• Welcome to SC4 Devotion Forum Archives.

NAM Traffic Simulator Development and Theory

Started by z, August 02, 2008, 05:07:50 PM

Previous topic - Next topic

0 Members and 3 Guests are viewing this topic.

ldog

Quote from: z on November 07, 2009, 11:56:36 PM
I have also been doing more tests along these lines, and the more I do, the more I think we're dealing with two different constants:  1) There's the "perfect" PH, which always produces the fastest paths.  Determining this exactly experimentally is extremely difficult, as it means calculating and comparing the travel time of hundreds, if not thousands, of paths.  We know it's somewhere near .003; as Tropod claimed that that was the number, I'm willing to go with it until proven otherwise.  2)  There's the "no abandonment" PH, which is the level that no abandonment due to commute time ever occurs because the pathfinder fails to find a valid path.  Preliminary experiments seem to show that the value of this second constant is somewhere between .002 and .0025.

I would reach a different conclusion.
Some of us around here are really stuck on "perfect ph = .003" :P
At any rate, the experimentation is providing new information, which I think we can all agree is a good thing.

Quote from: z on November 07, 2009, 11:56:36 PM
I assume you are using your simulator on your testbed.  If so, try using Simulator Z and see if that makes a difference.  If you get a lot of abandonment with Simulator Z, I'd be interested in looking into this more.

Yup. You guys keep plugging your sims. I get it, I get it. Part of why I am experimenting with my own was cause I wanted to avoid this whole raging "coke vs pepsi" debate. If I used A or Z, then everything would be seen as biased, and then I'd have to do run every experiment twice or you guys would never be happy. I'm just trying to get to the bottom of how the blasted thing really works.  I have appreciated the valuable insight I get from BOTH of you. Now I didn't go and pick B as "the third neutral candidate"...I decided to do my own thing. I looked at B after and noticed how similar it was, but then that was no shocker. Mott's theories were a big influence. I suspect they were for you and Jason as well and you guys probably started with something similar and went from there.

At this point I am pretty sure with my super craptastic building methods I can make any simulator cry uncle :P

Seriously though, if I jacked my commute time up to 600 in mine even the abandonment would likely go away. I am still not overly concerned about the abandonment at the moment; I was just adding that my findings tend to confirm yours, even when they weren't the thing I was trying to find lol. Likely it will still come down to a design choice at some point on when and how much abandonment is allowable from the simulator to as Jason says "preserve some level of difficulty".

Just making the ph really low and the max commute time very high is still not the answer for all occassions. Besides abandonment, what other penalty can the traffic simulator impose on us for bad citybuilding decisions? Chaos of cars spam? I admit that is annoying but not enough. If we could fully congest all available networks to red all the time but everything still worked then as long as you provide a valid route you can do whatever you want. Of course there could be other more tangible effects but I can't think of them at the moment (which doesn't mean I think I have been exhaustive; I am still asking the question...I don't feel I have answered it)

b22rian

Quote from: z on November 08, 2009, 03:23:51 PM
A quick but important correction:  Simulator B actually has a PH eight times higher than Simulator Z's:  .025 vs. .003.  It's easy to miss that extra zero.  But this has major implications for what follows in Brian's post.

More on some other topics later...


Thanks, for that important correction Steve...

          Both the rest of my post above , and your recent findings on the testing of the different PH values..
Make a lot more sense to me now.. I apologize to those I may have caused additional confusion to, on account of
my oversight .

Brian

xxdita

Quote from: ldog on November 08, 2009, 04:04:20 PM
Yup. You guys keep plugging your sims. I get it, I get it. Part of why I am experimenting with my own was cause I wanted to avoid this whole raging "coke vs pepsi" debate. If I used A or Z, then everything would be seen as biased, and then I'd have to do run every experiment twice or you guys would never be happy. I'm just trying to get to the bottom of how the blasted thing really works.  I have appreciated the valuable insight I get from BOTH of you. Now I didn't go and pick B as "the third neutral candidate"...I decided to do my own thing. I looked at B after and noticed how similar it was, but then that was no shocker. Mott's theories were a big influence. I suspect they were for you and Jason as well and you guys probably started with something similar and went from there.

I'm in the same boat. I've learned enough from Jason, and by tinkering under the hood, to be able to personalize the traffic sim to suit my particular needs, and my playing style. It may just be a jacked up version of A, but I knew the research involved in it, and had tested it thoroughly, along with any changes I've made.

I've always been willing to test anything, but for me to actually be able to give Barby my thumbs up on any mod going to the LEX, or to recommend it to someone else... that's gonna take some serious convincing.

QuoteAt this point I am pretty sure with my super craptastic building methods I can make any simulator cry uncle :P

I can make em scream... But this comes in handy when people want testers for their latest simulator, huh?

QuoteSeriously though, if I jacked my commute time up to 600 in mine even the abandonment would likely go away. I am still not overly concerned about the abandonment at the moment; I was just adding that my findings tend to confirm yours, even when they weren't the thing I was trying to find lol. Likely it will still come down to a design choice at some point on when and how much abandonment is allowable from the simulator to as Jason says "preserve some level of difficulty".

If I increase my commute time any at all, my abandonment issues go away almost instantly. Why? Because now my Sims can travel 3 tiles over to find work... and that always ends up in an unplayable region for me. It may have been ok for me to have to travel an hour or more to work, in real life, but the problems it can create in SC4 are just not worth it to me. And as there is no truly regional traffic simulator (unless I've missed something?), it would be my guess that the game isn't designed for extended commutes between tiles.

QuoteJust making the ph really low and the max commute time very high is still not the answer for all occassions. Besides abandonment, what other penalty can the traffic simulator impose on us for bad citybuilding decisions? Chaos of cars spam? I admit that is annoying but not enough. If we could fully congest all available networks to red all the time but everything still worked then as long as you provide a valid route you can do whatever you want. Of course there could be other more tangible effects but I can't think of them at the moment (which doesn't mean I think I have been exhaustive; I am still asking the question...I don't feel I have answered it)

It seems like Jason and I tested this out too... in a thread at Simtropolis. By providing ped malls as the ONLY path from residential zones to any jobs, we proved that Sims will do everything they can to get to work, even if it means going beyond the commute time set up in the traffic sim, as long as there is a valid route provided. (I don't have a link... Jason might?)

z

#303
So many posts to respond to.  I'll just start here, get as far as I can, and continue tomorrow.

