• 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 1 Guest are viewing this topic.

z

Traffic simulator aficionados, rejoice!  &dance &dance &dance ()flower() ()flower() ()flower() &hlp &hlp &hlp  Certain secrets of the Maxis traffic simulator, last heard in the hallowed but long-deserted Halls of Maxis, have been rediscovered!  :o "$Deal"$ :o  Formerly known only to a select group of Maxis programmers  ()meeting(), these secrets are about to see the light of day.  :sunny:


It all started when...  (You know it's going to be a long post when it begins "It all started when..."  But knowledge comes at a price!  ::)  Patience, dear readers, patience...  :sleeping:)

Anyway, it all started when Jason challenged me to a duel.  ?$%kar&%h  (No, that can't be right.  ???  What was it?  ()what()  Oh yes!  &idea) to prove an assertion I had made based on experiments I performed when I first built Simulator Z.  I had found that increasing the pathfinding heuristic above .003 would produce abandonment in various situations.  Jason's response was:

Quote from: jplumbley on October 27, 2009, 09:12:48 PM
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 [pathfinding heuristic] 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.

So I did exactly what Jason had asked.  I took Simulator Z, raised the PH to .009 and showed him abandonment where none had existed before, and I did it all publicly in this long post.  However, in a later post, Jason stated the following:

QuoteIt would be interesting to see how this one change in the PH effects other play styles.  There is a big question of why do I not get abandonment with a PH of .009 but you do?  Maybe there was an anomoly in your experiment?  I simply have not seen anything like that happen in my cities other then when I was using the MAXIS Default Simulator.

I am not trying to challenge your test, but I think it needs more testing.

OK.  I'm sure that there are other skeptics out there besides Jason.  So I designed a number of new, different tests.  And as they say, "Around every silver lining, there's a dark cloud." ()testing()   "Inside every dark cloud, there's a silver lining." :)  And in this case, the silver lining was unexpected revelations about the inner workings of the traffic simulator.

Disclaimer:  These tests were performed only at Jason's request, and were merely designed to provide further evidence for previously made assertions about properties of the traffic simulator in general.  They are not designed, nor is it claimed that they show, anything in particular about any particular simulator; specifically, they are not claimed to show the fitness of any simulator for any particular purpose.  They do not claim to show whether a simulator is "good," "bad," "realistic," or a "sandbox simulator."  These tests merely illustrate the function of particular parameters in SC4 traffic simulators in general.  Actual simulators were used because this helped illustrate the effect of these parameters most effectively.  No Sims were harmed in the running of these tests.

OK, now that we've gotten the legalese out of the way, let's move on to the tests.

We start once again with the Near South Side of Chicago.  Jason raised the question of playing style.  Since all of my cities are arguably built in my playing style, that's the only style I can test directly.  However, there are other factors that come into play here.  First of all, as I mentioned earlier, this is not really "my" style, but is a reconstruction of an actual city, built and zoned to scale.  This is not an infrequent way of playing, and anyone can duplicate tests such as these using large cities on large tiles.  Secondly, if the tests can successfully illustrate principles of traffic simulator operation (which is their goal), then it really doesn't matter what playing style is used, or even if it is actually used by anyone at all.  For even in the latter case, the principle that has been discovered or proven can be applied to understand the workings of any city built with any style of play.

[Update:  Since I began writing this post several days ago, ldog (Lenny) has published confirming experiments of his own design with a custom-built simulator in his thread.  This would seem to address both the playing style and anomaly issues.  Jason has accepted this, but quite reasonably, still wants to know why he hasn't seen these effects with Simulator A.  In ldog's thread, he asks ldog, "Is it possible for you to test a City with Simulator A and show this effect happening?"  This post addresses that issue; it's actually somewhat coincidental that I had already done the following tests.]

We would like to make these tests as different as possible from the first set we did with Simulator Z, but designed in a way that they end up testing the same properties, most notably the Pathfinding Heuristic (PH).  So for all of the tests in this post, we will use variations of Simulator A (Hard).  First of all, this seems very appropriate, since this whole issue arose over the question of what function the PH played in the workings of Simulator A.  Secondly, reproducing the same effects in both Simulators A and Z would be further proof that the effect shown in the previous tests was not just some anomaly, as hypothesized by Jason.

But some may ask, Is it fair to use the Hard version of Simulator A on a city with a population of two million?  After all, the NAM documentation recommends that the Hard version be used only on cities up to about 250,000 Sims, while the Easy version is recommended for cities less than half the size of the Near South Side.  The answer is yes, it's fair, because we are not expecting the simulator to perform properly.  We are deliberately testing it out of spec to see if we can find evidence of problems that might only show up occasionally, or that might even show up frequently, but might be missed due to other factors.  If we don't test things out of spec, we run the risk of ending up in a position similar to that of the engineers who built the first Tacoma Narrows Bridge.

We start with an unmodified version of Simulator A (Hard), and we use the same starting point as before for the Near South Side - January, 433.  We run the game for 12 years and take a look:



I'm sure that the results here are not unexpected.  What's interesting is that if you compare this picture with the second picture in the previous test post, which showed Simulator Z modified to run with a PH of .009, you see that virtually all the buildings that were abandoned in that picture are abandoned in this picture as well.  There are also additional buildings that are abandoned in this picture as well.  There seem to be three main possibilities for the abandoned buildings in the current picture:


  • The buildings that are abandoned in both pictures are abandoned due to the higher PH, while the buildings abandoned only in this picture are due solely to the combined effects of congestion and the one-way commute time's being limited to 8.5 minutes.
  • All the abandoned buildings in this picture were abandoned due to the higher PH.
  • All the abandoned buildings in this picture were abandoned solely due to the combined effects of congestion and the limited maximum commute time.  The fact that many of the same buildings are abandoned is just a coincidence.

This picture alone does not provide enough data to choose between these three options.  Meanwhile, I've noticed that the speed of the game running this city on my computer under Simulator A is the same that it's been in previous tests - three game years per hour in Cheetah mode, which is half of the six game years per hour that I've found running under Simulator Z.  That's the game as a whole; the actual Simulator A running time seems to be about four times as long as Simulator Z.  In a later test, for the first time we'll find out exactly what's responsible for that large difference in speed.

Now we run the game for an additional 13 years, making 25 years in total, which means that this single test has taken more than eight hours.  The abandonment is essentially the same, so there's no need to post another city picture.  But now we have enough data for some interesting graphs:


Again, we have the same pattern of rising population that we had in the first testing post.  That turned out not to be a good thing there.  What about now?


Sure enough, we have almost the exact same diverging population lines as in the first testing post.  Once again, high-wealth Sims are deserting their residences (presumably fleeing to the suburbs), while their former residences are being occupied by a larger number of low-wealth Sims.  In the previous test, this change appeared to be linked solely to the change in the PH, as that was the only change made to the simulator.  So it's beginning to look like the PH does at least play a factor here.

