I expect that a large enough city with a not-too-good health (such as average) would stop generating ghosts once it got close to steady state (perhaps taking several hundred years). For details, see reply #3 of Timeless Tarraco. Trurl's city had 125120 people (with a binary .map file modification to allow 2x2 crummy houses to form anywhere). My Immortal Iunet with 12000 people at average health generated about 1 ghost per century, and my guess is that C3 treats deaths the same as Pharaoh. I thought that increasing the population by 20% (to 14400) would be enough to eventually stop making ghosts, but that's just a guess.

...taking several hundred years

I initially thought that it would take around 70 years to get the city to steady state, but building an "eternal" city I found out that I was wrong and it indeed takes several hundred years.

Getting census to stable state was also harder than I guessed it would be. The first wave of unemployment and then, most importantly, the first worker shortage wave were huge. During the latter I not only deleted all culture buildings that are not needed for services, but also had to devolve patricians back to plebeians and at the peak of the wave I even had to delete half of industries for a short time, to keep my city from falling apart.

I also encountered a devolution (and then evolution) of one house after 55 years of stable houses. Since Caesar's 3 walkers don't have a random part of the walk, I thought that if a city is stable for that long, then it will be stable forever. I am puzzled as to why that devolution happened and it makes me look more cautiously at players' claims of their cities being stable.

I thought that increasing the population by 20% (to 14400) would be enough to eventually stop making ghosts

This is a lot less than I initially guessed. I thought that an enormous population is required to stop the ghost production, probably around 100k people, as in Trurl's city. Although ghost production is hardly ever a problem, it's still kind of a good news that this is not true. To think of it, it is also fairly realistic that to survive for long the city has to be big.

EDIT:

a devolution (and then evolution) of one house after 55 years of stable houses

Looking at my earlier saves, I realized that it was actually 84 years, which makes it even more impressive.

[This message has been edited by equi (edited 03-29-2013 @ 08:56 PM).]

Brugle,

I've not seen that Timeless Tarraco thread before. If I had, I might have solved the cartpusher warehouse selection problem earlier since there is a big clue in your report in reply #38 to the fact that the bay is the destination and not the warehouse. There is also a clue in reply #45 where a behaviour was noted to have changed (we can now say that probably something was built/deleted which changed the distance to the map entry point).

equi - sorry to interrupt your thread If you've not alreay seen it my research on death rates contains some explanation of ghost formation. I see Timeless Tarraco is mentioned there too but I didn't look at it and hadn't realized that Trurl had already established some of the information in my post.

[This message has been edited by Trium (edited 03-29-2013 @ 09:39 PM).]

I encountered a devolution (and then evolution) of one house...

I think I'll share what was the problem. There are 2 markets in a housing block, and one of them regularly, but for a very short time, runs out of oil. It's like 1 month it doesn't have oil and 100 months it does.
Then there is also a rare moment when the walks of market traders arrange in a way that one house, for a couple of months, is only passed by a trader from this market.
When both these rare events coincided with each other and with the moment when the house is short on oil, it devolved. After 84 years.

Bottom line is: I moved the oil warehouse a little closer and the problem was solved

This is a request for personal preferences. I am thinking of Trium as I write it, but I would like to hear from anyone.

whisperwind777 had some very well-built cities. equi said that whisperwind777's Lutetia was solid for over 800 years (with changes to health and buildings, but I don't think that affected its stability). That's somewhat short for what I would consider eternal (ignoring the occasional gifts), but is still superb.

In your opinion, what makes an "eternal" city like whisperwind777's Lutetia so good? In my opinion, the one thing that stands out is that all roads are connected and that there are no gatehouses except in outer walls. But is there something else? I'd like to know what people really think makes an eternal city well-done (before putting a lot of work into another).

One of my unusual cities (at the time) was my Lugdunum Cliff Dwellers, which required a lot of trail-and-error to get the walkers to work properly. It would be easier to make something like that with our current knowledge, but I think it would still take a lot of trial-and-error. In contrast, most of my cities were designed to be built without testing how walkers behave. Is it more important to design a city beforehand or to get it running the best way with trial-and-error?

Note: I am not committing to building anything. If I had to guess, the next city I build might not be eternal, or it might be my Improved Immortal Iunet in Pharaoh (partly to see how much a few thousand more people would reduce ghost formation).

[This message has been edited by Brugle (edited 04-08-2013 @ 07:32 PM).]