[Lenny quotes my conclusion that there are two different pathfinding constants.]
Quote from: ldog on November 08, 2009, 04:04:20 PM
I would reach a different conclusion.
Some of us around here are really stuck on "perfect ph = .003" :P
At any rate, the experimentation is providing new information, which I think we can all agree is a good thing.

Here's some more information, which is part of what led to my conclusion.  When I take the standard Simulator Z and drop the PH to .0025, abandonment doesn't decrease at all.  In fact, it's the same level that it was at .0025 with my older version of Simulator Z, even though that version showed more abandonment at .003.  ()what()  This certainly isn't proof, but it does make me think that there are two different mechanisms going on here.

The .003 number for the perfect PH is just a hypothesis, but it seems a pretty good one, based on the evidence.  I also put a lot of weight in what the7trumpets and Tropod say.  If you don't think .003 is the right number, all you have to do is find one path that's not the fastest path, in any city that's been running with that PH for at least several decades.

From me:
QuoteI assume you are using your simulator on your testbed.  If so, try using Simulator Z and see if that makes a difference.  If you get a lot of abandonment with Simulator Z, I'd be interested in looking into this more.
From Lenny:
Quote
Yup. You guys keep plugging your sims. I get it, I get it. Part of why I am experimenting with my own was cause I wanted to avoid this whole raging "coke vs pepsi" debate. If I used A or Z, then everything would be seen as biased, and then I'd have to do run every experiment twice or you guys would never be happy.

I think you have missed my point here.  I am not trying to "plug" my simulator; I just want to see if what you have found is a general principle, in which case I should look at it, or it's just a property of your simulator, in which case it's not my concern.  As I have shown with my example of the older version of Simulator Z, there are other factors beside maximum commute time and PH that affect the amount of abandonment due to commute time.  If you want the claims in your post to be more than mere speculation, you need to test Simulator Z.  You may be right about your conclusion; I don't know, since I haven't played your city.  But you'll never really know without testing.

QuoteAt this point I am pretty sure with my super craptastic building methods I can make any simulator cry uncle :P

As Jason said to me, "Prove it."  I thought his request was reasonable, and I did exactly what he said.  What about you?  ;)

QuoteSeriously though, if I jacked my commute time up to 600 in mine even the abandonment would likely go away.

It seems to me that you are getting deeper and deeper into mere speculation.  Let's do a real test.  We'll take Simulator Z (Low), and get rid of that 600 maximum commute time.  We could use Simulator A's value of 17, but I always like to test the extremes, as they often give the most interesting results.  So I'll use the original Maxis value of 6 for the maximum commute time, with 4 for the mass transit maximum commute.  That gives the Sims all of three minutes to get from their homes to their jobs.  I'll admit, I'm not expecting the results to be pretty when I start this test.

[EDIT:  I later reran these test with a maximum commute time of 4, giving the Sims a total of two minutes to get from their homes to their jobs.  There was a little bigger bump in the commute time graph during the transition, but all the other results were identical to that of this test.]

We start with the same city, same time - January, 433.  I run the simulator for eleven years and take a look around.  Very interesting!  I took a picture, but there's no need to show it, because after running the simulator for another 20 years, the picture is identical.  Here's what my city looks like after running with a maximum commute time of 6 for 31 years:



There is no abandonment anywhere in this picture.  There's not even any additional downgrading.  The only two downgraded buildings in this picture have been downgraded since before the test began.  Would anyone have predicted this?  I didn't.

These results were so surprising to me that I had to question them.  Again, was I really using the proper simulator?  How could I tell?  Well, normally I don't pay any attention to the commute time graph, since Simulator Z normally has effectively unlimited commute time, but this time that graph should show something.  If the simulator really did rearrange all the Sims' routes so that they could reach their jobs in three minutes, you would really expect to see a significant change in the commute time graph.  Let's see:


This certainly is about what we would expect to see.  Commute time had plenty of room to fluctuate under the standard simulator.  But it starts dropping sharply at the transition, and levels out about 60% down from its peak.  It eventually settles a line that is almost flat; I would expect that the average commute time is very close to the maximum here.  And although I always stress that the numbers in the commute time graph cannot be made perfectly accurate, they're not that far off here; maximum commute time is 6, and the line levels out around 7.5.

OK, to me that's excellent evidence that we're using the right simulator.  There's more evidence coming later.  But first, let's check in with our old friend, the Pop & Jobs Graph:


This is another interesting graph.  There's certainly no divergence here - there shouldn't be if there's no abandonment or downgrading in the city.  Instead, the trends that were in place before the switch simply continue.  The R$$$ population continues to increase, as it has been doing since I introduced Simulator Z v1.2, although the increase is gradually leveling off.  Meanwhile, the R$$ population is decreasing somewhat faster than the R$$$ is increasing.  Total population has decreased slightly.  This is just the opposite of what we had seen before; in this case, as I verified by looking around the city, R$$$ buildings that had been downgraded to R$$ for centuries are slowly being upgraded to R$$$.  Other R$$ buildings are being replaced with R$$$ buildings.  How is the simulator doing this when the maximum commute time has been cut by 99%?  The Traffic Volume Graph gives us another clue:


The Sims are moving out of their cars and taking the subway more.  (The downward spikes are artifacts that occur when I save the game).  This movement to the subway makes sense, as the subway is much faster than cars.  Even so, it's not a very big difference, and it's especially impressive in terms of pathfinder ability when you remember that subway speeds were reduced by 30% in Simulator Z v1.2, to 105 kph.  Car speeds were left unchanged at 50 kph.

So everything is consistent here; I'm just really surprised at how well the pathfinder is doing.  If .003 isn't the perfect PH, then it's awfully close.  What do you think, Lenny?

Finally, here's the congestion minimap for the city:


All in all, it looks pretty reasonable.  My conclusion:  The effect of the maximum commute time in the game is highly overrated.  The PH appears to be the most important parameter by far in the traffic simulator.

QuoteI am still not overly concerned about the abandonment at the moment; I was just adding that my findings tend to confirm yours, even when they weren't the thing I was trying to find lol. Likely it will still come down to a design choice at some point on when and how much abandonment is allowable from the simulator to as Jason says "preserve some level of difficulty".

Just making the ph really low and the max commute time very high is still not the answer for all occassions. Besides abandonment, what other penalty can the traffic simulator impose on us for bad citybuilding decisions? Chaos of cars spam? I admit that is annoying but not enough.