Now some skeptics may say, "But the PH doesn't behave like that!  At least not in Simulator A.  Are you sure you have Simulator A installed properly?  Are you sure that there's not one of your many versions of Simulator Z hanging around, overriding it?"  Well, it's easy enough to show that this is indeed Simulator A.  Here is the Traffic Volume Graph:


The left side is Simulator Z; the right side is Simulator A.  There are some very obvious differences.  Car usage is higher is Simulator Z, simply because that's the way the simulator is set; still, not counting pedestrian traffic, car traffic comprises only about 20% of total traffic in this city under Simulator Z, compared to 11% for Simulator A.  Bus traffic is almost nonexistent under Simulator Z simply because this city was built with a much older simulator, and subway and bus lines are almost completely congruent.  In Simulator A, bus traffic exceeds car traffic; as we will see in later examples, the bus traffic in this city in Simulator A is mostly due to the fact that buses in that simulator do not contribute to traffic or congestion.

Speaking of congestion, here's the congestion minimap for the unmodified Simulator A:


Congestion is pretty bad, but then it should be; this simulator is operating way out of spec.

That completes the testing of the unmodified version of Simulator A.  There definitely seemed to be some indication that the PH is involved in both abandonment and downgrading, but how much?  And can we be sure?  Yes, we can, but to see why, and to answer the first question, we need to do different tests.

One possible explanation for some of the effects that we have observed is congestion and commute time limitation.  Of course, congestion can cause such effects only if there is a commute time limitation.  So for the next experiment, we raise the maximum commute time of Simulator A to the level of Simulator Z, which is effectively infinite.  With this modified simulator, no abandonment or downgrading effects can be attributed to commute time limitations, because there are none.  So if the commute time limitations were responsible for the abandonment and downgrading, we should now not see either of those two.  Making the commute time unlimited should give us a "sandbox simulator," right?  We'll soon see.

Capacities of Simulator A are similar to those of Simulator Z (Low), which has no abandonment or downgrading problems in this city.  However, Simulator Z's rail capacities are somewhat higher.  So that we can completely rule out different capacities as being the cause of different simulator behaviors, we give our modified version of Simulator A (Hard) the capacities of Simulator Z (Low).  Now we restart our city in January, 433 again, and run it for 7 years.  Here is the result:



This is interesting.  Even though the maximum commute time is now unlimited, abandonment is almost as bad as in the previous test.  It doesn't look like making the maximum commute time unlimited creates a "sandbox simulator."  Those really dark buildings are all truly abandoned due to commute time, too.  As in the previous testing post, I checked them all out.  Here's just one example:



How can they be abandoned due to commute time, if commute time is unlimited?  There are only two ways this can happen.  If there aren't enough jobs for these Sims, they'll abandon their residences and leave town.  But we saw from the picture at the beginning of the last testing post that the city was in perfect health under Simulator Z.  And then there are all those graphs, too.  So there are plenty of jobs in this city.

There's only one other possible explanation:  The pathfinding heuristic is too high.  The jobs are there, but the Sims can't find valid paths to them.  It looks like the PH is playing a very big role in abandonment here, being responsible for at least most of it.

Meanwhile, I mentioned earlier that one of the tests would show what was responsible for the large difference in speed between Simulators A and Z - basically a factor of four in this city.  I had suspected that it had something to do with the Congestion vs. Speed Curve, as the implementation of this curve has a direct effect on the speed of the A* algorithm.  But although this is true, it turns out that there is a factor that has an effect many times greater on the speed of the simulator.  It's the maximum commute time.

Now if you think about it, you might thin, "That's reasonable - with a higher maximum commute time, the pathfinder can explore more paths, which means that there's more work for it to do, which means that it will run slower."  But it turns out to work just the opposite way:  Other things being equal, then the higher the maximum commute time, the faster the simulator runs.  This effect dwarfs all other effects on simulator speed.  Of course, there comes a point where raising the commute time will no longer raise the simulator speed; where this point is varies by city.

I discovered this when I ran the current test.  Whereas the game had always run at about three years per hour in this city before when using Simulator A, now it was running at a bit more than six years per hour, which is a bit faster than using Simulator Z.  (This makes sense because once the maximum commute times have been equalized, Simulator A's higher PH will cause it to run slightly faster than Simulator Z.)  This means that Simulator A itself was running about five times as fast as it had before I raised the maximum commute time.  This has since been verified repeatedly.

At first I thought that this made no sense at all.  Then I realized what was happening.  With a low maximum commute time, the simulator was having to try extra hard to get the Sims to their jobs before running out of commute time.  With a high maximum commute time, the simulator was freer to pick the first valid route that came along that met the PH criteria.  This also leads to a wider range of routes.  So with a higher maximum commute time, there was actually less work for the simulator to do.  Apparently, a lot less.

I remember when I installed the NAM for the very first time and switched from the Maxis simulator to one of the NAM simulators on a large tile.  There were many benefits, of course, and among them was that my game started running faster.  I didn't understand much about how anything worked at that point, so I didn't think about it a lot then.  But now I see why it happened.

So what's the benefit of a low commute time?  I can see none.  If commute time is going to be great enough so that abandonment does not occur, then you might as well raise it way up and make the game run faster.  This may decrease congestion, but if you want more congestion, the best way to get that is to lower network capacities, as doing so has no unexpected side effects.

So that's what the city looks like after seven years.  Was that enough time for the full effects of the simulator switch to manifest?  We'll run the game for another 23 years, for a total of 30 years for this test, to find out:



Things look pretty much the same.  Some abandonment in the smaller buildings in the upper right of the picture has cleared up, but one of the big residential towers has gone from completely healthy in the previous picture to abandoned now.  This looks like a typical abandon/reoccupy cycle.

What about the rest of the city?  Here's another neighborhood:



Notice that one instance of the Comfort Arms is selected.  Most instances of this building in this picture are dark gray, meaning downgraded, while a few are black (along with some other buildings), meaning abandoned.  The downgrading is new, as we'll see in the following graph.  It's part of the whole abandon/reoccupy cycle going on in this city.  Here's the Pop & Jobs graph:


Once again, we have the diverging lines, with the same meaning as before.  And what they show certainly corresponds with the picture above.  It looks like raising the maximum commute time didn't buy us much in this city, at least in terms of abandonment and downgrading.

Finally, let's take a look at the Traffic Volume Graph:


This looks basically the same as the Traffic Volume Graph from the previous test, with one exception:  Bus traffic is now less than car traffic, instead of greater.  This is a direct result of increasing the commute time.  With a longer commute time, the pathfinder is able to use less congested routes for the Sims, such as subways, and there is less necessity for putting them on buses, which effectively have infinite capacity in Simulator A.

So if making the maximum commute time unlimited doesn't solve the abandonment problem, what's causing it?  None of the abandonment and downgrading we have seen in this set of pictures has come from commute time limitations, since there are none.  Is there a single parameter in the traffic simulator that actually is causing this problem?

