obtain the Z you need to build Q until you scrape together N pounds of L. This kind of progression-based system makes for a good self-directed open-worldy sort of experience.
Sometimes the limits are on the scope available to you. For example, you only have a certain limited amount of space, or you have to work within a specific topology, or so on. These are typically level-oriented games like Evil Genius rather than open-ended games, because fundamentally having a limited scope means that you'll want to change scopes sometimes. (IE, change levels.) Of course, many games combine the two kinds of limits.
A third kind of limit that we tend to overlook is how much effort is required to build something. For example, in Minecraft, even if you play in creative mode it's going to take a lot of time to put things together. You have to place every brick, even though you technically have all the bricks you'll ever need. This is also very popular with social games, where it simply flat-out states that you need to wait N hours before you can build something. Often combined with the other limits.
Aside from all of these limits are the final goals you can aim to achieve.
In most such games, self-expression is a major goal. To keep bringing Minecraft back into the equation, 95% of the cool things constructed in Minecraft are self-expression on top of a small kernel of functionality. This is a good goal, but it's just one goal, and you should have others.
Another goal is expansive capabilities. Allowing your constructions to help you in the game world is a major benefit. "Expansive" sounds militaristic, but in this case it can and often does mean exploration and resource creation. A farm is an expansive construction because it produces valuable resources that you'll use. A minecart transport system is expansive because it helps you get around the world easier. A classy outfit will allow you to enter the high court regions. And, of course, a massive cannon you can carry around with you is expansive because it helps you kill monsters. Building "stuff" does include non-buildings.
Expansive capabilities can be a bit of a limited appeal, though, because they more or less transparently target specific game elements. For example, allowing the player to build a better gun is normally not going to involve any kind of self-expression. They just choose the next gun up the tier and shove resources at it until they get an exact duplicate of the same weapon everyone else is using. Similarly, farms tend to have a few optimal arrangements or even simply be reduced to a tile labeled "farm". To allow players to have extended fun with these expansive things, you have to make every expansive thing have a large number of tweakable elements to it. Think Borderlands guns. There is no "best" gun, because every gun has a variety of strengths and weaknesses, special effects, and ammo preference. You can implement this for everything, making every construction unique.
Another goal is defensive capabilities. How well the things you construct hold up to the rigors of the game world. In Minecraft and Terraria, this would be how well your house keeps monsters out. But it can be made a lot more complex. For example, building a space station might involve how well it survives asteroid strikes, alien invasions, power failures, and so on. Sometimes these are simulated events - IE, they don't actually damage your construction. In other games, your construction actually gets damaged and part of the game is building it up again after every challenge. Bridge building games are a good example of the simulated kind of defensive game.
More games are starting to come out with defensive constructions as core gameplay. However, these days there is a twist: the defensive play is often aimed at other players rather than at some set game world challenge. So other players get to invade your fortress and try to get your gold, or whatever. I think this is great, but I don't see why it is limited to defensive play. Just hold onto that thought while we move into another goal:
The last goal I want to discuss is complexity. Having complexity in how your constructions can work is really fun.
As an example, in Minecraft there is little complexity in your constructions. It's a struggle to even have doors that automatically open. Structural integrity is not important and not calculated, so you can have floating floors. There's no NPCs simulated, and creating reactive elements is quite difficult.
On the other hand, there are a lot of games where complexity is actually an issue, and I think we're going to see more and more of them. For example, in a bridge building game, every element is physically related to every other element. Bad design in one area can cause the whole bridge to collapse.
Similarly, designing dungeons for other players to explore (or NPCs) means that the interconnectivity of the rooms is as important as what rooms you have. Actually, even choosing what stats and skills to get as you level up in a MMORPG is this kind of complexity, as certain skills and stats work better together than others.
One problem with complexity is that it creates optimal builds. This is really clear in MMORPGs where a character you build to express yourself turns out to be awful at actually playing the game. Similarly, there is a specific optimal circuit for putting out redstone clock pulses in Minecraft, but at least there you don't have players tying their self-expression to the circuit design.
