I've played SimCity 2000 and SimCity 4, but have only played 4 for a few hours. I concede that maybe I didn't play 4 enough to really get it and appreciate the underlying simulation but I still prefer the new game's presentation to 4's. Despite the abstractions the new game uses it still feels more real than 4 did to me.
I've spent a fairly large amount of time on Simcity 4 and while the simulation is pretty robust, it was the tweaking and mods that really made it something spectacular.
I wish I could post some screenshots but i'm at work, if you're willing to spend the time downloading community addons it really adds tremendously to the realism. 4 also benefits from 10 years of more powerful computers which means that the huge regions that quickly became unplayable are kind of just steamrolled by the amount of RAM and processing power in home pc's, even with a huge lack of optimization.
It will all come down to the addon and modding that Maxis introduces into 5 which hopefully will increase the depth of the game as much as they did for 4.
This manufacturing city seems to have the problem of the trains not coming fast enough to collect all the goddamn alloy.
Ah in my case I am turning all my alloy into processors, and then into TV's and Computers so the train delivery has been pretty solid for me. Before that all my alloy deliveries were via truck until my processing plants were set up.
NNID: delphinidaes Official PA Forums FFXIV:ARR Free Company <GHOST> gitl.enjin.com Join us on Sargatanas!
Factories still complaining about unskilled workers.
What is your tech level? Should be 3 yellow gears below your education level. That is what determines your standards for industry.
Also be sure you are educating them in a university or community college, I don't believe grade school, library, or highschool will count for the college educated workers thing. Also be sure that your buildings are green colored when you check your education data map and that your industry is located in a position to readily receive those residents.
It coul dbe you have outside workers coming in who are not skilled.
NNID: delphinidaes Official PA Forums FFXIV:ARR Free Company <GHOST> gitl.enjin.com Join us on Sargatanas!
0
ShogunHair long; money long; me and broke wizards we don't get alongRegistered Userregular
Factories still complaining about unskilled workers.
What is your tech level? Should be 3 yellow gears below your education level. That is what determines your standards for industry.
2 1/2 gears. Though, my education level can't be any higher. I have 3500/3500 students enrolled in my universities.
they will complain and move out until it fills to full and then they will come back. There are several stages of high tech industry. I've also noticed odd things with education and tech level where my education would get stuck at 1.5 hats but my tech level would quickly climb to full. Then it would bottom out completely and my education would start to rise.
From what I can tell, it shouldn't matter where the students get educated. The students go to a school building, get education, and then return and make the building educated. Any sims spewed out by that building will then have the educated tag. The exception is the library, which acts as a shop.
From what I can tell, it shouldn't matter where the students get educated. The students go to a school building, get education, and then return and make the building educated. Any sims spewed out by that building will then have the educated tag. The exception is the library, which acts as a shop.
Possibly, but there is a request you get from industry saying "Not enough college educated workers". It could just be flavor text, but all those messages went away when I bulldozed my grade school and highschool and just had a community college and university.
*Edit* Actually, isn't there a big image of all the messages that pop up and what they mean? I think it's from the strategy guide but I can't find it off hand
Delphinidaes on
NNID: delphinidaes Official PA Forums FFXIV:ARR Free Company <GHOST> gitl.enjin.com Join us on Sargatanas!
In fact my theory is that the only reason this is being reported as a launch worse than D3 is because in general people like Blizzard and hate EA, which compounds any actual issues that arise for this game vs D3.
Uh, I don't recall D3 leaving me with any "wtfbbq" bugs, like buildings just stop working, or infinite people flooding into a building, or invisible people who won't ever leave, etc,
There's also the part where it'd be fucking easy to do a lot better than they're currently doing. I mean, the traffic AI is literally just as dumb as rocks. It's the sort of childish model I'd expect from a first-year CS student. Actually, last term, we made our first-year CS students do an assignment with pathfinding and trucks that "claimed" locations to visit and still used shortest-path pathfinding. So to be clear, if you were to ask my first-year CS students on how to implement SimCity, they'd come up with a superior model.
Seriously. I could not in good conscience give a CS degree to someone who thought that was an appropriate traffic simulation model.
The population counter being fudged while the number of sims using public transportation doesn't appear to be fudged means you should go into the menus to see if your public transportation is actually effective, right?
Couscous on
0
SarksusATTACK AND DETHRONE GODRegistered Userregular
In fact my theory is that the only reason this is being reported as a launch worse than D3 is because in general people like Blizzard and hate EA, which compounds any actual issues that arise for this game vs D3.
Uh, I don't recall D3 leaving me with any "wtfbbq" bugs, like buildings just stop working, or infinite people flooding into a building, or invisible people who won't ever leave, etc,
There's also the part where it'd be fucking easy to do a lot better than they're currently doing. I mean, the traffic AI is literally just as dumb as rocks. It's the sort of childish model I'd expect from a first-year CS student. Actually, last term, we made our first-year CS students do an assignment with pathfinding and trucks that "claimed" locations to visit, so really, you could expect a superior model from those first-year CS students....
I think it's easy to think it's that cut and dry, that Maxis is just filled with idiots, but I think that's a little presumptuous. There is probably a reason why they went in this direction. There is obviously a lot of information we are not privy to.
And I don't play D3, and I'm aware that there are definitely folks who didn't have a lot of problems with the game, but from the media coverage of the game's launch and thereafter it was pretty clear that a lot of people were having issues even if it wasn't universal.
In fact my theory is that the only reason this is being reported as a launch worse than D3 is because in general people like Blizzard and hate EA, which compounds any actual issues that arise for this game vs D3.
Uh, I don't recall D3 leaving me with any "wtfbbq" bugs, like buildings just stop working, or infinite people flooding into a building, or invisible people who won't ever leave, etc,
There's also the part where it'd be fucking easy to do a lot better than they're currently doing. I mean, the traffic AI is literally just as dumb as rocks. It's the sort of childish model I'd expect from a first-year CS student. Actually, last term, we made our first-year CS students do an assignment with pathfinding and trucks that "claimed" locations to visit, so really, you could expect a superior model from those first-year CS students....