Not surprisingly, the answer is "yes."  It's the same parameter that caused all the abandonment in the modified version of Simulator Z in the previous testing post:  the Pathfinding Heuristic.  How do we prove this?  It would seem that the easiest way would be to start with an unmodified copy of Simulator A and lower the Pathfinding Heuristic from .009 to .003, leaving absolutely everything else unchanged.  So that's what we do.  We start at the same point, and we take a look around after 15 years:



I found this to be a very surprising picture.  In the previous testing post, I had shown that raising the pathfinding heuristic in Simulator Z resulted in abandonment where there had been none before.  This was not surprising, since Simulator Z has an effectively unlimited maximum commute time.  But Simulator A's maximum commute time of 8.5 minutes (one way) has been kept intact here, and there are only two abandoned buildings:  the Comfort Arms on the upper edge of the park, and the tall building near the top of the picture.  (The Al  Attar building lower down looks dark, but it has merely been downgraded to low wealth.)  Compare that with the first picture in this post, where there are about a dozen abandoned buildings.  The only difference in the simulators used is in the value of the pathfinding heuristic.

The above picture is after 15 years.  What happens if we run longer?  Does the abandonment get worse, better, or stay the same?  We run for 10 more years, for a total of 25 years, and take another look around:



Another surprising picture:  The abandonment is gone completely.  Again, this is just standard Simulator A (Hard), with the only change being that the PH has been reduced from .009 to .003.  Although this simulator was designed for cities with a maximum population of about 250,000, it seems to be doing quite fine on a city eight times that size, once we reduce the PH.

What about the graphs?  We would expect these to look quite different from the graphs in the previous two tests, although we would not expect them to look identical to Simulator Z's graphs.  The simulators are still different in other ways.

First we start with the cumulative RCI population graph:


As before, there's a rise in population in the middle of the graph when we switch from Simulator Z to Simulator A, but this time it's much less pronounced.  (The rise is preceded by a brief dip, which is where the temporary abandonment occurs.  As a population rise in the previous tests always had negative implications, this would seem to be a good sign.  For details, we look at the Pop & Jobs Graph:


Sure enough, the wide divergence between the R$ and R$$$ populations has disappeared.  In both previous tests in this thread, and in the test of the degraded Simulator Z in the previous thread, the divergence was always present.  So it's been present in the three tests where the PH was .009, and absent in the two tests where the PH was .003.  Based on these five tests, it certainly appears that the higher PH not only causes more abandonment, but also results in the city's being less attractive to high-wealth Sims and more attractive to low-wealth Sims.

Looking at the graph more closely, we can see a little bit more of what's been going on.  There was an initial drop off in high wealth Sims, which was presumably accompanied by abandonment.  The most likely cause for this is the reduced maximum commute time; routes that worked before were suddenly too long.  But the pathfinder is much smarter now, and over the next 25 years, it gradually found paths for everybody, so that at the end of this period, the population of high-wealth Sims is the same as before the drop.  There's just a tiny drop right at the end of this period; this is most likely not the beginning of a reverse trend, but simply a small fluctuation indicating that the rise has leveled off.

So there's essentially no net change in the R$$$ population.  However, during this period, the R$ population rises, while the R$$ declines, though by a smaller amount.  This change is due to other differences between the two simulators.  Based on previous experience and also on checking out various buildings, what's happening here is that there has been some downgrading of buildings from R$$ to R$.  However, this is unrelated to the PH, since the PH is identical on both sides of the graph.

Let's take a quick look at the Traffic Volume Graph:


It looks very similar to the previous one.  So traffic volume patterns really haven't changed significantly here.  This seems quite natural, until you remember that the previous graph was from the experiment where Simulator A had been modified to have an unlimited maximum commute time, whereas in this experiment it has the standard 8.5 minutes (one way) commute time.  In both cases, bus usage (which is used here as an escape valve for too much congestion) dropped.  This is more evidence that dropping the PH is similar in some ways to extending the commute time.  This is also one reason Nate's city worked fine with the CAM traffic simulator, which has a very large PH of .0465 (more than five times that of Simulator A); the large PH was greatly offset by a large commute time of 30 minutes one-way.

Here's a congestion minimap of the modified Simulator A:


There's still plenty of congestion, even though there's a low PH, no abandonment and no loss of R$$$ population.  Compare the above graph with the congestion graph of the unmodified Simulator A, which was displayed earlier in this post, but is reproduced directly below for easy comparison:


As you can see, there's somewhat more congestion here.  The lesser congestion in the first picture is a result of the smarter pathfinder rearranging the routes of the Sims so that they all can get to their jobs within 8.5 minutes.  Another way of putting this is that a lower PH makes congestion more of a motivating factor for the pathfinder in finding faster routes for Sims.

Finally, for comparison, here's the congestion minimap of the recently introduced Simulator Z Classic:


It looks extremely similar to the congestion minimap of the unmodified Simulator A directly above, although if you look closely, you can see that there's a bit more congestion.  I introduced Simulator Z Classic with a (hopefully) somewhat amusing post, but as you can see, it's no joke.  Its network capacities are all no more than half that of the version of Simulator A whose congestion minimap is displayed directly above.  This means that congested networks with similar congestion levels in Simulator Z Classic are carrying about half the traffic as their counterparts in Simulator A (Hard), at least in this city.

Does the similarity in congestion minimaps mean that Simulator Z Classic is roughly equivalent to Simulator A (Hard)?  Not at all.  It simply means what you can see:  In this city, their congestion minimaps are very similar.  In other cities, that may or may not be the case.  What I think this does show is that the combined power of a low PH and an unlimited commute time can greatly reduce the need for network capacity.  At the same time, Simulator Z Classic is like all other versions of Simulator Z in that abandonment due to commute time is essentially nonexistent, as is downgrading.  This last feature, of course, is dependent on many other factors besides the traffic simulator; the Near South Side has been designed and maintained to attract as many high-wealth Sims as possible.  (Disclaimer:  I am not a particularly noteworthy game player, so there is undoubtedly a lot more that could have been done in this area.)

At the beginning of this post, I promised that I would reveal new secrets about the inner workings of the traffic simulator engine.  I'll summarize what I have shown here:


  • Other things being equal, then the higher the maximum commute time, the faster the simulator runs.  This effect dwarfs all other effects on simulator speed.
  • Using a lower commute time narrows the range of routes used by the Sims.
  • If the maximum commute time is at least that of Simulator A (8.5 minutes one way), there are sufficient jobs available, and the Pathfinding Heuristic is higher than .003, then in the vast majority of cities, all abandonment due to commute time will be caused by the PH, and none will be due to commute time limitations.  In the small minority cities where this is not true, the vast majority of abandonment due to commute time will still be due to the PH.
  • A higher PH not only causes more abandonment, but also results in the city's being less attractive to high-wealth Sims and more attractive to low-wealth Sims.

