SimCity 5

[cybrxkhan]

I have to generally agree with warpus here. While I have decided to show no love about the servers and stuff, I'll hold out my final judgment about gameplay in and of itself until I actually get my hands on the game - that said, from what my friend has been telling me about the game (he has it), it doesn't sound too good for me. Even though he said he enjoyed the game (once the servers got working... sort of), he blatantly said that modded SC4 is a much better single-player experience, and I'll trust him on that one - judging from the reviews and whatever information I can get, the new SimCity really is more appropriately SimCity Online rather than SimCity, unfortunately.

 

[sophie]

Absolutely agree. See for instance old RPGs. You were placed into a large world without any indication at all. It was supposed you would be able to find your way asking the characters, reading books you find, etc. Now even the RPGs which give you more "freedom" have this tiny marker showing you the way at any moment. It looks like today gamers cant think or read and must be constantly directed so they dont get lost. (For more information on the issue see my thread at the tavern: "Are we getting dumber?" )

Judging by how poor gamers like Arin and JonTron are at following directions or reading information put right in front of them, is it really that surprising?

 

[cybrxkhan]

And well, apparently this server hooplah wasn't needed in first place, according to a Maxis insider: http://www.rockpapershotgun.com/2013/03/12/simcity-server-not-necessary/

Normally obviously the source would be suspect, but given how apparently some people have been able to play for some time even without an online connection, I suppose it holds some merit.

 

[Thorgalaeg]

It was expected. Think on 100,000 or 1 million people playing the game simultaneously all around the world. How much calculation power would they need? Obviously even EA would not shoot themselves in the foot that way.

 

[CivCube]

And well, apparently this server hooplah wasn't needed in first place, according to a Maxis insider: http://www.rockpapershotgun.com/2013/03/12/simcity-server-not-necessary/

Normally obviously the source would be suspect, but given how apparently some people have been able to play for some time even without an online connection, I suppose it holds some merit.

Someone added in the comments that the Glassbox AI apparently only covers a small segment of the city, with the rest being a "shadow" population. After 100k citizens, the simulation stops working.

 

[SouthernKing]

Someone's going to have some explaining to do.

 

[cybrxkhan]

Someone added in the comments that the Glassbox AI apparently only covers a small segment of the city, with the rest being a "shadow" population. After 100k citizens, the simulation stops working.

Yeah, if that turns out to be true (which seems quite possible, and would make sense in my opinion), that would be very, very problematic - not only would the servers be an issue, but the very gameplay itself? Oh my.

 

[NBAfan]

Someone added in the comments that the Glassbox AI apparently only covers a small segment of the city, with the rest being a "shadow" population. After 100k citizens, the simulation stops working.

That makes the tiny cities decision even stupider

 

[SouthernKing]

Also, apparently, the glassbox traffic "AI" isn't AI at all; it uses the same system for traffic/services as it does for utilities, with the agents and whatnot.
More here, I found it in the comments of the RPS article.

 

[PieceOfMind]

That's the excuse, but today's computers are so much more powerful than the ones we had when SC4 came out, it's really not a very good one.

My computer is much more powerful than the typical one from 2003 too, yet SC4 still brings it to its knees. Mind you that's also partly because it can't use more than one core.

Gaming worlds should be getting larger, not smaller!

If you design your game well enough, a large map is always possible these days, unless you want to simulate every single blade of grass individually or some other such silly thing.

But that's kinda the point. Maybe not every blade of grass, but every silly sim. Imagine doing pathing for 200,000 little agents who want to go to work, and then doing the same for 200,000 power (agents) to provide homes with power, then water, then garbage, and so on. Computationally, assuming all agents are modelled as they were described to us in all the previews, this should be much more demanding than SC4. So much so that even with more powerful hardware than 2003 we should still expect smaller cities.

I guess we can't know for sure without access to some sort of profiler, but we might have to agree to disagree there. If we supposed the alternative - that Maxis/EA forced small cities on players for a reason other than computers being able to handle them - what would that be?

Calculations on server? I think that one has been proved wrong.
Later selling larger cities as DLC? Plausible, considering it's EA, but I think even for them it's too damaging (PR-wise) a route.
What else? Originally you said to combat piracy - how would that work?

 

[Maniacal]

Wow, even I didn't expect that many exaggerations and lies from EA about their simulation, and it seems increasingly pointless, unnecessary and not even living up to their clams.

Also something I have always liked about the Tropico games:

 

[PieceOfMind]

Also, apparently, the glassbox traffic "AI" isn't AI at all; it uses the same system for traffic/services as it does for utilities, with the agents and whatnot.
More here, I found it in the comments of the RPS article.

And now the evidence is building that going for a very fine-grained simulation may have been a mistake even for increased realism. The one that takes the cake for me is a sim can be high wealth one day, and then medium wealth the next simply because he found a medium wealth home with 2 kids on the way home from work.

The funny thing is, I'm pretty sure that's not a bug but working as designed. This is something that worries me about Simcity5, that severe 'shortcuts' were needed to keep the gameplay going. What's the point of going with the higher fidelity model if it's going to produce just as many absurdities as the statistical model while being more computationally expensive to boot!?