I think it's easy to think it's that cut and dry, that Maxis is just filled with idiots, but I think that's a little presumptuous. There is probably a reason why they went in this direction. There is obviously a lot of information we are not privy to.
And I don't play D3, and I'm aware that there are definitely folks who didn't have a lot of problems with the game, but from the media coverage of the game's launch and thereafter it was pretty clear that a lot of people were having issues even if it wasn't universal.
Server issues are different from in-game bugs. Both launches have had server issues, but D3 had very few "wtfbbq?!" bugs. You can sort of get a sense of them here:
I just don't understand this hurgleburgle "SimCity's not that bad" defense. I can literally describe for you in a short paragraph a superior Sim/traffic model that would be well within your computer's ability to process. I was discussing this with a friend of mine who's an AI post-doc, and we both estimated that implementing a Tropico 4-esque model would require just less than a millisecond in computing time for thousands of agents. (Hence, see how people have modded Tropico 4 to work with thousands of citizens, each of which are substantially more detailed than the Sims in SimCity.) With pathfinding occurring at specific intervals, it wouldn't be too hard to implement something for hundreds of thousands of agents, imo, if we fixed paths after a pathfinding cycle and only recomputed if the any paths were invalidated. (Managing hundreds of thousands of agents, of course, gets a little bit heavier in terms of memory consumption, but that's another story.) Mass transit's slightly harder, but we, in minutes, came up with several heuristics that were superior to "randomly turn at intersections".
It's a sad day when I point at Maxis' SimCity and say, "Boy, Kalypso Media, the German publishers of Airline Tycoon 2, did this a whole lot better."
This manufacturing city seems to have the problem of the trains not coming fast enough to collect all the goddamn alloy.
I regularly see sims asking for me to add more trains but this doesn't seem an option?
I assume this is done by building a rail network into the city using the heavy rails and plopping more stations.
Yeah, adding more stations adds more trains, but take a guess as to what happens with those.
Is the answer pathfinding related by any chance.
I think so. At least, when I tried with three stations there were more trains but one station (the first one the trains arrived at) would have about 6000 riders per day whilst the other two had around 400-500.
Also, thinking about the sims and their homes and jobs, it made me think this guy was in charge of the simulation;
In fact my theory is that the only reason this is being reported as a launch worse than D3 is because in general people like Blizzard and hate EA, which compounds any actual issues that arise for this game vs D3.
Uh, I don't recall D3 leaving me with any "wtfbbq" bugs, like buildings just stop working, or infinite people flooding into a building, or invisible people who won't ever leave, etc,
There's also the part where it'd be fucking easy to do a lot better than they're currently doing. I mean, the traffic AI is literally just as dumb as rocks. It's the sort of childish model I'd expect from a first-year CS student. Actually, last term, we made our first-year CS students do an assignment with pathfinding and trucks that "claimed" locations to visit, so really, you could expect a superior model from those first-year CS students....
I think it's easy to think it's that cut and dry, that Maxis is just filled with idiots, but I think that's a little presumptuous. There is probably a reason why they went in this direction. There is obviously a lot of information we are not privy to.
And I don't play D3, and I'm aware that there are definitely folks who didn't have a lot of problems with the game, but from the media coverage of the game's launch and thereafter it was pretty clear that a lot of people were having issues even if it wasn't universal.
Server issues are different from in-game bugs. Both launches have had server issues, but D3 had very few "wtfbbq?!" bugs. You can sort of get a sense of them here:
I just don't understand this hurgleburgle "SimCity's not that bad" defense. I can literally describe for you in a short paragraph a superior Sim/traffic model that would be well within your computer's ability to process. I was discussing this with a friend of mine who's an AI post-doc, and we both estimated that implementing a Tropico 4-esque model would require just a millisecond in computing time for thousands of agents. (Hence, see how people have modded Tropico 4 to work with thousands of citizens, each of which are substantially more detailed than the Sims in SimCity.) Mass transit's slightly harder, but we, in minutes, came up with several heuristics that were superior to "randomly turn at intersections".
It's a sad day when I point at Maxis' SimCity and say, "Boy, Kalypso Media, the German publishers of Airline Tycoon 2, did this a whole lot better."
It's just easier for me to give them the benefit of the doubt. In my experience there are almost always variables and circumstances that are only apparent when you are very close to the problem, and I don't just mean computing problems but life in general.
I mean I just can't see them failing to implement something you say is really easy without a good reason.
In fact my theory is that the only reason this is being reported as a launch worse than D3 is because in general people like Blizzard and hate EA, which compounds any actual issues that arise for this game vs D3.
Uh, I don't recall D3 leaving me with any "wtfbbq" bugs, like buildings just stop working, or infinite people flooding into a building, or invisible people who won't ever leave, etc,
There's also the part where it'd be fucking easy to do a lot better than they're currently doing. I mean, the traffic AI is literally just as dumb as rocks. It's the sort of childish model I'd expect from a first-year CS student. Actually, last term, we made our first-year CS students do an assignment with pathfinding and trucks that "claimed" locations to visit, so really, you could expect a superior model from those first-year CS students....
I think it's easy to think it's that cut and dry, that Maxis is just filled with idiots, but I think that's a little presumptuous. There is probably a reason why they went in this direction. There is obviously a lot of information we are not privy to.
And I don't play D3, and I'm aware that there are definitely folks who didn't have a lot of problems with the game, but from the media coverage of the game's launch and thereafter it was pretty clear that a lot of people were having issues even if it wasn't universal.