So I think I have demonstrated that the same effects I have shown happening in Simulator Z can also happen in Simulator A.  But obviously, they don't happen all the time, as Jason has pointed out.  So now it's time to answer his question:  Why hasn't he seen these effects?  I have also gone through the entire testing thread he referenced, and certainly, I think that the general feeling any reader of that thread would get would be that this simulator was working quite well, was a big improvement over previous simulators, and had no major problems.  I think one summary Jason posted is also relevant:

QuoteTelltale signs that the Simulator is unbalanced:
1.  Abandonment when there is sufficient demand and jobs available.
2.  Sims taking obscure paths.
3.  Possible Abandonment/Repopulation cycle. (can be caused by other things aswell, such as poorly designed network structure, other Simulator issues, etc.)
There are others aswell but these are big ones.

I would agree with this completely, and in fact, this is what I have concentrated on showing in this post.  Why didn't Jason see any of this?

There are two parts to this answer.  The first part is that the cities tested tended to be on the small side.  As I mentioned earlier, performance differences in the simulator tend to manifest exponentially with city size.  Mott made mention of the exponential nature of A* in October, 2007, months before the tests began.  Yet from what I can tell, all of the test cities except one had a population less than a million, many of them much less, which made them a size in which it was far less likely for them to run into problems with a PH of .009.  This is especially true since many of these cities were originally built with a traffic simulator with a much larger PH; in such cases, using Simulator A with its PH of .009 would cause abandonment due to commute time either to decrease markedly or disappear entirely.

The second part of this answer concerns the city with the biggest population - Nate's, which started out with a population of over two million.  Although as Jason has pointed out, the congestion data view is unreliable, I have found enough data in Nate's posts to be able to show that either all or virtually all of his abandonment was caused by the PH.  At the same time, showing this unveils the presence of the Congestion Data View Bug, not only the one in the congestion data view, but also the related one in Simulator A itself (which has since been fixed).  However, showing this will take a separate post, as I need a break now (as do my readers, assuming there are any left  ::)).

Jason, do you have a record of exactly what capacity numbers were used in your test simulator?  I saw some of the numbers mentioned, and in one place you mentioned that the simulator had capacities close to what is now the Medium version of Simulator A.  However, if I had the exact capacities, I could be more precise in my explanation.

I hope that I have provided convincing evidence of what I have found; any questions are welcome.  The follow-up post concerning Nate's city will come as soon as possible.

Shadow Assassin

Interesting post, though... I still have a question: what affects how frequently the game "refreshes" its paths for the entire city? Say you bulldoze a road, and you get the "flying car" bug using the path query tool (and in the congestion data view it still shows up)... how long does it take, usually, before the game refreshes itself and corrects all the paths for its Sims when they travel?

Does the PH affect this, or is it something else within the simulator, or even as a direct result of computer speed?
New Horizons Productions
Berethor ♦ beskhu3epnm ♦ blade2k5 ♦ dedgren ♦ dmscopio ♦ Ennedi
emilin ♦ Heblem ♦ jplumbley ♦ moganite ♦ M4346 ♦ papab2000
Shadow Assassin ♦ Tarkus ♦ wouanagaine
See my uploads on the LEX!

xxdita

One dumb question, which I think needs to be asked.
Can a building become abandoned due to it's desirability factors changing? And if so, is that abandonment still defined as "Abandoned Due To Commute Time"? I only ask because I can't recall ever seeing an abandoned building with "Abandoned Due To Pollution", or any other reason aside from commute time. Perhaps all abandonment is blamed on commute times?
Taking a look at your pics, I can see you're using CAM, which has stronger desirability requirements in place, to prevent abandonment, but even those aren't foolproof. Overly congested transit networks deminish an area's desirability, which would cause much of the abandonment issues you've shown.
Not to discount your tests by any means, I just think we still don't have enough information. Perhaps get pics of the area in question's desirability before & say 10 years after switching traffic sims? Especially R$$$.

RippleJet

This was discussed in Lenny's thread not so long ago. :)


Quote from: RippleJet on November 04, 2009, 07:39:23 AM
There are only five of them in the game:


  • Abandoned due to lack of power
  • Abandoned due to lack of water
  • Abandoned due to low desirability
  • Abandoned due to commute time
  • Abandoned due to low demand

The third one is the one that you would get due to low customers in a commercial building, or due to any other desirability factor being too low (all in all, for any RCI type due to the desirability falling below the abandonment threshold).

The fourth one is the only one that you'd normally get in residentials, unless you cut off power or water.

The fifth one is the one you'd get if the overall demand for that RCI type is negative, and that can occur for all RCI types.

b22rian

 A real eye opener of a post here Steve  :o

Thanks for all the time and efforts you put into this highly informative tests of the traffic sims..
Back a long time ago, when I helped you run all those comparison tests of the different traffic sims.
i often wondered later what the true reasons were for sim Z running faster than the other despite
its more precise PH setting than the others..  :-\ i really appreciate you solving that little mystery for me  &apls

I think your post here is obviously going to generate a lot of further discussion about the traffic sims as you
have presented many new concepts which we did nt have a handle on previous to your testing here..
I greatly look forward to those discussions in your thread here..

your good friend, Brian

z

Quote from: Shadow Assassin on November 07, 2009, 04:15:15 AM
Interesting post, though... I still have a question: what affects how frequently the game "refreshes" its paths for the entire city? Say you bulldoze a road, and you get the "flying car" bug using the path query tool (and in the congestion data view it still shows up)... how long does it take, usually, before the game refreshes itself and corrects all the paths for its Sims when they travel?

Does the PH affect this, or is it something else within the simulator, or even as a direct result of computer speed?

What you're really asking is, "How often does the traffic simulator run?"  The answer is about once every four months, more or less.  The exact frequency is somewhat dependent on city population, but nothing else.

There are some exceptions here.  For example, when you first load a game, it may be well over a year before the traffic simulator first runs, but then it will resume its normal schedule.  That's the only exception I know of that affects your case.  Other exceptions may affect when certain parts of the traffic simulator run.

Quote from: xxdita on November 07, 2009, 04:26:41 AM
One dumb question, which I think needs to be asked.
Can a building become abandoned due to it's desirability factors changing?

Possibly, to the extent that desirability affected demand.

QuoteAnd if so, is that abandonment still defined as "Abandoned Due To Commute Time"? I only ask because I can't recall ever seeing an abandoned building with "Abandoned Due To Pollution", or any other reason aside from commute time. Perhaps all abandonment is blamed on commute times?

Definitely not.  There's a whole category of abandonment due to lack of demand.  There are only three ways you can get abandonment due to commute time:  1) There are no valid jobs for your Sims to reach.  This is fairly rare.  2) Your Sims are unable to reach valid jobs within the maximum commute time limit.  3) Due to the setting of the PH, the Sims are unable to find valid paths to their jobs.

QuoteNot to discount your tests by any means, I just think we still don't have enough information. Perhaps get pics of the area in question's desirability before & say 10 years after switching traffic sims? Especially R$$$.

Assuming what I say is true about abandonment, do you still think we still don't have enough information?