OK, how about an increase in noise that makes residential areas less desirable, an increase in pollution, and a decrease of up to 14 points in the local Mayor Rating?  These all occur when your congestion gets too bad, and they all have side effects themselves.  As you may remember, Shadow Assassin found that the lower Mayor Rating was a contributing cause to riots in his city.  If your congestion is widespread, this can have a significant effect on your global Mayor Rating, which affects which rewards you can get.  To quote the Prima Guide regarding the awards:

QuoteIn addition... they improve your city in myriad other ways (enhancing desirability, boosting EQ or HQ, further increasing Mayor Rating, to name a few).

Note the last point.  This means that bad traffic can eventually cost you a lot more than 14 points in your Mayor Rating, ending up having a very major effect on the game, even with no abandonment due to commute time.  And abandonment due to demand may still happen, of course, and if you look at the above quote, you can see how bad traffic can start a cycle than can result in exactly that.

Quote from: xxdita on November 08, 2009, 09:04:32 PM
If I increase my commute time any at all, my abandonment issues go away almost instantly. Why? Because now my Sims can travel 3 tiles over to find work... and that always ends up in an unplayable region for me. It may have been ok for me to have to travel an hour or more to work, in real life, but the problems it can create in SC4 are just not worth it to me. And as there is no truly regional traffic simulator (unless I've missed something?), it would be my guess that the game isn't designed for extended commutes between tiles.

I agree - the game wasn't designed for extended commutes between tiles.  That doesn't mean that we can't make it work, though.  One of the main original design goals of Simulator Z was to provide what you call a truly regional traffic simulator.  That's what the Ultra capacity level is really for - it's for when you have Sims commuting not only in your city, but coming and going from various surrounding cities as well, thus being counted as commuters in all of them.  But as you indicate, the limitations in the game to doing this are serious, and they aren't something that the traffic simulator can completely overcome on its own.  However, with CAM 2.0, I think we will finally have a platform where effective intercity commuting over many tiles can actually work well.  The ESURE project, providing express subways that allow you to control to some extent where the Sims commute, will also help in this area.

I'll get to more posts tomorrow...

b22rian

Quote from: z on November 09, 2009, 02:03:43 AM

[Lenny quotes my conclusion that there are two different pathfinding constants.]
Here's some more information, which is part of what led to my conclusion.  When I take the standard Simulator Z and drop the PH to .0025, abandonment doesn't decrease at all.  In fact, it's the same level that it was at .0025 with my older version of Simulator Z, even though that version showed more abandonment at .003.  ()what()  This certainly isn't proof, but it does make me think that there are two different mechanisms going on here.




   Steve ,
.. on the second .0025.. I assume you meant to type .003 right ?
In fact, it's the same level that it was at .xxxx

Thanks, brian

z

No, it's correct as it reads.  I'm comparing two versions of Simulator Z here.  The story began back in Lenny's thread somewhere, which may make it a little confusing.

b22rian

Quote from: z on November 09, 2009, 03:44:12 AM
No, it's correct as it reads.  I'm comparing two versions of Simulator Z here.  The story began back in Lenny's thread somewhere, which may make it a little confusing.

   Ok, thanks for clearing that up Steve.. I never knew the older Z had actually a lower PH , that's quite
interesting.. Although I think i was more surprised when I found out Sim B had a PH actually 8 times higher
than our current sim Z.. I mean i expected it to be higher , but perhaps not to that degree,,

  However, although it seems  like quite a bit (8 X), I take it the actual differences the average gamer sees from
the much high PH setting alone would not be quite as dramatic... and it takes all the different property changes
between sim Z and the other 2 traffic sims,  (A and B ) to make a noteworthy difference ?

Thanks, brian

z

Quote from: b22rian on November 09, 2009, 04:05:34 AM
   Ok, thanks for clearing that up Steve.. I never knew the older Z had actually a lower PH

Actually, it didn't.  I just gave it one for a particular test.

Quote
  However, although it seems  like quite a bit (8 X), I take it the actual differences the average gamer sees from
the much high PH setting alone would not be quite as dramatic... and it takes all the different property changes
between sim Z and the other 2 traffic sims,  (A and B ) to make a noteworthy difference ?

Simulator B has a longer maximum commute time:  24 minutes, vs. 17 for Simulator A.  This helps make up for the difference in PH, at least to some extent.  Although I just went through a long test showing how maximum commute time was overrated, it becomes more and more important as the PH rises.

As a postscript to my test, I think it's important to ask, How reasonable are the results I got from a theoretical point of view?  Let's assume that the pathfinder, which appears to be very smart here, put the Sims with the longest commute on subways.  The time in minutes required to cross a single square is given by the formula .96/speed, where speed in this case is 105 kph.  Therefore, traveling by subway, the time required to cross a single square is .009 minutes.  So in a single minute, a subway can cross 111 squares; in three minutes, the maximum one-way commute, it can cross 333 squares.  Since a large tile is only 256 squares long, we seem to have met the reasonableness criterion.  Of course, this doesn't give the Sims time for leisurely walks to the station.

Just for fun, I tried reducing the maximum commute time from 6 to 4 - that's two minutes one way.  Early results seemed to indicate that the game was able to handle that, but it crashed after less than an hour.  I can't say that I blame it...  :D

ldog

Quote from: z on November 09, 2009, 02:03:43 AM
[Lenny quotes my conclusion that there are two different pathfinding constants.]
Here's some more information, which is part of what led to my conclusion.  When I take the standard Simulator Z and drop the PH to .0025, abandonment doesn't decrease at all.  In fact, it's the same level that it was at .0025 with my older version of Simulator Z, even though that version showed more abandonment at .003.  ()what()  This certainly isn't proof, but it does make me think that there are two different mechanisms going on here.

The .003 number for the perfect PH is just a hypothesis, but it seems a pretty good one, based on the evidence.  I also put a lot of weight in what the7trumpets and Tropod say.  If you don't think .003 is the right number, all you have to do is find one path that's not the fastest path, in any city that's been running with that PH for at least several decades.

I am of course poking a little fun but if lowering the ph further changes performance then that would tend to indicate to me the perfect ph is indeed lower.
Trying to find a path that isn't the shortest would be like trying to find a needle in a haystack. Like I said early on about testing methods; one would have to count out every tile in a given route, modify it by congestion effects, figure out how long it actually took, and then start doing the same for all the alternate paths.