Server issues are different from in-game bugs. Both launches have had server issues, but D3 had very few "wtfbbq?!" bugs. You can sort of get a sense of them here:
I just don't understand this hurgleburgle "SimCity's not that bad" defense. I can literally describe for you in a short paragraph a superior Sim/traffic model that would be well within your computer's ability to process. I was discussing this with a friend of mine who's an AI post-doc, and we both estimated that implementing a Tropico 4-esque model would require just a millisecond in computing time for thousands of agents. (Hence, see how people have modded Tropico 4 to work with thousands of citizens, each of which are substantially more detailed than the Sims in SimCity.) Mass transit's slightly harder, but we, in minutes, came up with several heuristics that were superior to "randomly turn at intersections".
It's a sad day when I point at Maxis' SimCity and say, "Boy, Kalypso Media, the German publishers of Airline Tycoon 2, did this a whole lot better."
It's just easier for me to give them the benefit of the doubt. In my experience there are almost always variables and circumstances that are only apparent when you are very close to the problem, and I don't just mean computing problems but life in general.
I mean I just can't see them failing to implement something you say is really easy without a good reason.
Model transport system as a graph, with intersections and transport hubs as vertices, roads and transit paths as edges. Edges are costed according to road / path length. (Optional: weight edges according to type, so that, for example, high-density roads of a fixed length are preferred over low-density roads of the same length.) Insert every job in the city into the network as sink vertices. This will drastically increase the number of vertices in the graph, but won't increase complexity, as those vertices are essentially leafs, and no decisions really need to be made, since they're all degree 2. For each worker agent, use weighted-BFS to find the closest suitable job. Note that we can agglomerate multiple agents together, rather than performing a unique search for each agent, and iterative searches are likely to be related but whatever. The agent now has the closest suitable available job, and immediately knows the shortest / lowest cost path there, which it can immediately follow. Filled sink vertices can be removed. Commute homes are automatic reversals of the to-work path.
Now, notably, we can also change the model so that once agents have found a job, they retain that job. If a city is massively rebuilt, that would be problematic, but the nature of the current game model is that this process needs to happen only every 1440 seconds. So we actually have something like 2 minutes to perform all these calculations for the majority of the agents, and only actually need real-time calculations for the select group of citizens / jobs that get fucked with in the last few seconds before rush hour, so practically speaking, we could use a much more complex algorithm even, and this would have the advantage that Sims retain certain patterns. Use some sort of marker to indicate when recomputation needs to happen (or make it timed, so Sims follow the same pattern for a month or two, then refind jobs, etc,)
Hell, the emergency vehicles all rushing to the same emergency? Solution: each emergency has a single slot. Gets claimed by the nearest vehicle, found using weighted-BFS again, running in max O(n) time, average O(lg n) time. If you want, insert a little fudging to assign agents to their nearest intersection rather than along roads. Non-ideal, but still easily understood and manipulable, and would allow for a lot more agents, since the number of BFSes done would correlate to the number of intersections rather than agents. (Optional: insert "fake" intersection vertices along long, unintersected roads to alleviate adversarial road design.)
None of the content here is beyond a 2nd-year CS student, btw, maybe a 3rd-year. Coming up with it might test them, but I'd certainly expect, in a class of 2nd years, that a number of them would come up with this or something similar.
How hard would it have been to have the agents path independently of each other? The way I'm thinking of it is in layers; layer 1 would be workers, layer 2 shoppers, layer 3 water, layer 4 sewage, etc. They all follow the same rules, they all follow the same path (the roads) but they act independently of each other so one doesn't affect the other. But what we apparently have is all of those 'layers' mashed together and interfering with each other. There's no logic to sewage backing up or power or water not getting into a city because the roads are congested.
Also, do the agents all clock on and off at the same time? Or are they staggered, e.g. industry starts at 8am, commercial and school 9am, school finishes at 3pm, industry at 5pm and commercial 6pm. Because that seems a logical thing to do.
0
SarksusATTACK AND DETHRONE GODRegistered Userregular
edited March 2013
I don't disbelieve that you know what you're talking about, I'm just saying, is Maxis being idiots really the best reason for why they didn't implement this? I really don't like shit-talking game devs so I give them the benefit of the doubt. This works in my favor too because if I think there might be mitigating circumstances it makes me less annoyed, and I like not being annoyed.
Edit: It's like when somebody cuts you off in traffic, what if they really had to poop?
Wait, wait, wait. Just reading over some posts I missed... Are these 130,000 mysterious visitors in my city actually the ghosts of the city that I levelled? God I hope that isn't the case.
Until they patch it im afraid your city will permanately count those visitors. The same thing is occuring in my city, completely bulldozed yet every month it thinks i have 600k visitors and 60k commuting shoppers.
I also noticed an invisible object telling me it needs a road connection when I have the road menu up and I could not figure out what it was until I randomly saw some bush thing in the middle of the sky
I don't disbelieve that you know what you're talking about, I'm just saying, is Maxis being idiots really the best reason for why they didn't implement this? I really don't like shit-talking game devs so I give them the benefit of the doubt. This works in my favor too because if I think there might be mitigating circumstances it makes me less annoyed, and I like not being annoyed.
Edit: It's like when somebody cuts you off in traffic, what if they really had to poop?
Hell, I dunno why they didn't do it. I'm just saying, this is an unnecessarily inferior product. Tropico 4 can officially handle 2000 agents with homes, jobs, and dynamic real-time shortest-path pathfinding. OpenTTD can handle ... at least 15000 vehicles with user-specified destinations using dynamic, real-time shortest-path pathfinding using, I think, a modified version of A* search, and also has to deal with train signals. Note that OpenTTD caps at 15000 because they're of the belief that nobody would ever use more than 15000 vehicles, not because that's a system-imposed maximum, and this is an open-source project (i.e. FREE).
It's just mind-boggling as a computer scientist. I mean, this is the sort of thing we daydream about as students, how would we build SimCity if we grew up to be game developers... and what was delivered, at least in terms of traffic design, is way way dumber than what we conceived of even as undergrads.
How hard would it have been to have the agents path independently of each other? The way I'm thinking of it is in layers; layer 1 would be workers, layer 2 shoppers, layer 3 water, layer 4 sewage, etc. They all follow the same rules, they all follow the same path (the roads) but they act independently of each other so one doesn't affect the other. But what we apparently have is all of those 'layers' mashed together and interfering with each other. There's no logic to sewage backing up or power or water not getting into a city because the roads are congested.
Also, do the agents all clock on and off at the same time? Or are they staggered, e.g. industry starts at 8am, commercial and school 9am, school finishes at 3pm, industry at 5pm and commercial 6pm. Because that seems a logical thing to do.
I'm sorry, but what?
I have never seen my water, power, or sewage slow down because of traffic congestion. Are you talking about something that happens on the region?
NNID: delphinidaes Official PA Forums FFXIV:ARR Free Company <GHOST> gitl.enjin.com Join us on Sargatanas!
How hard would it have been to have the agents path independently of each other? The way I'm thinking of it is in layers; layer 1 would be workers, layer 2 shoppers, layer 3 water, layer 4 sewage, etc. They all follow the same rules, they all follow the same path (the roads) but they act independently of each other so one doesn't affect the other. But what we apparently have is all of those 'layers' mashed together and interfering with each other. There's no logic to sewage backing up or power or water not getting into a city because the roads are congested.
Also, do the agents all clock on and off at the same time? Or are they staggered, e.g. industry starts at 8am, commercial and school 9am, school finishes at 3pm, industry at 5pm and commercial 6pm. Because that seems a logical thing to do.
Actually, the problem of pathing them all together is exponentially harder than dealing with the layers individually. Just from a computational complexity standpoint.
We've actually been having quite a good laugh in the planning group as to how horrible the pathfinding is. There is basically an entire academic conference (or half of two, the Symposium on Combinatorial Search and Artificial Intelligence in Interactive Digital Entertainment) discussing "How the hell do I path thousands of things without collisions in a way that's not super expensive".
I mean, I don't expect to see bleeding edge pathfinding research in all things, but I've definitely failed students that put forth better efforts. What's hilarious is that EA has had some *major* pathfinding brilliance in house. The pathfinding algorithms in the newest C&C game and Dragon Age are actually still pretty relevant to the research community despite being a few years old at this point. Just, you know, big big breakthroughs. I guess the left hand doesn't know what the right is doing over there.
In fact my theory is that the only reason this is being reported as a launch worse than D3 is because in general people like Blizzard and hate EA, which compounds any actual issues that arise for this game vs D3.
Uh, I don't recall D3 leaving me with any "wtfbbq" bugs, like buildings just stop working, or infinite people flooding into a building, or invisible people who won't ever leave, etc,
There's also the part where it'd be fucking easy to do a lot better than they're currently doing. I mean, the traffic AI is literally just as dumb as rocks. It's the sort of childish model I'd expect from a first-year CS student. Actually, last term, we made our first-year CS students do an assignment with pathfinding and trucks that "claimed" locations to visit, so really, you could expect a superior model from those first-year CS students....
I think it's easy to think it's that cut and dry, that Maxis is just filled with idiots, but I think that's a little presumptuous. There is probably a reason why they went in this direction. There is obviously a lot of information we are not privy to.
And I don't play D3, and I'm aware that there are definitely folks who didn't have a lot of problems with the game, but from the media coverage of the game's launch and thereafter it was pretty clear that a lot of people were having issues even if it wasn't universal.
Server issues are different from in-game bugs. Both launches have had server issues, but D3 had very few "wtfbbq?!" bugs. You can sort of get a sense of them here:
I just don't understand this hurgleburgle "SimCity's not that bad" defense. I can literally describe for you in a short paragraph a superior Sim/traffic model that would be well within your computer's ability to process. I was discussing this with a friend of mine who's an AI post-doc, and we both estimated that implementing a Tropico 4-esque model would require just a millisecond in computing time for thousands of agents. (Hence, see how people have modded Tropico 4 to work with thousands of citizens, each of which are substantially more detailed than the Sims in SimCity.) Mass transit's slightly harder, but we, in minutes, came up with several heuristics that were superior to "randomly turn at intersections".
It's a sad day when I point at Maxis' SimCity and say, "Boy, Kalypso Media, the German publishers of Airline Tycoon 2, did this a whole lot better."
It's just easier for me to give them the benefit of the doubt. In my experience there are almost always variables and circumstances that are only apparent when you are very close to the problem, and I don't just mean computing problems but life in general.
I mean I just can't see them failing to implement something you say is really easy without a good reason.
Model transport system as a graph, with intersections and transport hubs as vertices, roads and transit paths as edges. Edges are costed according to road / path length. (Optional: weight edges according to type, so that, for example, high-density roads of a fixed length are preferred over low-density roads of the same length.) Insert every job in the city into the network as sink vertices. This will drastically increase the number of vertices in the graph, but won't increase complexity, as those vertices are essentially leafs, and no decisions really need to be made, since they're all degree 2. For each worker agent, use weighted-BFS to find the closest suitable job. Note that we can agglomerate multiple agents together, rather than performing a unique search for each agent, and iterative searches are likely to be related but whatever. The agent now has the closest suitable available job, and immediately knows the shortest / lowest cost path there, which it can immediately follow. Filled sink vertices can be removed. Commute homes are automatic reversals of the to-work path.
Now, notably, we can also change the model so that once agents have found a job, they retain that job. If a city is massively rebuilt, that would be problematic, but the nature of the current game model is that this process needs to happen only every 1440 seconds. So we actually have something like 2 minutes to perform all these calculations for the majority of the agents, and only actually need real-time calculations for the select group of citizens / jobs that get fucked with in the last few seconds before rush hour, so practically speaking, we could use a much more complex algorithm even, and this would have the advantage that Sims retain certain patterns. Use some sort of marker to indicate when recomputation needs to happen (or make it timed, so Sims follow the same pattern for a month or two, then refind jobs, etc,)
Hell, the emergency vehicles all rushing to the same emergency? Solution: each emergency has a single slot. Gets claimed by the nearest vehicle, found using weighted-BFS again, running in max O(n) time, average O(lg n) time. If you want, insert a little fudging to assign agents to their nearest intersection rather than along roads. Non-ideal, but still easily understood and manipulable, and would allow for a lot more agents, since the number of BFSes done would correlate to the number of intersections rather than agents. (Optional: insert "fake" intersection vertices along long, unintersected roads to alleviate adversarial road design.)
None of the content here is beyond a 2nd-year CS student, btw, maybe a 3rd-year. Coming up with it might test them, but I'd certainly expect, in a class of 2nd years, that a number of them would come up with this or something similar.
A job is taken out of the graph for the first worker that reaches it, but what if the first worker in your list is on the other side of the city from where the remaining jobs are, and there are actually workers down further in the list who are closer to those jobs. The workers on the other side of the city would get the jobs and travel all the way across town for them, while the workers down the list who are closer may not get any job.
A better way to do it is first divide the city graph using a 2-d spacial sub division structure (quadtree for example, though others will be more optimal since cities will not be uniform), and perform the job search at the lowest level of the tree, then recurse to the next level and perform the search for the remaining workers, etc until finally you perform the search at the level of the entire city. This way, you will assign workers closer to the jobs first.
Jephery on
}
"Orkses never lose a battle. If we win we win, if we die we die fightin so it don't count. If we runs for it we don't die neither, cos we can come back for annuver go, see!".
Given that they have said they are already going about improving the traffic system, it is kind of hard not to see it as either incompetence or just being really goddamn rushed. Their GDC presentation even stated that the D* could weigh traffic and whatnot.
0
CarbonFireSee youin the countryRegistered Userregular
I don't disbelieve that you know what you're talking about, I'm just saying, is Maxis being idiots really the best reason for why they didn't implement this? I really don't like shit-talking game devs so I give them the benefit of the doubt. This works in my favor too because if I think there might be mitigating circumstances it makes me less annoyed, and I like not being annoyed.
Edit: It's like when somebody cuts you off in traffic, what if they really had to poop?
Hell, I dunno why they didn't do it. I'm just saying, this is an unnecessarily inferior product. Tropico 4 can officially handle 2000 agents with homes, jobs, and dynamic real-time shortest-path pathfinding. OpenTTD can handle ... at least 15000 vehicles with user-specified destinations using dynamic, real-time shortest-path pathfinding using, I think, a modified version of A* search, and also has to deal with train signals. Note that OpenTTD caps at 15000 because they're of the belief that nobody would ever use more than 15000 vehicles, not because that's a system-imposed maximum, and this is an open-source project (i.e. FREE).
It's just mind-boggling as a computer scientist. I mean, this is the sort of thing we daydream about as students, how would we build SimCity if we grew up to be game developers... and what was delivered, at least in terms of traffic design, is way way dumber than what we conceived of even as undergrads.
How much do you think that having to host these cities on EA's servers and have them "always simulating" to a certain extent might effect the simulation? When you shut down Tropico 4, your city is paused. That is not entirely the case with SimCity 5, though I'm not sure exactly to what extent the simulation is still running on their servers. Inputs and outputs to the region are surely running, but what's inside...is somewhat less clear.
I wonder....can you "visit" someone else's city and watch it burn to the ground while they're not logged on?
How hard would it have been to have the agents path independently of each other? The way I'm thinking of it is in layers; layer 1 would be workers, layer 2 shoppers, layer 3 water, layer 4 sewage, etc. They all follow the same rules, they all follow the same path (the roads) but they act independently of each other so one doesn't affect the other. But what we apparently have is all of those 'layers' mashed together and interfering with each other. There's no logic to sewage backing up or power or water not getting into a city because the roads are congested.
Also, do the agents all clock on and off at the same time? Or are they staggered, e.g. industry starts at 8am, commercial and school 9am, school finishes at 3pm, industry at 5pm and commercial 6pm. Because that seems a logical thing to do.
I'm sorry, but what?
I have never seen my water, power, or sewage slow down because of traffic congestion. Are you talking about something that happens on the region?
If you buy in services from your neighbours it quite often just stops working because of problems elsewhere with the pathing.
How hard would it have been to have the agents path independently of each other? The way I'm thinking of it is in layers; layer 1 would be workers, layer 2 shoppers, layer 3 water, layer 4 sewage, etc. They all follow the same rules, they all follow the same path (the roads) but they act independently of each other so one doesn't affect the other. But what we apparently have is all of those 'layers' mashed together and interfering with each other. There's no logic to sewage backing up or power or water not getting into a city because the roads are congested.
Also, do the agents all clock on and off at the same time? Or are they staggered, e.g. industry starts at 8am, commercial and school 9am, school finishes at 3pm, industry at 5pm and commercial 6pm. Because that seems a logical thing to do.
Actually, the problem of pathing them all together is exponentially harder than dealing with the layers individually. Just from a computational complexity standpoint.
We've actually been having quite a good laugh in the planning group as to how horrible the pathfinding is. There is basically an entire academic conference (or half of two, the Symposium on Combinatorial Search and Artificial Intelligence in Interactive Digital Entertainment) discussing "How the hell do I path thousands of things without collisions in a way that's not super expensive".
I mean, I don't expect to see bleeding edge pathfinding research in all things, but I've definitely failed students that put forth better efforts. What's hilarious is that EA has had some *major* pathfinding brilliance in house. The pathfinding algorithms in the newest C&C game and Dragon Age are actually still pretty relevant to the research community despite being a few years old at this point. Just, you know, big big breakthroughs. I guess the left hand doesn't know what the right is doing over there.
Well to be fair EA is a publisher (among other things) and C&C and Dragon Age were developed by Phenomic Game Development (most recently at least) and Bioware respectively.
Neither of which are Maxis.
Not to say that there isn't some sort of resource sharing going on (there most likely is), but that's like saying that because of an innovation in World of Warcraft, Prototype by Radical Entertainment should have certain things because they are all under Activision.
Not that i'm excusing the traffic pathfinding mind you, that's a separate issue.