I see that Tage has posted as well, confirming what I said...

xxdita

Yeah, I was in the process of modifying my earlier post... after talking with Tage.
I'll read over your tests again after a good day's sleep...
But I already know that the only way for me to be truly convinced is to test it myself, in order to analyze every little question that pops up in my mind, in real time. So... looks like I have a new city to build, as soon as I wrap up my current project...

ldog

 &apls
Wow. I really don't even have much of anything to say for a change :P
Great job testing and documenting (I think this gets me out of doing the last test Jason asked me for :D )
:thumbsup:

sumwonyuno

#288
Wow, z, that's the most animated intro I've seen of a post of yours.  :P

Good job on the investigating!   :thumbsup:  I do agree that higher commute time = more paths available (if any).  The initial gut feeling is that will lead to more CPU power needed, but the performance of Simulator Z refuted that.  We've been saying all along that Simulator Z runs faster than pretty much all other simulators out there and it has a much higher max commute time.  Why didn't any of us see that connection between the two any sooner?   ???
[Edit: jplumbey's post below casts a shadow of doubt on what I've said in this post.]

The "low" commute times causing abandonment makes sense.  But that counter-intuitive higher max commute time = faster simulator makes sense also, after reading that line a few times.  If the max commute time were set to the minimum time that one commuter could to get to an available job (ignoring congestion), the simulator would find that job sooner than if it had to do an exhaustive search of the city tile of paths within the max commute time.  If the max commute time were set to one minute less, the simulator wouldn't find a path, and it would have to do the exhaustive search (leading to more time because of more calculations).


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

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

jplumbley

Quote from: z on November 07, 2009, 03:43:12 AM
Anyway, it all started when Jason challenged me to a duel.  ?$%kar&%h  (No, that can't be right.  ???  What was it?  ()what()  Oh yes!  &idea) to prove an assertion I had made based on experiments I performed when I first built Simulator Z.  I had found that increasing the pathfinding heuristic above .003 would produce abandonment in various situations.  Jason's response was:

So I did exactly what Jason had asked.  I took Simulator Z, raised the PH to .009 and showed him abandonment where none had existed before, and I did it all publicly in this long post.  However, in a later post, Jason stated the following:

OK.  I'm sure that there are other skeptics out there besides Jason.

Before I even read your post I have to comment on this...

I did not "challenge you to a duel" and I think you should stop acting that I am "dueling" with you.  I am tired of your attitude and the way you relentlessly attack my work with impunity.  And due to your attitude, yes I am very skeptical of your tests, I will be honest and tell you when I read them I feel bias in them from you.  Especially when you are trying to prove the "legitimacy" of your work compared to Simulator A.

Please realize that I have not been attacking your work and I am only defending mine.  I have not attacked your work and I would greatly appreciate that you stop attacking mine the way you have been.  If you were to stop and maybe change your attitude towards me a little bit then things wouldn't be so tense.  Believe me I have been trying really hard not to fight with you and give you some respect, unfortunately I do not feel you see it, nor are you even trying to give me that same respect.

I am sorry you have seen this as some kind of duel....  Where it is clearly a one-sided duel and one I am not trying to participate in but collaborate in a group effort.
"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

jplumbley

I have a problem with your assertion that "Higher Max Commute Time" = faster game speed.....  You have pointed out that Simulator A runs slower than other Simulators, which is true.  So, then if Maximum Commute Time is what creates this "slowness" why does Simulator A run slower than the default Simulator when Simulator A has a Max Commute Time almost 3x higher?

Your post is interesting.  You have shown that the 0.009 value can cause abandonment in any Simulator, but still there are questions why it happens in some cities but not all.  You again have attacked me in trying to say I have missed it... but it is hard to say I miss it when I don't run into the problem.  Your entire post seems hostile towards me.

I also disagree with your assertion that the City is "less attractive" to R$$$ Sims and more attractive to R$ Sims.  The PH should not be effecting the demand as this statement would suggest.  What is more likely is this is a side-effect of the Pathfinding itself, something more along the lines of something I posted earlier where the buildings are dilapidating due to the number of paths in that building being "not valid" but not enough to the point where the building has abandoned.  Quoted below is that theory I had posted over a week ago:

The relevant part in this quoted post is where it discusses the part of the theory is where the game determines how to abandon a building.  Dilapidation would come into effect here because it would create a buffer zone between "normal occupancy" and "outright abandonment".

Quote from: jplumbley on October 29, 2009, 06:48:06 PM
I dont begin to trivialize this as "elementary"...  I will say I do not know exactly how A* is used in SC4, but we do know it is a function in a larger equation or a function of a series of equations which determine the  IF, HOW, WHERE, WHEN, etc. for a Sim to reach thier destination.  What you have read only strengthens what my belief of the PH is used for, if in fact what you have read is true.

To me...

PH is the HOW of the Simulator.  It is what the Simulator uses to determine WHICH path to choose to get from point A to point B and the value of the PH, from my understanding, is how accurately it compares the routes that it finds.  But, this has no direct relationship to IF the path it chooses is a successful or unsuccessful path.  Now, from time to time depending on the accuracy of the value it may select a route that is slightly "slower".

Speed is the cost per tile, we know the Simulator will use Speed with multiple variables like the CvS, Intersection vs Speed properties and more to determine the cost of each route it checks.

The IF part, is where abandonment will come into play, and it should be as simple as an IF/THEN statement for the Simulator to compare the "best" route it finds for the Sim against the Maximum Commute Time.  If the total round trip cost of the trip is over 17, then the building becomes abandoned.

Now, the part that makes this theory dicey is the fact that there are 1000s of Sims in one building, with multiple destinations.... I do not know how MAXIS has handled this but it is very easy to see that in many cases your buildings dont have all of thier Sims commuting.  One might chalk this up to so Sims are kids and go to school instead of work, etc.  But maybe there is another reason for this, a more practical one in reference to the Traffic Simulator.  In a building of 1000s of Sims, even 100s of Sims, there will be routes that work and routes that dont.  The way I believe that MAXIS has handled the failed routes is by comparing the number of successful and unsuccessful routes from the same building and if the number of failed routes reachs a certain threshhold, such as 40% for example, then the building will become abandoned.  But, by the same token if the building has a much higher percentage of successful routes then the building will stay inhabitted and depending on how many unsuccessful routes are found could be used to determine the descrpancy between "building occupancy" and "building capacity".

So for the sake of an example for those who don't understand my theory:  (this will in no way be totally accurate, but an idea of how this may work)

Lets take a building that has a
Building Capactiy = 1000 Sims

Now, the 1000 Sims in the building will search for routes to the random destinations they have chosen for work.  The Simulator will do it's work, calculate the cost of each trip, compare them using the value of PH then compare the cost of each the best route for each Sim with the Commute Time allowed.

After the routes have been compared and the best ones for each of the 1000 Sims has been compared to the Commute Time, the Simulator finds that 700 or the 1000 Sims found a successful route to work.  Which, surpasses the 60% requirement (just an example remember) for the building to be considered "Occupied" so the building does not abandon.

So, you will probably see 700 Sims leaving the building and following thier route to work.  But when you query the building, it might say there are 893 Sims occupying the building.  Why are the other 193 Sims still in the building you may ask?  Well, it may only take a percentage of the Sims who did not find a successful route out of the building, but left the others in to possibly increase population for stage caps, demand and other purposes in the game that these Sims would effect.

It is a theory, but to be honest I think this is how the game really deals with pathfinding, abandonment and how they two together act to effect other parts of the game.  It is probably by no means a "perfect" description of how it works, but a general idea and I am sure there are other things that are not taken into consideration that take this basic theory and make it more complex under the same basic prinicpal I have just described.
"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

catty

Can I make a suggestion, the old SimCityscape site has quite literally 1000s of cities ranging from ones with only a few hundred sims to ones with millions of sims all of which can be downloaded.

http://simcity.ea.com/scape/index.php

If @jplumbley and @z were to select one of these cities as a test city that the testers could all go and download and set up some agreed test conditions;

a. Simulator A or Z
b. contents of plugins
c. changes allowed in the city
d. how long to run and what snaphots you want taking and when
e. ...... and so on

I for one would be happy to act as a tester and hopefully it might give us a clearer idea of the difference between SIM A and SIM Z and what's happening in the game.

Cathy
I meant," said Ipslore bitterly, "what is there in this world that truly makes living worthwhile?" DEATH thought about it. "CATS," he said eventually, "CATS ARE NICE.

ldog

Quote from: jplumbley on November 07, 2009, 07:06:35 PM
I have a problem with your assertion that "Higher Max Commute Time" = faster game speed.....  You have pointed out that Simulator A runs slower than other Simulators, which is true.  So, then if Maximum Commute Time is what creates this "slowness" why does Simulator A run slower than the default Simulator when Simulator A has a Max Commute Time almost 3x higher?

Your post is interesting.  You have shown that the 0.009 value can cause abandonment in any Simulator, but still there are questions why it happens in some cities but not all.  You again have attacked me in trying to say I have missed it... but it is hard to say I miss it when I don't run into the problem.  Your entire post seems hostile towards me.

I also disagree with your assertion that the City is "less attractive" to R$$$ Sims and more attractive to R$ Sims.  The PH should not be effecting the demand as this statement would suggest.  What is more likely is this is a side-effect of the Pathfinding itself, something more along the lines of something I posted earlier where the buildings are dilapidating due to the number of paths in that building being "not valid" but not enough to the point where the building has abandoned.  Quoted below is that theory I had posted over a week ago:

The relevant part in this quoted post is where it discusses the part of the theory is where the game determines how to abandon a building.  Dilapidation would come into effect here because it would create a buffer zone between "normal occupancy" and "outright abandonment".


Not just the max commute time, but the ph as well...I think. I haven't observed the speed effect myself, but then Steve was doing timed tests. I've had too much else going on to do any kind of timed tests yet.

I am also still able to get quite a lot of abandonment with .003 , same testbed. It takes a bit of congestion at least, but still the network is overkill and there are plenty of alternate routes. I just upped the max commute to 30 but haven't had time to get results (yes it is likely if I up it to 60 it will stop, but I am trying to find the lower end :P ). I also set everyone to 100% fastest method. Either it will help me isolate MT vs car or it will just fubar the game ;D .

I haven't had time to try to observe the attractiveness theory, although in regards to the ph, we know it shouldn't have anything to do with it; yet it is pretty clear it seems to effect more things than it should. As far as it being the pathfinder though, changing the ph does directly affect the pathfinder, so if you think it is the pathfinder then you can't disagree with the ph.

I still think your dilapidation/abandonment theory is valid, but I don't see how this contradicts it.

Anyway, I know your post was directed to Steve, but I had to throw my 2¢ in  :P

jplumbley

Quote from: ldog on November 07, 2009, 09:38:51 PM
I haven't had time to try to observe the attractiveness theory, although in regards to the ph, we know it shouldn't have anything to do with it; yet it is pretty clear it seems to effect more things than it should. As far as it being the pathfinder though, changing the ph does directly affect the pathfinder, so if you think it is the pathfinder then you can't disagree with the ph.

I still think your dilapidation/abandonment theory is valid, but I don't see how this contradicts it.

Anyway, I know your post was directed to Steve, but I had to throw my 2¢ in  :P

I think Steve's wording was probably misleading.  His tests seem to confirm the theory I posted last week, but the PH isn't effecting the demand (the way I would read attractiveness would in my mind be translated to demand in technical terms)... I do not think dilapidation is a mechanic of a demand related situation, but a technical response to how the game is handling the finding of "valid commuters" and "invalid commuters".

A question I would have now would be does the demand for R$ go down and the demand for R$$$ go up with the changes that occur?  If this is the case, then as a side-effect of that mechanic it would effect demand but instead R$$$ demand is rising instead of falling as Steve suggested.  But this would be due to the drop in population for R$$$ and rise in population for R$.  If the demand stays the same, the game may actually still interpret many of those buildings to be still considered R$$$ buildings even though they are dilapidated and still fill the higher wealth jobs available.

I simply dont see the direct relationship to demand that Steve has suggested.  I see a mechanic the game employs while handling "invalid traffic routes".
"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

z

Quote from: jplumbley on November 07, 2009, 06:05:04 PM
I did not "challenge you to a duel" and I think you should stop acting that I am "dueling" with you... I am sorry you have seen this as some kind of duel....  Where it is clearly a one-sided duel and one I am not trying to participate in but collaborate in a group effort.

There is a misunderstanding here.  I do not see this as a duel, or that you are dueling with me.  What you quoted was intended to be a joke, and to lighten things up a bit, especially as I was about to drag people through a long and complex post.  The "dueling" part was crossed out; I stated, "No, that can't be right," and I proceeded to make a little fun of myself for not seeing what was really going on.  It seems to me that if I was attacking anyone in that statement, it was myself.  Between that, the various animals jumping up and down, the other animated icons, and the other jokes, I really didn't expect anyone to take that part of the post seriously.  I am sorry that you did; please simply see this as humor that did not work as intended.

QuoteI am tired of your attitude and the way you relentlessly attack my work with impunity.

I don't understand this part.  I went over the main body of the post several times, doing my best to keep it clear of anything that might be interpreted as an attack on your work.  I changed a number of things that were slightly ambiguous for this very reason.  I threw in the "Disclaimer" which was also designed to be a little humorous, but which clearly stated that the purpose of my post was to understand properties of the simulator in general, and not to compare the worthiness of any simulators.  You had asked if the effects I had previously demonstrated with Simulator Z could be shown with Simulator A, and my tests were designed to show that.  The tests were focused on how various properties of the simulator worked; I just don't see where there was any attack.

QuoteBelieve me I have been trying really hard not to fight with you and give you some respect, unfortunately I do not feel you see it, nor are you even trying to give me that same respect.

