Hacker News new | past | comments | ask | show | jobs | submit login
The Quest to Make Massive Gaming Worlds Realistically Complex (vice.com)
38 points by DiabloD3 on Apr 17, 2015 | hide | past | web | favorite | 28 comments



The article is not good.

It is true that games tend to be overly simplistic in this way, but it is really a design problem, not a technical problem.

Technically, we can litter the ground with stuff and have plenty of permanent-world changes. Design-wise, it is usually unclear how to make that a playable game. Using someone's middleware is not going to help this.

The premise of this article is another example of what Frank Lantz calls the Immersive Fallacy: https://www.youtube.com/watch?v=6JzNt1bSk_U


Right. Such a world would need huge numbers of semi-intelligent NPCs cleaning up after the players.

In GTA, after you bashed a car or knocked over a sign, tow trucks, ambulances, fire trucks, and repair crews would show up. NPC firefighters and EMTs would extracted injured drivers from the wrecked cars and put them in ambulances, which then drive away to hospitals. Tow trucks tow away the damaged cars to impound lots or scrapyards. Repair crews show up and replace poles, restring wires, replace signs. Damage to buildings brings out owners, then insurance adjusters, then contractors, then construction crews with all their equipment. Like Sim City, but with more detail.

Running the world where nobody's watching uses a lot of compute power. Game design has traditionally avoided that. It might be affordable today.

Once the NPCs get good enough at running realistic simulated worlds, they can start taking jobs in the real world.


The amount of data per item can degrade over time.

During a wreck you have full details. Once the clean up crew leaves, you just need to record that 'NPC is hosptalized' and 'blue car (model 123) has 35% damage'. Later on, NPC goes by a mechanics, you can query the list of damaged vehicles, maybe load one being worked on in the garage, and if so remove it from the list. After a few hours items roll off the list (representing them being taken care of outside of the NPCs knowledge). The real trick is getting it to happen enough to be noticed but not enough to be seriously investigated so you can increase immersion without breaking it.

I've done a variant of this in a D&D campaign to simulate the world around the PCs, increasing detail only as it became more relevant to the PCs. Not something I recommend doing by hand w/ dice rolls.


"NPC firefighters and EMTs would extracted injured drivers from the wrecked cars and put them in ambulances, which then drive away to hospitals."

That happens to a small degree in Grand Theft Auto games. It tends to be more for comic effect than simulation, but it's a nice detail nonetheless. I hope one day games reach the fidelity you describe. It would be a fun world to wreak havoc in.


>Running the world where nobody's watching uses a lot of compute power.

That should not be true with proper design. The game would just store "car accident of type X and severity Y at time Z" until someone comes to watch. Then it would show the appropriate cleanup efforts that the watcher would witness.


That's what games do now. Switching between levels of detail like that tends to create artifacts, and requires lots of special casing.


Such a world would need huge numbers of semi-intelligent NPCs cleaning up after the players.

Not really. You can make many things evaporate magically after about three hours. The attention span of modern people gives you about an order of magnitude margin with that amount.


I felt this a while back doing CGI. My mind tries to recompute the world, so my models would aim toward simulation, endlessly precise, physics worth models. Whereas people actually working in the industry (Disney, Dreamworks, etc) followed the opposite philosophy, coming from traditional movies and animations where reality is so far from reach that you don't even think about trying. They just think about a minimal set of artefacts fitting a field of view that would tell what need to be told. Doesn't matter if you only create half a builing, the left side of a face, it's irrelevant as long as it helps incarnating the experience.

Now I've seen this in video games, since the PS3, games are more and more 'real' but the gameplay isn't better, only the magazine reviews are, with these 1080 8xAA Global Lit glamorous shots.

And as with movies, I now take great delight seeing how far people could make you feel and travel with limited means. You couldn't represent much, you had to pick and organize things in a way to tickle the mind of the viewer/player.


Crysis 2 takes place in downtown Manhattan? Can't enter a single building...

That's not what gets me excited about games. I want to go exploring. If I just wanted a fancy story and nice visuals, I'd watch a movie.


Tiny but nicely world crafted vs dull but gigantic place ?


>The premise of this article is another example of what Frank Lantz calls the Immersive Fallacy: https://www.youtube.com/watch?v=6JzNt1bSk_U

That's an interesting talk and is really asking: what is the existential endpoint of gaming?

I think the answer for me is the world as it is now, minus the things that are cumbersome or harmful and with unlimited options. So that means if you fall out of a plane and hit the ground you don't die, but it also means if you happen to like eating then you can sit and eat for days without getting obese. So yes, lets build in eating and shitting (If the internet is right then apparently some people like shitting on other people) as options for people.


It's also not QA testable. That's very important part of game development.

A linear game (or multi-linear) has a way of testing, anything else would require the entire team to dedicate one or two days from their week to play the game and report problems.

This also means your QA have to be all the time with you, rather than borrow small QA for the begining, then increase 10fold year or 6 months before shipping.


Ten years ago some friends and I did some preliminary design on an MMO game. Size and scope was basically the continental US, with thousands of players and hundreds of thousands of NPCs.

The data involved, we estimated, would fit comfortably into less than 10 TB.

It's not a technical problem, even at that time. It's strictly a design problem--once you've built such a thing, how do you populate it? What do you do to make it interesting?

And as I've gotten older, sadly, the appeal of creating virtual worlds has gotten less and less. I don't want people spending time frantically toggling state in a row in a database, even one I operate and get revenue from. :(


What this software is trying to propose is to get rid of the curated game experience and replace it with an algorithm. That works for some games, but I don't see how it could possibly be used to drive a consistent plot.


You don't necessarily need a consistent plot, or at least very much of one. There does have to be some careful and subtle design around an overarching goal, however.