 

[Maniacal]

I think this .gif sums up the AI:

 

[warpus]

But that's kinda the point. Maybe not every blade of grass, but every silly sim. Imagine doing pathing for 200,000 little agents who want to go to work, and then doing the same for 200,000 power (agents) to provide homes with power, then water, then garbage, and so on. Computationally, assuming all agents are modelled as they were described to us in all the previews, this should be much more demanding than SC4. So much so that even with more powerful hardware than 2003 we should still expect smaller cities.

That's incredibly bad design, then. You can achieve virtually the same thing in terms of gameplay via other means.. Other means that don't force you to limit the size of the map so much.

I mean.. imagine if the next Elder Scrolls game was being discussed by devs in very early stages of the game and one guy said: "Okay.. We're going to add an exciting new feature to the game.. but it's going to shrink the playable area down from what we saw in Skyrim down to a roughly 2km by 2km piece of in-game land"

You just don't.. do that. You think of other ways of making it work that would allow you to keep the large maps.

I don't know why they went with that design choice, but it doesn't add anything to the game for me, and takes a LOT of it away.

 

[Polycrates]

Yeah the more I read about this, the more impressed I am with the Tropico games; the numbers are smaller and Tropicans may not always make the most sensible decisions, but the simulation of individual citizens in a pretty complex system works remarkably well (well, for the most part) into building a functional macro-level whole, without any real behind-the-scenes cheating to keep the edifice from collapsing.

 

[PieceOfMind]

That's incredibly bad design, then. You can achieve virtually the same thing in terms of gameplay via other means.. Other means that don't force you to limit the size of the map so much.

I mean.. imagine if the next Elder Scrolls game was being discussed by devs in very early stages of the game and one guy said: "Okay.. We're going to add an exciting new feature to the game.. but it's going to shrink the playable area down from what we saw in Skyrim down to a roughly 2km by 2km piece of in-game land"

You just don't.. do that. You think of other ways of making it work that would allow you to keep the large maps.

I don't know why they went with that design choice, but it doesn't add anything to the game for me, and takes a LOT of it away.

Personally I think there is great appeal in having a very high fidelity simulation. The problem is the side-effects or consequences. Like your Skyrim example, you may end up sacrificing something that was more important all along.

Reminds me of the sort of reasons that water isn't modelled very well in most games, except the ones where basically the whole game is based on water (e.g. Wave Race 64, Hydrophobia on PC). In your typical FPS where maybe 5% of time playing the game is in a locale near water, and then probably 5% of players ever notice the detail of the water, what's the point in modelling water to such a nice degree (extremely computationally expensive if it's to look at least half-decent, by the way) that it causes a 10fps hit for average computers?

Also similar is the cost of going to 1upt in Civ5. I am fairly sure the designer/s didn't fully anticipate the sorts of path-finding issues and unit shuffling that would result. Basically at the design stage of each of these projects they have been too ambitious for certain features and could have benefited from having a computer scientist on the team.

 

[Marla_Singer]

Computationally, assuming all agents are modelled as they were described to us in all the previews, this should be much more demanding than SC4. So much so that even with more powerful hardware than 2003 we should still expect smaller cities.

CitiesXL or Cities in Motion are both able to simulate correctly pathfinding on much larger city maps than SimCity. And they've been produced several years ago by smaller companies than Maxis/EA.

 

[PieceOfMind]

Well I have not played much each of those games, but I'm fairly sure that each would have its own "shortcuts" that SimCity doesn't get to enjoy.
CIM for example - I don't think you can change any of the buildings or roads? So you have a static network. That makes it orders of magnitude easier to do path-finding tasks as you don't have to constantly update what the network is and therefore where the best paths are.

Cities XL sounds like it still has the benefit of not having to path-find anything except its traffic. What I was saying about Simcity5 is that it's a bit over the top to be path-finding for things like power supply, water supply, garbage collection, sewage, and the list probably goes on. If those tasks take a similar amount of effort as path-finding for traffic, then you've already got 5 times the resource requirement.

It sounds, too, like Simcity5 is relying too heavily on a "greedy" path-finding heuristic. Sims all moving to the closest applicable destination, in a horde, and then when it's filled moving onto the next available. "Greedy" because each agent only thinks of themselves and are very short-sighted. Either this was lazy programming (possibly also a type of skeleton code in the hopes of improving later via patches) or necessary because of performance issues - I suspect the former.

Anyway, I will have to try out Cities XL and see if it does indeed do these tasks better.

 

[warpus]

Personally I think there is great appeal in having a very high fidelity simulation. The problem is the side-effects or consequences. Like your Skyrim example, you may end up sacrificing something that was more important all along.

Yeah, exactly. Makes me wonder what other reasons they might have had to make the maps that small. "Hey, if we go down the extreme agent route, we'll need small maps. Then it'll be easier to __________" ?

 

[cybrxkhan]

I think SimCity 4's "pathfinding" system, particularly the one with the Rush Hour expansion, while problematic in its own ways, worked out fine overall... at least compared to what the new SimCity appears to be. They could have just improved a bit on that instead of trying to overdo themselves with something that wouldn't have been worth it anyways.