Delphinidaes on
NNID: delphinidaes Official PA Forums FFXIV:ARR Free Company <GHOST> gitl.enjin.com Join us on Sargatanas!
So how does the water, sewage, and power pathing work? The sewage seems to all go towards one outflow or processing plant until that one is full. The water and power, however, I have no fucking clue.
Buying services from neighbors doesn't work because the game doesn't give you an excess amount of the utility. Because of this, it can take a long time for each node to get to the right place.
I'm looking through the thread, but I haven't seen it mentioned yet - it looks like some frigging geniuses managed to get the game running offline, including a debug mode and the possibility of pretty extensive mods. That's seriously impressive. Has anyone tried it yet, at a risk to their Oigin account?
How hard would it have been to have the agents path independently of each other? The way I'm thinking of it is in layers; layer 1 would be workers, layer 2 shoppers, layer 3 water, layer 4 sewage, etc. They all follow the same rules, they all follow the same path (the roads) but they act independently of each other so one doesn't affect the other. But what we apparently have is all of those 'layers' mashed together and interfering with each other. There's no logic to sewage backing up or power or water not getting into a city because the roads are congested.
Also, do the agents all clock on and off at the same time? Or are they staggered, e.g. industry starts at 8am, commercial and school 9am, school finishes at 3pm, industry at 5pm and commercial 6pm. Because that seems a logical thing to do.
I'm sorry, but what?
I have never seen my water, power, or sewage slow down because of traffic congestion. Are you talking about something that happens on the region?
If you buy in services from your neighbours it quite often just stops working because of problems elsewhere with the pathing.
Is that when you are only relying on their resources or when you are using it to just cover part of your own?
Because I know there were issues with region sharing in general (most likely related to the server issues) and that things could be delayed due to that (Gifts for example, and I would assume power, water, etc) but I don't think any of that was ever related to traffic issues. At least certainly not within the city as you can clearly see the stuff flowing fine despite having crippling traffic issues within your own city.
Now whether or not it all compresses in to one thing in between cities, I have no idea. But I would doubt it, it's more likely related to the server issues managing the stuff between cities for multiple players. (Partly why I would think the cheetah speed helped, but that's just conjecture)
NNID: delphinidaes Official PA Forums FFXIV:ARR Free Company <GHOST> gitl.enjin.com Join us on Sargatanas!
A job is taken out of the graph for the first worker that reaches it, but what if the first worker in your list is on the other side of the city from where the remaining jobs are, and there are actually workers down further in the list who are closer to those jobs. The workers on the other side of the city would get the jobs and travel all the way across town for them, while the workers down the list who are closer may not get any job.
A better way to do it is first divide the city graph using a 2-d spacial sub division structure (quadtree for example), and perform the job search at the lowest level of the tree, then recurse to the next level and perform the search for the remaining workers, etc until finally you perform the search at the level of the entire city. This way, you will assign workers closer to the jobs first.
I was just writing that off the top of my head. Obviously there are superior versions possible. I'm just confident that this basic model could handle ... probably tens of thousands of agents, and is simple enough that laypeople could understand it and play with it. But I'm not sure the concern you bring up is actually an issue. The total distance traveled by the worker Sims would be roughly the same, whether it's a worker traveling 10 and another traveling 1 or both traveling 5, though one might be preferable to the other for traffic-easing reasons (or maybe some random combination of them would be best for simulation purposes). Might be able to find some edge cases that break that with specific road layouts. Or in situations with unemployment. But that one's relatively easy to fix, even using the same basic algorithm - if you have unemployment, flip the search so it comes out from workplaces rather than homes. (Or just tell them to stop having unemployment, lolololol.)
Which trucks deliver the coal and metal to smelter plants, the trade port trucks or the smelter plant trucks?
I want to say the Trade port trucks, but i'm not 100% on that. I think the smelter ones deliver the finished product back to the Trade port.
Trade trucks (if the goods are in storage at the trade depot/port) and trucks that are garaged at coal/metal mines will deliver to smelter plants. Kind of like how coal mines will deliver Coal to Coal power plants.
Trucks handle outgoing shipments in most (all?) cases, I think.
Posts
I've spent a fairly large amount of time on Simcity 4 and while the simulation is pretty robust, it was the tweaking and mods that really made it something spectacular.
I wish I could post some screenshots but i'm at work, if you're willing to spend the time downloading community addons it really adds tremendously to the realism. 4 also benefits from 10 years of more powerful computers which means that the huge regions that quickly became unplayable are kind of just steamrolled by the amount of RAM and processing power in home pc's, even with a huge lack of optimization.
It will all come down to the addon and modding that Maxis introduces into 5 which hopefully will increase the depth of the game as much as they did for 4.
Ah in my case I am turning all my alloy into processors, and then into TV's and Computers so the train delivery has been pretty solid for me. Before that all my alloy deliveries were via truck until my processing plants were set up.
Official PA Forums FFXIV:ARR Free Company <GHOST> gitl.enjin.com Join us on Sargatanas!
Factories still complaining about unskilled workers.
What is your tech level? Should be 3 yellow gears below your education level. That is what determines your standards for industry.
Shogun Streams Vidya
2 1/2 gears. Though, my education level can't be any higher. I have 3500/3500 students enrolled in my universities.
Also be sure you are educating them in a university or community college, I don't believe grade school, library, or highschool will count for the college educated workers thing. Also be sure that your buildings are green colored when you check your education data map and that your industry is located in a position to readily receive those residents.
It coul dbe you have outside workers coming in who are not skilled.
Official PA Forums FFXIV:ARR Free Company <GHOST> gitl.enjin.com Join us on Sargatanas!
they will complain and move out until it fills to full and then they will come back. There are several stages of high tech industry. I've also noticed odd things with education and tech level where my education would get stuck at 1.5 hats but my tech level would quickly climb to full. Then it would bottom out completely and my education would start to rise.
Shogun Streams Vidya
I regularly see sims asking for me to add more trains but this doesn't seem an option?
I assume this is done by building a rail network into the city using the heavy rails and plopping more stations.
Possibly, but there is a request you get from industry saying "Not enough college educated workers". It could just be flavor text, but all those messages went away when I bulldozed my grade school and highschool and just had a community college and university.
*Edit* Actually, isn't there a big image of all the messages that pop up and what they mean? I think it's from the strategy guide but I can't find it off hand
Official PA Forums FFXIV:ARR Free Company <GHOST> gitl.enjin.com Join us on Sargatanas!
Yeah, adding more stations adds more trains, but take a guess as to what happens with those.
Is the answer pathfinding related by any chance.
Uh, I don't recall D3 leaving me with any "wtfbbq" bugs, like buildings just stop working, or infinite people flooding into a building, or invisible people who won't ever leave, etc,
There's also the part where it'd be fucking easy to do a lot better than they're currently doing. I mean, the traffic AI is literally just as dumb as rocks. It's the sort of childish model I'd expect from a first-year CS student. Actually, last term, we made our first-year CS students do an assignment with pathfinding and trucks that "claimed" locations to visit and still used shortest-path pathfinding. So to be clear, if you were to ask my first-year CS students on how to implement SimCity, they'd come up with a superior model.
Seriously. I could not in good conscience give a CS degree to someone who thought that was an appropriate traffic simulation model.
I think it's easy to think it's that cut and dry, that Maxis is just filled with idiots, but I think that's a little presumptuous. There is probably a reason why they went in this direction. There is obviously a lot of information we are not privy to.
And I don't play D3, and I'm aware that there are definitely folks who didn't have a lot of problems with the game, but from the media coverage of the game's launch and thereafter it was pretty clear that a lot of people were having issues even if it wasn't universal.
Server issues are different from in-game bugs. Both launches have had server issues, but D3 had very few "wtfbbq?!" bugs. You can sort of get a sense of them here:
http://www.diablowiki.com/Diablo_III_Patches
I just don't understand this hurgleburgle "SimCity's not that bad" defense. I can literally describe for you in a short paragraph a superior Sim/traffic model that would be well within your computer's ability to process. I was discussing this with a friend of mine who's an AI post-doc, and we both estimated that implementing a Tropico 4-esque model would require just less than a millisecond in computing time for thousands of agents. (Hence, see how people have modded Tropico 4 to work with thousands of citizens, each of which are substantially more detailed than the Sims in SimCity.) With pathfinding occurring at specific intervals, it wouldn't be too hard to implement something for hundreds of thousands of agents, imo, if we fixed paths after a pathfinding cycle and only recomputed if the any paths were invalidated. (Managing hundreds of thousands of agents, of course, gets a little bit heavier in terms of memory consumption, but that's another story.) Mass transit's slightly harder, but we, in minutes, came up with several heuristics that were superior to "randomly turn at intersections".
It's a sad day when I point at Maxis' SimCity and say, "Boy, Kalypso Media, the German publishers of Airline Tycoon 2, did this a whole lot better."
I think so. At least, when I tried with three stations there were more trains but one station (the first one the trains arrived at) would have about 6000 riders per day whilst the other two had around 400-500.
Also, thinking about the sims and their homes and jobs, it made me think this guy was in charge of the simulation;
It's just easier for me to give them the benefit of the doubt. In my experience there are almost always variables and circumstances that are only apparent when you are very close to the problem, and I don't just mean computing problems but life in general.
I mean I just can't see them failing to implement something you say is really easy without a good reason.
Can I reset the region or do I just abandon everything and let EA worry about it?
Model transport system as a graph, with intersections and transport hubs as vertices, roads and transit paths as edges. Edges are costed according to road / path length. (Optional: weight edges according to type, so that, for example, high-density roads of a fixed length are preferred over low-density roads of the same length.) Insert every job in the city into the network as sink vertices. This will drastically increase the number of vertices in the graph, but won't increase complexity, as those vertices are essentially leafs, and no decisions really need to be made, since they're all degree 2. For each worker agent, use weighted-BFS to find the closest suitable job. Note that we can agglomerate multiple agents together, rather than performing a unique search for each agent, and iterative searches are likely to be related but whatever. The agent now has the closest suitable available job, and immediately knows the shortest / lowest cost path there, which it can immediately follow. Filled sink vertices can be removed. Commute homes are automatic reversals of the to-work path.
Now, notably, we can also change the model so that once agents have found a job, they retain that job. If a city is massively rebuilt, that would be problematic, but the nature of the current game model is that this process needs to happen only every 1440 seconds. So we actually have something like 2 minutes to perform all these calculations for the majority of the agents, and only actually need real-time calculations for the select group of citizens / jobs that get fucked with in the last few seconds before rush hour, so practically speaking, we could use a much more complex algorithm even, and this would have the advantage that Sims retain certain patterns. Use some sort of marker to indicate when recomputation needs to happen (or make it timed, so Sims follow the same pattern for a month or two, then refind jobs, etc,)
Hell, the emergency vehicles all rushing to the same emergency? Solution: each emergency has a single slot. Gets claimed by the nearest vehicle, found using weighted-BFS again, running in max O(n) time, average O(lg n) time. If you want, insert a little fudging to assign agents to their nearest intersection rather than along roads. Non-ideal, but still easily understood and manipulable, and would allow for a lot more agents, since the number of BFSes done would correlate to the number of intersections rather than agents. (Optional: insert "fake" intersection vertices along long, unintersected roads to alleviate adversarial road design.)
None of the content here is beyond a 2nd-year CS student, btw, maybe a 3rd-year. Coming up with it might test them, but I'd certainly expect, in a class of 2nd years, that a number of them would come up with this or something similar.
Also, do the agents all clock on and off at the same time? Or are they staggered, e.g. industry starts at 8am, commercial and school 9am, school finishes at 3pm, industry at 5pm and commercial 6pm. Because that seems a logical thing to do.