In Eve Online several years ago, many of the more serious players didn't give a ___ about any of the curated stuff, other than as a source of munitions and resources for the alliance power struggle. Forces fight for dominance: that's not very much "plot" but many people definitely cared. (I would agree that you need something to structure the experience, however, otherwise you wind up with Second Life, then people's brains go right up the Maslow Need Hierarchy and wind up at sex.)

Also, in my work at the MADE museum in Oakland, I find that people might talk about notable stories and art direction in award winning past games, but do you know what they actually do? They play Smash, Mario Kart, or Super Mario. Judging by actions, it's the gameplay/interaction that most strongly draws people back to a re-experience.

In a way, assuming the need for plot in games would be like assuming the need for music/sound in cinema. It clearly can enhance the core experience, and you'd almost never have a modern commercial release without it. However, it's synergistic to and not absolutely essential to the core experience.


I think that's a good point, however some counterexamples I would give are Dragon Age, Deus Ex, and Mass Effect. They are massively curated and plot-driven games which many gamers will replay a great deal to re-experience the plot rather than the gameplay. In some cases they will replay them to re-experience the plot in spite of the gameplay.


> They are massively curated and plot-driven games which many gamers will replay a great deal to re-experience the plot rather than the gameplay. In some cases they will replay them to re-experience the plot in spite of the gameplay.

But in those cases, do we have a story supporting a game, or a game as the substrate/presentation medium for a story? I think we are actually talking about two vastly different things mashed under the same overall label of "game." These two things can even exist in the same piece of software we would call a "game." You can have a very plot/story/roleplaying based experience in Eve Online. You can also basically ignore all of the above and just go for a fairly sandboxy/emergent experience. (I will also note that the two player populations in Eve Online were somewhat distinct, and even when people were in both, they tended to do this by "switching modes.")

Clearly, there are many kinds of computer games where plot can be mostly, if not completely nonexistent. There is also clearly a subset of story-centric games that use the interactive medium to primarily tell a story. They are clearly not the same thing. (There's a good analogy with movies, in which some are very focused on the music/soundtrack and others are more focused on action and visual spectacle or the story/plot.)

(EDIT: To determine how "core" something is, ask yourself: Could the same "thing" exist as a movie, book, or in some other kind of medium without drastically changing? I would say that the overall experience of Mass Effect would drastically change were it to become a TV miniseries. However, you could conceivably create a TV miniseries with nearly the identical plot of Mass Effect.)


get rid of the curated game experience and replace it with an algorithm.

This is part of an overall trend in the industry that has been building for the past several years. It started with the game Rogue and for a long time it was relegated to the genre of Roguelikes. Now, procedurally generated content is showing up everywhere, just like RPG mechanics and microtransactions before it.


I'm not getting any useful info out of that press release. It sounds like a work distribution system to improve scaling, but what has this to do with fetch quests? The 'boring' quests in MMOs are not designed that way because of technical restrictions, but game design restrictions to keep the game world under control despite thousands of players exploiting the game mechanics (that's the actual 'game play' in MMOs, trying to 'gamble the game rules'.

Simple linear quest lines, unlimited resources, limiting interaction between players (trading, etc...) help to keep the game world balanced for all players, and most importantly, enable the game to provide an interesting 'hero-story' to each player instead of just being a simple 'pawn'.

Older MMOs like Ultima Online provided much more freedom (and power) to players then current MMOs which are much more 'dumbed down'. I think the current state of MMO monoculture can be traced back to WoW's success. WoW was simple, linear, accessible, and extremely successful. Obviously this is what players wanted instead of the complicated and unforgiving Ultima Online, and everyone wanted to create a WoW killer.

May be the wheel of reincarnation is now turning again, there seems to be a small revival of 'sandbox MMOs' which are less about individual story and more about player interactions.


Minecraft flips this on its head. It has permanence, massive world scaling, and endless potential for per-player experiences thanks to its purely open-ended nature.


Back in the 90s when I wanted to be a game developer I thought that the perfect game engine would be a scaled up particle emulator - making every "object" just a complex of particles. In effect emulating the way the world is built with stable elements.

Having finally come back to development 2 decades later and doing work with virtual worlds I definitely think this is the right approach. It's basically what is already being built with nuclear simulation models and the like so I think there is definitely room to work between game engines, doing it at a more abstract level and the supercomputers doing it at the particle level.


Why do you think emulating our real world so closely would make the perfect game engine? How would you approach (semi-)abstract games in such engine?


Kinda like a dynamic voxel engine with a really high voxel per meter cubed count?


Oh, so worlds will no longer cater to the magic teenage suburban fantasy that everything magically fixes itself in the background after they mess it up and leave the room?


Sounds like they're abstracting away the backed for MMOs with some interesting defaults and ideas.

I'm sure it might be huge for MMO devs, but as a person that occasionally plays games, unless someone comes up with a way to have thousands of players in close proximity, in a FPS-complex world all served under 100ms I'm not really excited. That's pushing the frontier gaming-wise IMO.

I don't see this tech producing fundamentally better gaming experiences. It just might level the playing field for MMO ppl.

EDIT: they actually don't seem gaming focused, so perhaps it's just the article


What this curated press release reminds me of, is when OpenCyc was released. Defining an ontology it contains detail and rules, which model things from materials and processes, to projectile weapons and of course swords and what not.

In theory, something like this should provide a technical resource of standard things, which designers could then employ and build on. And which programmers, could generate a more immersive world to build gameplay on. In practice OpenCyc is a resource hog, and the details and rules it contains are when examined found to be very limited.


I'm surprised that neither the article, nor the people here have mentioned http://www.wurmonline.com as the probably most successful implementation of the sandbox concept, where one player's actions affect the virtual world for everyone else.


Sounds to me like they're using actors to build MMOs.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: