If I vaguely remember, there's this sense of major plot points, and then everything else is connective tissue. You can imagine the same thing in a city design – what are the parts of a city that make it distinctive, or fun, or notable? You randomly choose some distinctive design elements, and then fill in.
This generator has a bit of that. I like how the rivers or an ocean can dramatically redefine a city in this generator. But there could certainly be more. Interesting topology. Highways, instead of a couple equally-sized roads. A scattering of distinctive building forms, like a military base, convent, marketplace.
Agree.
> which procedural generation has been, thus far, unable to provide
Disagree.
I think Dwarf Fortress (and to a lesser extent, EVE Online) has shown how to do procedural generation that does result in interesting interactions. There are four important factors in successful procedural games:
1. Granular modulation (you don't just punch a dwarf, you punch out a dwarf's left eye).
2. Diverse interactions (lava can be used to smith weapons. Or burn invaders. Or mixed with water.)
3. Persistent state as a result of the above. In a game like GTA, you can stage a one-man stand against overwhelming military force, but when you return to the site of your last stand, not a trace is left. In Dwarf Fortress, even if you lose you leave behind traces that can be revisited in future games. This is also the magic of EVE Online's alliance endgame (establishing player structures in contested space and seeing your alliance people flying there), without which I think the game loses its luster.
4. Environments small enough to build up a history. You can remember sites of memorable conflicts in Dwarf Fortress because your play space is constrained to a single tiny piece of the larger world. In Minecraft, Starbound, or No Man's Sky, your play area is so large that you never build up memories of any particular place (Starbound offsets this somewhat by having set pieces, but loses that sense of location outside of those pre-crafted zones). In EVE Online alliances settle for months or longer at a time in a particular region of space so that they develop an emotional attachment to it.
It's a cool as hell set of concepts which would never make it into a modern game imo, but that's the strength of it and what keeps it ticking
> Disagree.
We'll have to agree to disagree here. I've played hundreds of hours of Dwarf Fortress and it becomes very same-y and uninteresting.
The problem I think is that there's nothing to interact with. And there's no interesting ways to interact with the things you can interact with. Like star control 2 is technically a small game, and a small world; it just feels big, because it actually requires time to traverse, and there's actually things to find. Zelda games tend to be even smaller (primarily a sequence of small rooms), and yet feel much larger. Again, time, secrets and interest play a large role in why it feels that way.
people try to use procedural generation as a crutch, to avoid actual game design (see: no man's sky, minecraft; note the latter is popular by accident, adventure mode was intended as the primary, but loses its appeal quickly. Creative mode became the primary appeal by accident.) but its simply not sufficient without giving a compelling reason to actually explore and interact with the map.
I agree. I guess what I’m saying is that procedural generation need not be a crutch. It should allow the game designer to treat a procedural world like he were a location scout. They can then build sets within that world where story elements can take place, locations modified as required and the remaining 99% left as a living world that swallows up the player. Perhaps as games slowly evolve out of their ‘save the world’ routine this might make more sense. Instead we get endless collectathons that require scouting every polygon of the world.
I’ve also wondered about reduced scale in open world games. Instead of building a hollow and empty city like San Andreas, why not build a fully populated and detailed city block? Every building has detailed stores and accommodation where characters live and work. Your character is perhaps a shut-in (Rear Window) or a beat cop (back to L.A. Noir) where you somehow interact with that compact world.
Remember the often limited worlds of the N64? Racing games could only fit a handful tracks on a single cart. Every inch of virtual real estate was utilised. Such games have gone from a hundred MB to almost a hundred GB. It’s probably the only medium where you don’t need to sit through everything to have a full experience. The virtual worlds no longer need to feel like oversized jumgle gyms.
I’m probably way off the mark but I think we’re in for interesting times ahead.
This used to be a popular strategy, my favorite being lego island 2. You don’t need anything as grand an excuse as a neet pc or detective to allow for it: just take away the fucking car.
>It should allow the game designer to treat a procedural world like he were a location scout. They can then build sets within that world where story elements can take place, locations modified as required and the remaining 99% left as a living world that swallows up the player
This is how I believe open world games are usually done now, if not actually generated, then done mentally (imagine a town by a river, add a simple crisis and let things play themselves out. Not necessarily likely events, and certainly not realistic, but sufficient).
Most open world games do have a decent number of setpieces to find. The problem, I think, is thats all they are. Oblivion, skyrim, witcher, etc all make the same mistake: “you can climb any mountain” and thats where it stops. They are open only in the sense that you can travel in a non-linear path, in an non-linear order of the story. But they forget that the main interesting facet of games as a medium is interaction. That the player has autonomy.
They’re merely open in story, but mechanically they’re extremely linear. They don’t provide open interactions. It doesn’t matter what backdrop you give it, or how well designed and populated things are, if I can’t interact with them.
Roguelikes (note the k), simulation-games and dwarf fortress (and from tabletops: adnd and its kin) are my favorite in this regard: you can interact with everything, in a somewhat limited and controlled fashion, with the potential for interesting, unexpected results. They essentially forgo any real plot or story in favor of letting player autonomy act as its own source of interest. The games and its internal interactions are mechanically complex, while the world around it is significantly less so.
But thus, my opinion is the opposite of yours: the advancements in technology surrounding games, outside of perhaps AR/VR, currently and in the near-future, offers nothing new. The same style of games have already existed, and will continue on independent of these advancements. The only difference will be their graphical complexity, and more likely than not, each advance in graphics will come at the cost of the game’s design, in general. We’re in for nothing new.
What I’ve love to experience is getting lost in the sprawl and scale of a metropolis. Oddly enough L.A. Noir did a pretty decent job of this.
Anyone familiar with the history of Kerbal Space Program knows this issue. KSP's developers releases a "contracts" system that was essentially a random story teller. A few variables were randomly selected to generate a task. Of course many of the tasks ended up being ridiculous and they had to back-fill with a series of rules, but it is still rather silly. Then in a more recent update they allowed users to create "missions" and share them via steam (yawn). But this was layer atop the previous contracts scheme, resulting in a grinding mess that many/most players simply ignore. They instead play according to self-created story arcs and self-imposed rules.
SimCity did it best. Give the players the tools to play, but don't tell them how to play. Build a world full of towns and potential then let the players loose to explore and play how they want. Don't try to randomly build quests. Create the interface so that players can write and share their own storylines. I envision something more like OpenOblivion, aka Elder Scrolls with community-created quests/factions/stories.
It is still in very deep beta state, but this doesn't stop richness and complexity of worlds generated in it. It is seriously underrated project, and whole blog if worth a read
I’m sure I read something at the time that said what I’m describing though.
I've lately been playing Dead Cells a lot, er too much. It's a roguelike-metroidvania hybrid with a bit of Prince of Persia sprinkled atop.
All levels are procedural. Every time you die you start from level one again. But since it looks different every time, it doesn't bother one much or at all.
Curiously, there is no narrative or story line at all. The Indie studio behind it, Motion Twin, simply doesn't have the resources or budget for doing this 'properly'.
The typical approach is then to just use text -- just because you have to have a story. But they opted not to even do that.
For me Dead Cells is a great example of on-the-fly procedural content done exactly right -- w/o any narrative.
Dwarf Fortress puts in a lot of work to keep worlds and interactions rich. Even games like Football Manager are fundamentally role play, and the constantly evolving rivalries and personnel involved keep things fresh well into the game. Honestly think small but rich villages would be more interesting than massive but realistic looking cities, but this is still impressive stuff.