Quote from: z on November 09, 2009, 02:03:43 AM
From me:From Lenny:
I think you have missed my point here.  I am not trying to "plug" my simulator; I just want to see if what you have found is a general principle, in which case I should look at it, or it's just a property of your simulator, in which case it's not my concern.  As I have shown with my example of the older version of Simulator Z, there are other factors beside maximum commute time and PH that affect the amount of abandonment due to commute time.  If you want the claims in your post to be more than mere speculation, you need to test Simulator Z.  You may be right about your conclusion; I don't know, since I haven't played your city.  But you'll never really know without testing.

As Jason said to me, "Prove it."  I thought his request was reasonable, and I did exactly what he said.  What about you?  ;)

I still don't understand what it has to do with Simulator Z. Jason asked you to prove it because you were making comparisons between A & Z.
I thought it was clear I was joking when I was bragging about being able to break any sim.
Newsflash, it isn't possible in a pretty vanilla game without the CAM to get a population to the kind of levels most of y'all consider routine.
Especially when you use at most about 60% of a large tile. As I've maintained since the begining, Z is too large scale a simulator for my use.
It is pretty pointless for me to run any tests using Z in any capacity. My own sim right now, I dropped the caps back to defaults because I couldn't get enough traffic.
I find the default simulator with better pathfinding and longer commute, cvs and intersection tweaks can handle things pretty well in this situation. I need to scale it back even further to maintain some challenge level.

Quote from: z on November 09, 2009, 02:03:43 AM
It seems to me that you are getting deeper and deeper into mere speculation.  Let's do a real test.  We'll take Simulator Z (Low), and get rid of that 600 maximum commute time.  We could use Simulator A's value of 17, but I always like to test the extremes, as they often give the most interesting results.  So I'll use the original Maxis value of 6 for the maximum commute time, with 4 for the mass transit maximum commute.  That gives the Sims all of three minutes to get from their homes to their jobs.  I'll admit, I'm not expecting the results to be pretty when I start this test.

We start with the same city, same time - January, 433.  I run the simulator for eleven years and take a look around.  Very interesting!  I took a picture, but there's no need to show it, because after running the simulator for another 20 years, the picture is identical.  Here's what my city looks like after running with a maximum commute time of 6 for 31 years:

All in all, it looks pretty reasonable.  My conclusion:  The effect of the maximum commute time in the game is highly overrated.  The PH appears to be the most important parameter by far in the traffic simulator.

I shortened the above. A very interesting test. Which of course pushes me even deeper into mere speculation.
Because the other day you "proved" that the max commute time was much more important than thought.
Today you just "proved" that it isn't.
The more testing I do, the more confusing my results as well.

Quote from: z on November 09, 2009, 02:03:43 AM
OK, how about an increase in noise that makes residential areas less desirable, an increase in pollution, and a decrease of up to 14 points in the local Mayor Rating?  These all occur when your congestion gets too bad, and they all have side effects themselves.  As you may remember, Shadow Assassin found that the lower Mayor Rating was a contributing cause to riots in his city.  If your congestion is widespread, this can have a significant effect on your global Mayor Rating, which affects which rewards you can get.  To quote the Prima Guide regarding the awards:

Note the last point.  This means that bad traffic can eventually cost you a lot more than 14 points in your Mayor Rating, ending up having a very major effect on the game, even with no abandonment due to commute time.  And abandonment due to demand may still happen, of course, and if you look at the above quote, you can see how bad traffic can start a cycle than can result in exactly that.

Thanks.

jplumbley

Quote from: z on November 09, 2009, 02:03:43 AM
As Jason said to me, "Prove it."  I thought his request was reasonable, and I did exactly what he said.  What about you? 

You can't make claims saying something is better and not proving it.  And I did not say "Prove it.", I was not an a-- about it like it seems you are trying to make it seem with this quote.  I said:

Quote from: jplumbley on October 27, 2009, 09:12:48 PM
Quote from: z on October 27, 2009, 05:24:51 PM
There are plenty of people who have said that Simulator A with its value of .009 produces too much abandonment, and if you look through the threads, you will find them.  Here's just one sample quote, from sumwonyuno:

I would like you to proove this...  I have not had any problems what-so-ever with abandonment in my cities due to the PH being higher than yours.  If you are so confident that the PH causes abandonment, proove it with Simulator Z and raise the value to 0.009 and show me the abandonment.  Abandonment is not caused by one simple factor, it is caused by many, mostly Commute Time and Congestion in your City.  And if you want to do this test there is absolutely no way you can compare Simulator A and Z you must do it with one or the other.

Obviously since I am having trouble getting abandonment in my Cities with Simulator A I cannot run this test since I am already at such a low Commute Time and high PH compared to Simulator Z.  All I can say is I cannot reproduce your abandonment issues with a "Better Pathfinding" instead of "Perfect Pathfinding".

You made a claim that others had abandonment in Simulator A which you claimed came from a higher PH.  I do not have this problem, and have not seen it in my cities and know others with "healthy" cities as well with the 0.009 PH.  You made this claim with absolutely no content or even anything to suggest this could be a problem and you used a 3rd party as the claimant who then refuted your use of his/her name as an example.  So, yes I have every right to say "Prove it." like an a--, but I was trying to be nice about it and explained why I have trouble believing that claim.

So now, you have shown that it does exist as a problem.  But you have not shown anything that can provide an explanation of why it happens in one City but not another considering there are many people who have "healthy" cities using Simulator A.

Your flat out claim seems to have some faults in it at least at this point.  And you were very pre-mature in making that claim, it would serve much better if you stopped doing things like that.
"You learn something new everyday."

http://img517.imageshack.us/img517/169/nhpjplumbleykv3.gif
Bringing the new horizons closer to reality.

Berethor ♦ beskhu3epnm ♦ blade2k5 ♦ dmscopio jplumbley ♦ moganite ♦ M4346 ♦ Dedgren ♦ Ennedi Shadow Assassin ♦  Tarkus ♦ wouanagaine
Street Addon Mod - SAM

Pharaon-Kheops

well, I do have abandonment because of commute time, in my CAM cities, even in the early stages (about 40k to 50k). This occurs not matter what simulator I use (both 3 at high capacity though, and even at "ultra" capacity in the case of sim z). Judging which sim has the "worse" is clearly too subjective at these pop levels. Of course, I did verify that there was no other problems like desirability, crime, demand, etc...

Thinking about all this discussion, a question popped up in my mind: "could this problem come from a conflict between simcity_1.dat and the simulator dat, as it is the case with the CAM (about workforce doubling)?"
Has anybody tried to merge simulator with simcity_1.dat with datpaker (or to modify directly simcity_1.dat...)?
"to be is to do" - Kant
"to do is to be" - Sartre
"to be or not to be" - Shakespeare
"to be do be do" - Sinatra

z

#311
Quote from: ldog on November 09, 2009, 09:05:36 AM
I am of course poking a little fun but if lowering the ph further changes performance then that would tend to indicate to me the perfect ph is indeed lower.

I understand; I just wanted to toss a little more data your way.

Quote
Trying to find a path that isn't the shortest would be like trying to find a needle in a haystack. Like I said early on about testing methods; one would have to count out every tile in a given route, modify it by congestion effects, figure out how long it actually took, and then start doing the same for all the alternate paths.

I agree completely.  That's a big reason why I'm not about to do this experiment.

Quote
I still don't understand what it has to do with Simulator Z. Jason asked you to prove it because you were making comparisons between A & Z.

It's not Simulator Z as such, but the way the various parameters are tuned in it.  I was basically saying that your statements about simulators in general conflicted with my experience with Simulator Z, and since you made them without testing Simulator Z, their validity seemed questionable.  The obvious way to resolve this would be to test Simulator Z.

Quote
I thought it was clear I was joking when I was bragging about being able to break any sim.

Well yes, I did see the smiley at the end of the sentence, but why was the statement made in the first place?  I for one would be very interested if you could break Simulator Z, and I'm sure there are others out there who would also be interested.  I should also add that my interest in such a test would be to show me where the simulator could be improved, as it is still a work in progress.  I'm not claiming that it's unbreakable, but if someone can break it, I'd like to know how they do it.

Quote
Newsflash, it isn't possible in a pretty vanilla game without the CAM to get a population to the kind of levels most of y'all consider routine.

I agree, having played large tiles in a vanilla game for far too long.  But it seems that a major purpose of this whole site is to develop content that goes beyond the vanilla game, which is why these issues are relevant.

QuoteIt is pretty pointless for me to run any tests using Z in any capacity. My own sim right now, I dropped the caps back to defaults because I couldn't get enough traffic.
I find the default simulator with better pathfinding and longer commute, cvs and intersection tweaks can handle things pretty well in this situation. I need to scale it back even further to maintain some challenge level.

I don't see how this follows.  You were saying earlier that you were getting a fair amount of abandonment with a PH of .003.  I would suggest testing Simulator Z Classic then.  It still has the Simulator Z core, but with the capacity levels of the original simulator.  I'm not suggesting that you use it as your simulator; you obviously want to build your own, and that's fine with me.  But there's no reason you can't do the tests to back up your statements.

Quote
I shortened the above. A very interesting test. Which of course pushes me even deeper into mere speculation.
Because the other day you "proved" that the max commute time was much more important than thought.
Today you just "proved" that it isn't.

Jason mentioned a similar point.  The tests do not provide contradictory results, however, my summary of them could have been clearer.  So let me summarize the two tests together in a way that's not contradictory:

The maximum commute time is the biggest factor by far in determining how fast the game runs.  However, when it comes to actual effects on game play, it has much less significance than previously thought; the value of the PH is much more significant in this case.

To me, that statement seems both internally consistent and consistent with both sets of test results.  Do you disagree?

Quote from: jplumbley on November 09, 2009, 10:03:15 AM
You can't make claims saying something is better and not proving it.  And I did not say "Prove it.", I was not an a-- about it like it seems you are trying to make it seem with this quote.

I was doing this from memory and made the mistake you point out.  I apologize.  To me, the statements said essentially the same thing, and I emphasized that I thought it was a reasonable request.

QuoteSo now, you have shown that it does exist as a problem.  But you have not shown anything that can provide an explanation of why it happens in one City but not another considering there are many people who have "healthy" cities using Simulator A.

I gave you an explanation in a recent post.  Do you disagree with that explanation?  If so, where?

From what I can tell, a majority of Simulator A users do not have this problem.  But I have heard from many who do, and for those who do have it, it can be serious.  As the very existence of this problem outside my cities is being called into question, perhaps it is time for me to gather together the quotes I had mentioned earlier.  This would at least have the effect of showing that my experience is shared by many others, something that I gather you don't believe at this point.  In some ways, the quotes are better than tests, because you don't know how often the test situations occur in actual games.  The quotes are from people who actually play the game and encounter this situation.

Quote from: Pharaon-Kheops on November 09, 2009, 01:36:59 PM
well, I do have abandonment because of commute time, in my CAM cities, even in the early stages (about 40k to 50k). This occurs not matter what simulator I use (both 3 at high capacity though, and even at "ultra" capacity in the case of sim z). Judging which sim has the "worse" is clearly too subjective at these pop levels. Of course, I did verify that there was no other problems like desirability, crime, demand, etc...

This definitely should not happen with Simulator Z if the city is built well.  What types of buildings are abandoning (i.e., R$, R$$, or R$$$)?  Could you please post a picture of your RCI Demand Graph, your Pop & Jobs graph, and your Traffic Congestion View minimap?  Also, how big is your abandonment problem?

QuoteThinking about all this discussion, a question popped up in my mind: "could this problem come from a conflict between simcity_1.dat and the simulator dat, as it is the case with the CAM (about workforce doubling)?"

No, the mechanisms in question are completely different.

EDIT:  I would be especially interested in knowing if you have enough jobs at the proper wealth levels for your Sims.  This wouldn't necessarily show up in the demand graph.

Pharaon-Kheops

#312
sadly, this was a test city and I have already wiped it out....

but I can give some more clues:

- as said, it was a test city. Large tile. Grid pattern, com zones evenly reparted along the main avenues. A large stripe of agri on the north border. Industrial zone in the SW corner.
About 56k residents 1000R$$ all the rest R$ (CAM with optional 5%). 30k ind jobs, 15k commercials. mayor rating above 60 (hospital and police coverage, no schools at all, FD onlyin the industrial zone). Water only in the Ind zone.

-Concerning traffic, some soft congestion (orange) at some intersections but nothing that could freeze totally an entire city.

-All R demand are positive, R$ quite at the top, R$$ slightly less. Comm and Ind demand low but positive.

-abandonment touch about 5 to 10% of the R$ houses (R$$ were too scarce to determine if they are also affected), evenly reparted all around the city.