Keeping complexity from stagnating is always going to be a challenge. The basic rule is that if there is a pretty obvious optimal pattern, it either needs to encompass all the variability so that players can express themselves, or it needs to be done automatically so players don't think it's supposed to be complex. For example, if you're laying down power relays, assuming that terrain isn't much of a factor, it might be best to simply allow the player to "paint" the region he wants to put power relays in and then automatically plant the relays for him.
This brings us to a few paragraphs ago, where we were talking about defensive constructions used to defend against other players. Why don't we use that for offensive, self-expressive, and complex play as well?
Let's posit a game.
You have a world of your own. It's a crap world, uninhabitable. You got it on the cheap. You live in a little bunker on the planet. Your job is to build the planet up as you prefer - or you could just focus on yourself, your bunker, interstellar politics, hacking other players, etc.
The construction is both large and small. You will often build up your bunker, of course. Not only to change the interior to better suit you, but also to have better resources for managing your world, increase the number of NPCs in your domain, etc. But you'll also spend a fair amount of time building on a larger scale, just popping down roads, factories, mines, launching pads, relay towers, terraforming centers, clone labs, NPC villages...
The key here is that the things you want to construct don't really cost resources. They cost effort.
If you want to put down a farm, you just plonk it down. But before it becomes functional, you get assigned a busy-work quest whose length is roughly equivalent to the value of the farm. Busy-work quests vary. For example, depending on your luck you might have to clear the farm out of space-arachnids that settled in the instant it landed. Or you might have to weld together the eight leads connecting the farm to the power grid. Or you might have to hack the farm's mainframe and replace the misconfigured software modules. Or you might have to rewire the nearby relays to accept the farm's slightly wonky notification protocols. Or you might have some combination of two or more - "space spiders knocked the nearby relays out of alignment..."
So a new player could plonk down five million acres of farms - just click "farm" and then highlight a whole continent. But he'd have millions of years of busy work to bring them on-line.
Obviously, a character's build would be all about the kind of busy work they are best at. If you're largely going the solo route, you might prefer to be moderately good at everything. But if you have some friends, you might make it so that each friend was good at a particular kind of work. If you plonk down the farm and it needs a software upgrade, you just go and handle it yourself because you're the hacker. But if you put down the farm and it's covered in space spiders, you might ask your friend the space marine to deal with it for you, inviting him on to your planet.
One key to this is that you can post these quests as publicly available, with a reward and priority of your choosing. This kind of farm might be a half-hour quest aimed at level 7 characters, and you might offer $200 space gold if someone comes along and does it. A person or group can accept and attempt it even if you're off line and have never met them before. They can't screw up your world (and they probably can't access your bunker unless you're unusually permissive), but they can see your planet and how you've got it set up.
As you grow to have a lot of construction, you may actually end up plonking down millions of acres of farms and then just posting those jobs with a $1000 reward each, relying on the willingness of people to come over and do that work for you.
All of these things are connected via topological connections. That is, the farm is connected by a road and a power relay system and a data relay system... unless you've specifically built the more expensive self-contained farm. So how you build your world matters as much as what buildings you put on it. Similarly, all the structures and computers and outfits and whatever all have a lot of cooperating statistics to them, like the guns in Borderlands. This farm doesn't simply produce X carbohydrates per day. It requires or produces a variety of soil-related chemicals, it requires a certain amount of energy, it generates a certain amount of heat and pollution, works best with a certain amount of sunlight or rain, and so on.
Of course, there's actually a whole part of the game involving data, science, and research as well. So you're not just building things to ship out, you're also building labs and computational resources and bandwidth to the shared galaxy network... which you can then use to unlock or customize things, or hook it together with your friends for a more powerful (efficient) network, or hack the galactic web or whatever. You can alter the characteristics of the blueprints you have or retrofit existing constructions by using these resources...
Well, it's a nice theory, anyway.