Edit: It's like when somebody cuts you off in traffic, what if they really had to poop?
Until they patch it im afraid your city will permanately count those visitors. The same thing is occuring in my city, completely bulldozed yet every month it thinks i have 600k visitors and 60k commuting shoppers.
I also noticed an invisible object telling me it needs a road connection when I have the road menu up and I could not figure out what it was until I randomly saw some bush thing in the middle of the sky
Hell, I dunno why they didn't do it. I'm just saying, this is an unnecessarily inferior product. Tropico 4 can officially handle 2000 agents with homes, jobs, and dynamic real-time shortest-path pathfinding. OpenTTD can handle ... at least 15000 vehicles with user-specified destinations using dynamic, real-time shortest-path pathfinding using, I think, a modified version of A* search, and also has to deal with train signals. Note that OpenTTD caps at 15000 because they're of the belief that nobody would ever use more than 15000 vehicles, not because that's a system-imposed maximum, and this is an open-source project (i.e. FREE).
It's just mind-boggling as a computer scientist. I mean, this is the sort of thing we daydream about as students, how would we build SimCity if we grew up to be game developers... and what was delivered, at least in terms of traffic design, is way way dumber than what we conceived of even as undergrads.
I'm sorry, but what?
I have never seen my water, power, or sewage slow down because of traffic congestion. Are you talking about something that happens on the region?
Official PA Forums FFXIV:ARR Free Company <GHOST> gitl.enjin.com Join us on Sargatanas!
Huh, I wonder if this actually syncs.
Actually, the problem of pathing them all together is exponentially harder than dealing with the layers individually. Just from a computational complexity standpoint.
We've actually been having quite a good laugh in the planning group as to how horrible the pathfinding is. There is basically an entire academic conference (or half of two, the Symposium on Combinatorial Search and Artificial Intelligence in Interactive Digital Entertainment) discussing "How the hell do I path thousands of things without collisions in a way that's not super expensive".
I mean, I don't expect to see bleeding edge pathfinding research in all things, but I've definitely failed students that put forth better efforts. What's hilarious is that EA has had some *major* pathfinding brilliance in house. The pathfinding algorithms in the newest C&C game and Dragon Age are actually still pretty relevant to the research community despite being a few years old at this point. Just, you know, big big breakthroughs. I guess the left hand doesn't know what the right is doing over there.
A job is taken out of the graph for the first worker that reaches it, but what if the first worker in your list is on the other side of the city from where the remaining jobs are, and there are actually workers down further in the list who are closer to those jobs. The workers on the other side of the city would get the jobs and travel all the way across town for them, while the workers down the list who are closer may not get any job.
A better way to do it is first divide the city graph using a 2-d spacial sub division structure (quadtree for example, though others will be more optimal since cities will not be uniform), and perform the job search at the lowest level of the tree, then recurse to the next level and perform the search for the remaining workers, etc until finally you perform the search at the level of the entire city. This way, you will assign workers closer to the jobs first.
"Orkses never lose a battle. If we win we win, if we die we die fightin so it don't count. If we runs for it we don't die neither, cos we can come back for annuver go, see!".
How much do you think that having to host these cities on EA's servers and have them "always simulating" to a certain extent might effect the simulation? When you shut down Tropico 4, your city is paused. That is not entirely the case with SimCity 5, though I'm not sure exactly to what extent the simulation is still running on their servers. Inputs and outputs to the region are surely running, but what's inside...is somewhat less clear.
I wonder....can you "visit" someone else's city and watch it burn to the ground while they're not logged on?
If you buy in services from your neighbours it quite often just stops working because of problems elsewhere with the pathing.
Well to be fair EA is a publisher (among other things) and C&C and Dragon Age were developed by Phenomic Game Development (most recently at least) and Bioware respectively.
Neither of which are Maxis.
Not to say that there isn't some sort of resource sharing going on (there most likely is), but that's like saying that because of an innovation in World of Warcraft, Prototype by Radical Entertainment should have certain things because they are all under Activision.
Not that i'm excusing the traffic pathfinding mind you, that's a separate issue.
Official PA Forums FFXIV:ARR Free Company <GHOST> gitl.enjin.com Join us on Sargatanas!
Buying services from neighbors doesn't work because the game doesn't give you an excess amount of the utility. Because of this, it can take a long time for each node to get to the right place.
http://arstechnica.com/gaming/2013/03/hackers-open-up-offline-play-modding-tools-for-simcity/
Edit: Nevermind, found the page with some posts about it.
Is that when you are only relying on their resources or when you are using it to just cover part of your own?
Because I know there were issues with region sharing in general (most likely related to the server issues) and that things could be delayed due to that (Gifts for example, and I would assume power, water, etc) but I don't think any of that was ever related to traffic issues. At least certainly not within the city as you can clearly see the stuff flowing fine despite having crippling traffic issues within your own city.
Now whether or not it all compresses in to one thing in between cities, I have no idea. But I would doubt it, it's more likely related to the server issues managing the stuff between cities for multiple players. (Partly why I would think the cheetah speed helped, but that's just conjecture)
Official PA Forums FFXIV:ARR Free Company <GHOST> gitl.enjin.com Join us on Sargatanas!
I was just writing that off the top of my head. Obviously there are superior versions possible. I'm just confident that this basic model could handle ... probably tens of thousands of agents, and is simple enough that laypeople could understand it and play with it. But I'm not sure the concern you bring up is actually an issue. The total distance traveled by the worker Sims would be roughly the same, whether it's a worker traveling 10 and another traveling 1 or both traveling 5, though one might be preferable to the other for traffic-easing reasons (or maybe some random combination of them would be best for simulation purposes). Might be able to find some edge cases that break that with specific road layouts. Or in situations with unemployment. But that one's relatively easy to fix, even using the same basic algorithm - if you have unemployment, flip the search so it comes out from workplaces rather than homes. (Or just tell them to stop having unemployment, lolololol.)
Trucks handle outgoing shipments in most (all?) cases, I think.