Hope this little bit of data could help... I promise to shoot my next city from all the points of view^^

PS: to answer your edit, as you can see above, there where plenty of jobs quite double than those needed...
"to be is to do" - Kant
"to do is to be" - Sartre
"to be or not to be" - Shakespeare
"to be do be do" - Sinatra

z

Quote from: Pharaon-Kheops on November 09, 2009, 03:07:09 PM
Water only in the Ind zone.

This could certainly be part of the problem, though in an indirect way.

Did you by any chance install the Park & Ride versions of the simulators?  That's the most common reason for getting this type of problem.

jplumbley

#314
Quote from: z on November 07, 2009, 11:56:36 PM
I had realized that I hadn't fully answered this part of the question, and I was planning to address it more.  I think that what it comes down to is what you call playing style.  There are two aspects of your playing style that you have publicly described that would explain why you don't see this effect.  First, you tend to play smaller cities than the one I used in my tests, and based on the way A* works, this would make it exponentially less likely that you would see such an effect.  As I mentioned, I went through your whole testing thread, and the only place I saw evidence of this effect was in Nate's city.  But that was by far the largest city in your tests, with all other cities being under a million, as far as I could tell.  So there's a huge connection between the likelihood of this happening and the size of the city, which your tests appear to confirm.  Nate's city when we started was about the same size as my test city.

The second factor is that in your cities, you tend to mix different zone types closely together.  This not only makes paths between Sims and their jobs shorter, but it means that there tend to be a lot of short, valid paths between Sims and their jobs.  This makes the job of the pathfinder much easier.  Combine that with the fact that in smaller cities, there are less paths that need to be searched, and the result is that the pathfinder does not need such a precise setting to find correct paths reliably.

I saw this effect with the Maxis simulator when I first played the game.  On medium-sized tiles, it worked reasonably well for me.  But on large tiles, things really fell apart.  (Please note:  I am not trying to compare the Maxis simulator to Simulator A; I am just comparing this one effect.  Simulator A is obviously vastly superior to the Maxis simulator in just about every conceivable way.)

Is this the quote you are referring to?  Essentially, you have explained to me what I have been trying to make you realize since you made your claim.  It seems to me that play style has a very large effect on what is the proper PH for a person which would really make "Perfect" Pathfinding less relevant than you are making it out to be.  Actually, one of the things I have been noticing is things overall are not as important as we originally assessed. (which would change our job descriptions a bit to be more helpful to others in providing them with information and guidance rather than "build it" for them.)  It seems to me that the play style which is most effected by not having "perfect" PH is where you are building griddy cities in overly populated areas.

http://en.wikipedia.org/wiki/Manhattan

According to this info on Wikipedia of Manhattan Island, one of the densest areas in the world, there is a population density of roughly 28,000 people per square KM.  Which means on the 16 square KM large city tile the densest spot in the world would only have roughly 450,000 Sims.  In real life, Manhattan is a griddy like city with the worst traffic in the world and if you take a look at the zoning, you will see condos next to office towers mixed together throughout the island.  Manhattan would not work without the proximity of home and work, which would only make sense for our game to force you to "shorten" your distance between home and work when you reach these ungodly population levels.  The very thing you have been trying to avoid is the realistic "crunch" of population on commute distance and allow your users to build griddy and make big sections of Residential and Commercial away from each other.  It doesnt work in real life why should it here?

But, then that is my opinion.  You are entitled to yours on what realism is.

This would be why I have been saying Simulator Z is great as a Sandbox Simulator for MDs and why Simulator A is better for those who would like a little more challenge and game play.
"You learn something new everyday."

http://img517.imageshack.us/img517/169/nhpjplumbleykv3.gif
Bringing the new horizons closer to reality.

Berethor ♦ beskhu3epnm ♦ blade2k5 ♦ dmscopio jplumbley ♦ moganite ♦ M4346 ♦ Dedgren ♦ Ennedi Shadow Assassin ♦  Tarkus ♦ wouanagaine
Street Addon Mod - SAM

sumwonyuno

#315
I've been doing some tests in my Downtown city tile and the results of the tests were not what I was expecting.

Here is a picture of the entire city tile:


The Sim population is ~310,000, while in RL, the equivalent population is < 100,000.

There is a single east-west freeway (Maxis highway) running through the center of the city tile.  Generally, north of the freeway is housing, commecial is to the south of the freeway and industrial jobs are on the southern edge of the map.  This pattern is common among city tiles on the southern shore of my region.  There are suburbs several tiles away and commuters need to skip across city tiles to get to jobs.

Back when I had abandonment problems with this particular city tile, the problem area is the at the center right of the map north of the freeway, where all the high-rises residential are.  There are only a couple of roadways in and out of the area.  I was using Simulator A at the time.  I had thought it was really a traffic problem (commute time).  To solve the abandonment problems, I used Simulator Z and beefed up the number of jobs in the region (particularly, tens of thousands of industrial jobs elsewhere in the region).  The mass abandonment went away, and I gave credit to Simulator Z.  Now that I think about it, it's more likely the increase in jobs did the trick.

I stuck with Simulator Z.  The next issue was getting Sims to work where I prefer them to, without affecting both the visual and functional aspects of the transportation system I built.  Back with Simulator A, Sims were working in the Downtown city tile, but a good chunk of commuters were leaving the city, not working in the same tile they had homes in.  Especially in that high-rise area north of the freeway.  These commuters were going to work in the city tile to the east.  The pattern stayed true with Simulator Z.

There is a huge amount of westbound commuters from the eastern suburbs.  I was able to achieve this after getting the abandonment issue "solved" and using Simulator z.  However, there are a lot of jobs in the Waikiki city tile (east of the Downtown city tile).  These commuters were just skipping that tile, skipping the Downtown tile, skipping the city tile west of Downtown, and I didn't want them to go any further, because there is the potential of an eternal commuter loop by the airport.  In the Downtown city tile, there were thousands of empty jobs, not just in the industrial areas (like I always had), but in the Downtown commercial high-rises as well.  Sims weren't going to those jobs.  I had the thought that Sims were just going to the closest job, but those commercial jobs were closer (to the perspective of the westbound commuter from the east suburbs) compared to the Maxis highway neighbor connection for the city tile to the west.

The simulator used in the following is Simulator Z Classic.  The congestion view was full of red like I'd never seen it before.  Since the network capacities are so low, and buses add to traffic, pedestrian usage was ~ 3/4 the amount of car usage in the city tile, and bus usage was ~ 1/5% the car usage.  Some people do walk 1/2 a mile to work in real life, but not in the range of tens of thousands.