I do believe you, and I have seen it.  And I have been trying to do the same with you, and trying to stick to dealing with facts.  So please, let's accept our mutual good intentions here (which I really do believe are genuine on both sides) and try to move forward.

Quote from: jplumbley on November 07, 2009, 07:06:35 PM
I have a problem with your assertion that "Higher Max Commute Time" = faster game speed.....  You have pointed out that Simulator A runs slower than other Simulators, which is true.  So, then if Maximum Commute Time is what creates this "slowness" why does Simulator A run slower than the default Simulator when Simulator A has a Max Commute Time almost 3x higher?

I haven't done any measurements on the speed of the default Simulator.  Have you, or have you seen any?

The effect of the maximum commute time on the speed is easy to verify, once you know to look for it.  Just take any city that's big enough so that the game slows down in Cheetah mode when Simulator A runs.  Time how fast the game runs.  Then change the maximum commute time to 600, and run the same test on the same city.  You should see a measurable difference.  Since the difference is connected with the pathfinding engine, it will vary exponentially with the size of the city, so you're unlikely to get the exact same results I did.  On the other hand, you should easily be able to replicate the effect whereby Simulator A runs slightly faster than Simulator Z.

Quote
Your post is interesting.  You have shown that the 0.009 value can cause abandonment in any Simulator, but still there are questions why it happens in some cities but not all.  You again have attacked me in trying to say I have missed it... but it is hard to say I miss it when I don't run into the problem.

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.)

QuoteI also disagree with your assertion that the City is "less attractive" to R$$$ Sims and more attractive to R$ Sims.  The PH should not be effecting the demand as this statement would suggest.  What is more likely is this is a side-effect of the Pathfinding itself, something more along the lines of something I posted earlier where the buildings are dilapidating due to the number of paths in that building being "not valid" but not enough to the point where the building has abandoned.

I agree with your theory here; this is the only explanation I could think of for why this was happening.  But at this point it's still a theory until this mechanism is proven to be the one in operation.  Nevertheless, I believe it; I do not believe, nor did I claim, that the PH has a direct effect on demand.  The reason I stated my conclusion in terms of attractiveness is that this is the observable effect; the rest is still theory at this point, even though I think it's a good theory.

Quote from: ldog on November 07, 2009, 09:38:51 PM
I am also still able to get quite a lot of abandonment with .003 , same testbed.

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 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.

Quote from: jplumbley on November 07, 2009, 10:03:36 PM
I think Steve's wording was probably misleading.  His tests seem to confirm the theory I posted last week, but the PH isn't effecting the demand (the way I would read attractiveness would in my mind be translated to demand in technical terms)... I do not think dilapidation is a mechanic of a demand related situation, but a technical response to how the game is handling the finding of "valid commuters" and "invalid commuters".

Just to confirm, I agree with this interpretation, and this is what I saw as the mechanism behind the results.

Quote
A question I would have now would be does the demand for R$ go down and the demand for R$$$ go up with the changes that occur?  If this is the case, then as a side-effect of that mechanic it would effect demand but instead R$$$ demand is rising instead of falling as Steve suggested.  But this would be due to the drop in population for R$$$ and rise in population for R$.  If the demand stays the same, the game may actually still interpret many of those buildings to be still considered R$$$ buildings even though they are dilapidated and still fill the higher wealth jobs available.

I did not see any change in demand, which makes sense to me.  The jobs are still there, but the R$$$ Sims can't find them.  I would think that the reason that the R$$$ Sims are affected in this way and not the R$ Sims is that there are fewer high-wealth  jobs than their low-wealth counterparts, and therefore a lot fewer valid paths between the R$$$ Sims and their jobs.  As a result, a decrease in pathfinder efficiency would hit the high-wealth Sims before it hit the low-wealth Sims.  From my tests, it appears that a value of .009 for the PH is high enough to create this problem for the R$$$ Sims, but not high enough to create it for the R$ Sims.  This is just a theory, though.

Quote
I simply dont see the direct relationship to demand that Steve has suggested.  I see a mechanic the game employs while handling "invalid traffic routes".

I don't see a direct relationship either, and I didn't mean to imply one.  It's an indirect relationship, and at this point we don't know for sure what the intermediate steps are.  But I do agree with your conclusion.

Now, should I put together that post about Nate's city that I mentioned earlier, or not?  If everybody's made up their mind one way or another about these issues, then there wouldn't seem to be much point in doing it, especially as it would take some time to do.  But if people think it would be helpful, I would be happy to do it.  Jason, if you are interested in such a post, you could help me by furnishing the capacity numbers for the simulator used in the tests.  Otherwise, I'll use my best estimate.

jplumbley

#295
Quote from: z on November 07, 2009, 11:56:36 PM
Jason, if you are interested in such a post, you could help me by furnishing the capacity numbers for the simulator used in the tests.  Otherwise, I'll use my best estimate.

I do not have exact numbers anymore....  As I have stated before the capacity numbers are very similar to those used in the Medium version +/- only a little bit within 50 to 100 in many cases.  I think if I remember back 2 years correctly the biggest change was in the MT capacities which were raised slightly from around 10,000 to where they are now.  It is very similar.

Quote from: z on November 07, 2009, 11:56:36 PM
There is a misunderstanding here.  I do not see this as a duel, or that you are dueling with me.  What you quoted was intended to be a joke, and to lighten things up a bit, especially as I was about to drag people through a long and complex post.  The "dueling" part was crossed out; I stated, "No, that can't be right," and I proceeded to make a little fun of myself for not seeing what was really going on.  It seems to me that if I was attacking anyone in that statement, it was myself.  Between that, the various animals jumping up and down, the other animated icons, and the other jokes, I really didn't expect anyone to take that part of the post seriously.  I am sorry that you did; please simply see this as humor that did not work as intended.

This was a very bad joke and one out of place given our history together.

Quote from: z on November 07, 2009, 11:56:36 PM

Quote from: jplumbley on November 07, 2009, 06:05:04 PM
I am tired of your attitude and the way you relentlessly attack my work with impunity.

I don't understand this part.  I went over the main body of the post several times, doing my best to keep it clear of anything that might be interpreted as an attack on your work.  I changed a number of things that were slightly ambiguous for this very reason.  I threw in the "Disclaimer" which was also designed to be a little humorous, but which clearly stated that the purpose of my post was to understand properties of the simulator in general, and not to compare the worthiness of any simulators.  You had asked if the effects I had previously demonstrated with Simulator Z could be shown with Simulator A, and my tests were designed to show that.  The tests were focused on how various properties of the simulator worked; I just don't see where there was any attack.

Quote from: jplumbley on November 07, 2009, 06:05:04 PMBefore I even read your post I have to comment on this...