The first test had to do with getting (some) Sims to where I wanted them to work.   A while ago, z suggested to me (quite often) to put rapid transit in my region, as it could "solve" some weird commuting issues (40,000 Sims walking to an avenue connection).  Even with Ultra, thousands of Sims were walking to their jobs, for both intra- and inter-city commutes.  Once Honolulu officially starts building its elevated rapid transit line, Capitalis will have one too.  I had the idea of a secret subway system.  It wouldn't intrude on the non-underground view.  I wanted to see if Sims really went to the "closest, available and appropriate job".

I built several, parallel, express subway north-south corridors.  One station was in the residential areas to the north of the freeway, and the other station was along the southern shore, by the commercial high-rises and industrial areas.  To my surprise, Sims were finally commuting to the southern shore.  Not by the dozens, but by the hundreds.  For these residential areas, the neighbor connections were the closest jobs, but these neighbor connections were extremely overcongested (hitting the 65535 limit).  I connected these parallel subway lines, and the usage jumped to 8000+ on some parts of the lines.  There were some bus commuters from neighboring cities that were transferring, but the overwhelming majority of subway users were residents of the Downtown city tile.  I didn't save the city tile after the test.

Today, I had the urge to try Simulator A again.  So, I replaced the traffic simulator and congestion view plugins.  I put in Simulator A Hard.  To my surprise, things were working out fine, and better than expected.  The westbound city-tile-skipping commuters were actually taking exits off the freeway.  Sims were working at Downtown and (some) industrial areas.  There wasn't abandoment.  I do have to give Simulator Z credit, as it moves traffic away from streets and roads, and on to avenues and highways, and also spreads traffic to alternate routes.  With Simulator A, Sims prefered certain routes, and there were those dreaded shortest-distance paths, but that was because nearby roads were many times over their capacity.  And speaking of cars, car usage was higher compared to Simulator Z Classic, pedestrians were way down, closer to bus usage.  There wasn't the overwhelming bus usage I was expecting, even though buses did not count to traffic.

Now, I'm interested in trying Simulator A in other neighboring city tiles and seeing if the patterns hold.  z, jplumbey, ldog, etc. you all can figure out what this all means.


The City & County of Honolulu, a Mayor Diary based on Honolulu, Hawai'i.

mark's memory address - I've created a blog!

z

#316
Quote from: jplumbley on November 09, 2009, 09:31:54 PM
Is this the quote you are referring to?

Yes, that is the quote I was referring to.

Now there's nothing in Simulator Z that prevents you from playing cities in your style and in a challenging way.  Various efficiencies I've made in Simulator Z have gradually made it require less and less network capacity to function in a challenging way.  This is why I started out with only the Ultra capacity level in my pre-release versions, introduced Low, Medium, and High for the release version, and have recently release Simulator Z Classic, which has half the capacities of the lowest capacity version of Simulator A.  Are you claiming that Simulator A is more challenging than Simulator Z Classic?  If so, on what basis?  The last time I asked you, you had never used Simulator Z.  Have you now?  If so, to what extent?

So although the current versions of Simulator Z will allow a challenging game to be played in all ranges of cities (and I am in the process of preparing documentation from users to back that up), Simulator A restricts users to your playing style, and fails in big cities with big populations.  (I'll address the "realistic" aspect later, with some statistics.)  So Simulator A allows you to play a subset of the cities you can play with Simulator Z.  Specifically, it restricts you to smaller cities with simpler paths - that's what it means to be playable with a higher PH.  So from that point of view, I could call Simulator A a "sandbox simulator," since it restricts you to a much simpler game, and Simulator Z a "more realistic" simulator (which is how many users have characterized it).  But this seems rather pointless to me.  So please, let's just get rid of terms like "sandbox simulator," and stick to verifiable facts.  Along those lines, unless you can prove that the vast majority of usage of Simulator Z is for MDs, which is contrary to what I've seen, please drop that assertion as well.  Repeating assertions not backed by verifiable facts isn't going to get us anywhere.

QuoteAccording to this info on Wikipedia of Manhattan Island...

We had this exact discussion a year ago, didn't we?  After months of discussion and research, two things emerged:


  • The game contains a fairly reliable scale of speeds and distances, based on the fact that a single game square is 16m long.  (At very small scales, the game sacrifices proportion for the sake of visibility, but this does not significantly affect game play.)
  • Population in SC4 cities is not comparable to population in RL cities.

The second point is supported by measurements made by David while building 3RR.  In a post in this thread made last December 1st, I said,

QuoteHowever, at one point dedgren concludes (using Chicago as a model, no less!) that population density in SC4 is 10 to 20 times too high; in another example, he concludes it's 10 times too high.  This corresponds with my experience exactly.  Having built a half dozen tiles of Chicago exactly to scale, right down to the streets, with correct zoning, I estimate that if I ever finish it, the total population will be about 30 million - about 10 times too high.  So taking this into account, the aforementioned model of Manhattan-style transport would start applying to SC4 cities somewhere in the 4.3 million to 8.6 million range.

You say here:

QuoteThe very thing you have been trying to avoid is the realistic "crunch" of population on commute distance and allow your users to build griddy and make big sections of Residential and Commercial away from each other.

And in an earlier post:

Quote from: jplumbley on October 31, 2009, 09:47:47 AM
What is happening in the downtown cores around the world which used to be reserved for commercial towers?
At least in Toronto anyways and I am sure in other parts of the world, there are more and more condos (residential units) being built right in the middle of the downtown core.  The initiative is there for the population to live closer to work instead of commute from the "bedroom" communities so that we can reduce traffic, promote walking or MT usage.

The implication I read here is that commute times either have gone down in Toronto or are expected to go down, and that the same is true ("I am sure") in other parts of the world.  No supporting figures are given.

Let's look at the actual figures.

In the issue of Time Magazine dated November 9th, at the bottom of page 11, there is the statement:  "15 minutes [is] the mean amount of time it takes people in Grand Forks, N.D. to get to work - the shortest commute of any U.S. metro area."  That's a one-way commute, so the mean round-trip commute is 30 minutes.  But in SC4, we talk about maximum commutes.  Now the maximum commute for an area is typically much more than twice the mean commute, but we'll be very conservative, and we'll use a figure of twice the mean commute.  This means that the shortest maximum commute of any U.S. metropolitan area is 60 minutes.  And of course, the average maximum commute in the U.S. is much more than that.  Compare this with the 17 minute maximum commute time of Simulator A.

Let's go specifically to Toronto.  You're Canadian; you seem to know Toronto.  But how do the actual figures stack up?  Here are the average round trip commute times for major Canadian cities, courtesy of the Government of Canada:


So the average round-trip commute time for Toronto commuters is 80 minutes, meaning that the maximum round-trip commute is somewhere over 160 minutes - about a factor of 10 higher than Simulator A's maximum.  Are those condos bringing that number down?  Or are they going to soon?  Here are excerpts from an article from the Toronto Star which use the same figures:

QuoteHe's among the growing ranks of extreme commuters in the Toronto area whose daily round trip to work routinely stretches to two hours or more. GTA residents have the longest commutes in the country – up from 68 minutes on average in 1992 to 79 minutes in 2005, according to Statistics Canada.

Experts predict the daily drive will likely get worse before it gets better as the GTA's population grows.

...

He's among the growing ranks of extreme commuters in the Toronto area whose daily round trip to work routinely stretches to two hours or more. GTA residents have the longest commutes in the country – up from 68 minutes on average in 1992 to 79 minutes in 2005, according to Statistics Canada.

Experts predict the daily drive will likely get worse before it gets better as the GTA's population grows.

Every weekday at 5 a.m., just about the time Barrett is arriving at his desk, Jennifer Case is pulling out of her Collingwood driveway with a huge travel mug for the 90-minute trip to Aurora, where she teaches three days a week.

But if she leaves any later than 5 a.m., Case runs into traffic. "Highway 400 is very well travelled by 6:15 a.m. By the time you get to Barrie it is very busy," she says.

Like most long-distance commuters, both Barrett and Case drive alone. Statistics show the farther the commute, the more likely it will be done in a car. Even transit advocates admit the alternative – expanding transit to more remote residential areas – could be contributing to the sprawl it serves.

"There's always a pressure to provide access (to transit) further out of the city. And it's a bit of a chicken-and-egg situation," says Michael Roschlau, president of the Canadian Urban Transit Association.

"It's making it easier to commute longer distances and encourages people to live further away from where they work. Transit is in a sense part of the problem," he said.

But transit isn't even a realistic option for the growing ranks of commuters who, like Barrett and Case, skirt or travel right through the city. These are the commuters who, along with trucking, have turned rush hour into an all-day affair.

Barrett, who's also a trustee of the Durham public school board, used to start work at 6:30 a.m. "Then I moved it to 6 a.m. and then I moved it to 5 a.m. I will not waste my time on the 401," he said. "If I leave the house at 6 a.m. it will take me an hour and a half. If I leave at 7 a.m., I likely won't see Mississauga for between an hour-and-three-quarters and two hours."


By the numbers

In the 1960s a typical suburban household would make 20 per cent of its trips by transit. By 2007 that had sdropped to 10 per cent.

Kilometres of carpool lanes on provincial roads in the Toronto region this year: 39

Kilometres of carpool lanes projected for 2031: 400

Number of car occupants on average in Toronto in the 1970s: 1.25

Number of car occupants in Toronto now: 1.1

Percentage of trips taken on transit in Toronto: 35 per cent

Percentage of trips taken on transit in York Region: 9 per cent

Percentage of Toronto-area residents who say traffic congestion is a severe problem: 41 per cent

Average Toronto-area round trip commute time in 2005: 79 minutes, up from 68 minutes in 1992

Proportion of Toronto-area workers with a round trip commute of an hour or more: 66 per cent

These are official government figures.  And the trend is just getting worse.  That's the real Toronto; the small number of condos built in the city center isn't making a difference, nor is it expected to in the near future.

I've just seen that sumwonyuno has posted.  At first glance, the results seem somewhat anomalous, but it will take time to analyze them.  Question for sumwonyuno:  Are you using Simulator Z v1.2?

sumwonyuno

#317
I had used Simulator Z Ultra with the latest NAM, then to Simulator Z Ultra v1.2, then I switched to Classic.


The City & County of Honolulu, a Mayor Diary based on Honolulu, Hawai'i.

mark's memory address - I've created a blog!

Pharaon-Kheops

Quote from: z on November 09, 2009, 06:05:22 PM
This could certainly be part of the problem, though in an indirect way.
hum... you'll have to be more factual on this point^^ how could having no water in R$ zones cause abandonment for commute time???

I do not use the P&R.

@jplumbey: your point about manhattan and the role of job proximity is right, but only to a certain extend as manhattan has far more jobs than inhabitants... the quality (I mean design cleverness, not quality in the sense of beautiful and clean^^) of the traffic network is a big factor here... but, indeed, should manhattan shelter only jobs, and every comuters come from the other NY bouroughs, this island would be what ressemble the most to hell  ;)
"to be is to do" - Kant
"to do is to be" - Sartre
"to be or not to be" - Shakespeare
"to be do be do" - Sinatra

z

#319
Quote from: sumwonyuno on November 09, 2009, 11:35:28 PM
I had used Simulator Z Ultra with the latest NAM, then to Simulator Z Ultra v1.2, then I switched to Classic.

What you reported in your main post is definitely a bug.  It's either a bug in Simulator Z or the game itself.  But since Simulator A can work properly in your city, there's no reason Simulator Z can't as well.  I can't reproduce the bug here, so I'd really appreciate your help in tracking it down.

I have one good idea about what the bug may be.  Since your city works better with Simulator A (Hard), I've created a special version of Simulator Z (Low) (which has similar capacities) which is modified in a way that may fix this bug.  I've attached this version of the simulator to this post.  I would appreciate it greatly if you would try it out in your main city and see how it works.  You should be able to tell after a few years, at most.


Quote from: Pharaon-Kheops on November 09, 2009, 11:35:42 PM
hum... you'll have to be more factual on this point^^ how could having no water in R$ zones cause abandonment for commute time???

As you know, lack of water severely limits the growth stages for all RCI types.  Although it does so symmetrically, this limit means that the game has much less room to play around with to adjust to imbalances in workers and jobs.  In this case, for example, if you ended up with too many workers in your residences, instead of your commercial buildings growing from Stage 3 to Stage 4, they'd just be stuck at Stage 3, and some of your workers wouldn't be able to find jobs.  If this affected enough of them, you'd get abandonment due to commute time.