You missed the very first line of my post... I had not read any of your post beyond your comment about "dueling".  The comment on your attitude has been a reflection of what I have felt since I've started posting again with your general posts.  But, yes there are points within your post above that I still feel you have either failed at humor or simply trying to undermine my questions.  Here is an example of what I am talking about:

Quote from: z on November 07, 2009, 03:43:12 AM
So I think I have demonstrated that the same effects I have shown happening in Simulator Z can also happen in Simulator A.  But obviously, they don't happen all the time, as Jason has pointed out.  So now it's time to answer his question:  Why hasn't he seen these effects?  I have also gone through the entire testing thread he referenced, and certainly, I think that the general feeling any reader of that thread would get would be that this simulator was working quite well, was a big improvement over previous simulators, and had no major problems.  I think one summary Jason posted is also relevant:

I would agree with this completely, and in fact, this is what I have concentrated on showing in this post.  Why didn't Jason see any of this?

How would you read this if the table was turned?  It seems to me you are trying to undermine my credibility even though in a later post you point out reasons why it really hasn't been a major problem in cities shown during testing, and the fact that my own playing style avoids that problem completely, which is answering your own question.

QuoteI haven't done any measurements on the speed of the default Simulator.  Have you, or have you seen any?

The effect of the maximum commute time on the speed is easy to verify, once you know to look for it.  Just take any city that's big enough so that the game slows down in Cheetah mode when Simulator A runs.  Time how fast the game runs.  Then change the maximum commute time to 600, and run the same test on the same city.  You should see a measurable difference.  Since the difference is connected with the pathfinding engine, it will vary exponentially with the size of the city, so you're unlikely to get the exact same results I did.  On the other hand, you should easily be able to replicate the effect whereby Simulator A runs slightly faster than Simulator Z.

I had a number of people question why their game ran slower after Simulator A was installed.  In a thread back in ST, long ago....  I don't have any test results on my own because I thought at the time it was related to the PH being lower.  But, I do remember the "slowdown" and I had been using the default Simulator prior to creating Simulator A.
"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

pierreh

I would like to second catty's suggestion in an earlier post above, for some 'neutral' testers to run tests on various cities found in the provided link. To the extent of my free time (which is not great these days and will get worse in the coming weeks because I will be moving) I would be pleased participate to such testing. Conditions of each test (city, simulator version/variant, duration, pictures to produce, etc) should be defined in detail to make sure that the tests produce valid and comparable results.

At any rate I wish that the discussions taking place in this thread - and in the related thread opened by ldog - remain as objective and factual as possible. They are invaluable as research and information about a particularly fascinating area of the game, and they allow players such as myself to learn a lot, to improve our playing styles and to try out new ways of building and interconnecting our cities. Please let's all try to remain dispassionate and zen about it.

pierreh

This is the report about the results of my second comparative tests between the Standard and the Euro Simulators, High variant. However, first I would ask Steve to open a new thread dedicated to the Euro variant of the Z Simulator and move the posts that pertain to it to that new thread, including this one, so that the discussion about the Euro variant can be separated from the much larger discussion taking place here.

The second test city is built on a large tile, on both shores of a wide river. There is still some room for expansion on the tile. The population at the start of the tests is very near 759k Sims.
The city has buses, a subway network that started to be developed when the city reached a population of 200k, and a railway network with two main stations (one on each side of the river) and some suburban stations. Once the basic subway network had been established, its development followed closely that of the city and in some cases anticipated it: when zoning new areas of residential or commercial, subway lines and stations were laid out right from the start, or new stations were opened on existing lines, sometimes with partial rerouting of some sections of tunnel. Bus stops are always present right from the start of any new zoning area, whether R, C, or I.
In many areas of the city the buses play a feeder role to the subway (and also to the railway); this can be observed by looking at various route queries.

Both tests were run for 10 years. The population increased moderately during the tests:
after 10 years with Standard High Simulator: 768k
after 10 years with Euro Hiigh Simulator: 770k

Each pair of pictures below shows first the Standard results then the Euro results.

Travel Time Graph





Car volume





Bus volume





Subway volume





Traffic densities; car, bus, subway and passenger trains






In the graphs, the flat lines during the first 18 months show the values produced by the standard Z High Simulator prior to the start of the test. The first run of the Simulator is about 18 months after the start of the test, and from that point on the graphs show variations.

Once again, not a lot of difference can be seen between the results produced by each variant of the Z High simulator. For buses, as mentioned above, the fact that there are a lot of short bus trips between the starting point and the next subway station, and the other subway station and the destination (feeder/distributor role of the buses) may explain the similarities observed between both tests. Subway volumes on the various segments of the network are a bit different between both tests but the global usage densities are quite similar.

b22rian

#298
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.)




    I had switched traffic sims from traffic sim B (hard) to traffic sim Z (low)..  So hope i can contribute
something worthwhile to the discussion..  Regrettably , i dont have any pics to confirm any of this but I'm
quite sure of my recollection as to why i changed traffic sims to Z (low)..

   At the time of the traffic Sim i had cities which were played all on large city tiles.. The largest of these cities
which as i recall, was right around a million sims, i began to experience abandonment due to commute times..
Trying some different strategies to deal with this issue, I eventually found the best solution was to mix the zones
as Steve was suggesting in the above post, and I found this solved my problem to a large degree , although
i still had spotty abandonment here and there.. it was also about this time steve  was putting the final touches
to sim Z, and i thought i would change traffic sims and give sim Z a try.. Once I changed all the commute
issues vanished for me.. , and ive been using sim Z without any further abandonment issues since..
( city is now approaching 2 million in population )..

    So anyways since there hasn't been nearly as much discussion on sim B in the various traffic sims I decided
to have a look at traffic sim B using the reader.. I found out the sim B hard i was using actually had a PH 8 times higher
value than sim Z does.. (0.025) as compared to .003..  So I just assumed the improvement i experienced to be
caused by a combination of (all) the changes Z made with sim Z.. (maximum commute times being longer), ect;
So I think its possible you can see changes based on just one factor sometimes like different (ph settings),
but i feel its more likely it often involves multiple factors , only because this area of the game is quite complex.
i do feel it does seem to take a few factors brought together , before the averge player (like myself) would
notice any worthwhile difference between the 3 traffic sims  (A, B or Z).. in my case..

     1.) .. city tile size (large) .
     2.) .. large city population  (over a million sims) .
     3.)... zone placed far apart.. ( fairly longer distances between where the sims live and have to commute
                                              .. to and from work ) .

over the past several months i have used and tested ( for steve).. all 3 of the traffic sims A, B. and Z..
and my position always was that any of the 3 traffic sims were excellent and high quality traffic sims for the
majority of players..  And i still recall back in the old days using that original maxis pathfinding engine, and would
feel very confident that any of the 3 traffic sims be it  (A, B or Z) are vastly superior to original traffic simulator
in many wayshave already been discussed and documented.. yet i hope i have contributed something to this 
discussion in a small way through my experiences playing all of them..

Thanks, Brian

z

Quote from: b22rian on November 08, 2009, 05:36:37 AM
I found out the sim B hard i was using actually had a lower PH value than sim Z does.. (0.025) as compared to .03.

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...