Yes, I've made no secret of the fact that I'm a great fan of ww777 (who I seem to remember was Chinese) but I don't think I was very interested in whether the cities were 'eternal'. Sure, I like a city which doesn't collapse to the ground within a year or two, but I liked his cities for other reasons. As you mention, interconnected roads without gatehouses impressed me a lot, as did his clever use of gardens, his extremely efficient use of space (his entire Valentia is crammed into one corner of the map) and his mastery of walker control - one engineer covering the whole of Tarsus was it?. Usually 100% hippodrome coverage too if I remember rightly (in cities with a hippodrome). I guess the 'eternal' goal forced some of the design efficiencies on him, but he took it to levels I wasn't used to seeing.

In some of his cities you'll see odd plaza tiles here and there and you realize that they mark walk targets for specific buildings. There may have been some trial and error or good fortune along the way, but he largely planned for people to go where he wanted them to. I do some of that myself these days, but at the time I was somewhat bedazzled. Added to all that, I simply found his cities aesthetically pleasing, and I suppose it's each to his own in that area. But you needn't be jealous Your highly original masterpiece in Lugdunum Cliff Dwellers remains the all-time classic career city.

Regarding the OP - I doubt there are any truly eternal C3 cities - at least not on non-northern maps. While in most cases, as I recall ww777 remarking, a cycle of four walks is enough to establish whether there is adequate coverage of that walker's service in perpetuity, the nature of fire risk means that no matter how long you run a city you can never guarantee that a building won't catch fire.

[This message has been edited by Trium (edited 04-09-2013 @ 05:46 AM).]

Quoted from Trium:

his extremely efficient use of space

I don't generally try to achieve that, except to some extent in military missions (such as my Timeless Tarraco) when I try to keep all of the city far from any place that might be an invasion point (without knowing where invasions occur). I'll keep it in mind.

Quoted from Trium:

one engineer covering the whole of Tarsus was it?

No, whisperwind777's Tarsus had 3 engineer posts, but that was very good. (My Happy Tarsus had 7.)

Quoted from Trium:

But you needn't be jealous

Jealous isn't quite right. I expect I'll eventually (although it may take years) build something similar (if it hasn't been done before)--a timeless C3 city (no gifts, etc) or a timeless Pharaoh city with all good houses--and I'd like to know what some additional targets might be.

Quoted from Trium:

I doubt there are any truly eternal C3 cities - at least not on non-northern maps ... you can never guarantee that a building won't catch fire

I'd consider a city "eternal" if it lasted for thousands of years and looked like it could last a lot longer. The lack of a fire for that long would be OK with me. (Something would probably break down eventually--perhaps cash would overflow into a negative number.)

Thanks very much. Anyone else have comments?

Quoted from Brugle:
the one thing that stands out is that all roads are connected and that there are no gatehouses except in outer walls. But is there something else?

No gatehouses and inter-connection is clearly not enough. I can build a city of dead-end roads using these goals, but it will be a lot worse than whisperwind's cities.
The rule of not having any dead-end roads (except for possibly 1-tile dead-ends in granaries) might be interesting. Many of his cities satisfy this rule and in others there is only 1 or 2 such roads.

Whisperwind's cities are very efficient in workers. His Lutetia (and it is likely that it's not the best example) for about 10k population has, among other things, 3 engineers, 2 baths and 3 warehouses. It's quite impressive.

I want to think up some good goal which imposes worker efficiency, but so far I haven't come up with one. Culture 100 together with everlasting are good enough for lower populations (say 10k and below), but no good for bigger cities.

Whisperwind's cities are also very efficient in space, but I don't think that one should set himself any rules in this regard. However, a city usually looks nicer when it's "dense".

As for aesthetics, I like his cities, but I'd give them 8 out of 10 I think I remember Brugle commenting somewhere that he didn't like clumps of several culture buildings of the same type. I definitely agree with that. It is perhaps most aesthetically unpleasing part in whisperwind's cities for me.

Quoted from equi:

I think I remember Brugle commenting somewhere that he didn't like clumps of several culture buildings of the same type

I was going to mention that in an eternal city it would be a challenge to blend in the required number of schools and academies for a high culture rating without using clumps of them. However, in real life major cities have theatre districts, culture districts, etc (when I lived in Liverpool my neighbourhood was renowned for its many posh schools) so I've never really objected to the zoning approach on grounds of realism.

a city usually looks nicer when it's "dense".

I agree. Too many of my cities have clearly separated housing/industrial/farming blocks scattered around.* If you look at the main clump of ww777's Massilia, for example, it looks very natural, busy and 'organic'. Without using an overlay it's difficult to see the underlying block-like structure and I like that.

I forgot to mention earlier - most (if not all) of ww's cities have wine in all/most grand Insulae. In some cities he has denied a second god to prevent evolution to villae but in some, eg Valentia, evolutionary control is purely through desirability. That is pretty cool. Of course, none of these individual properties are overwhelmingly difficult (though some are challenging). It's incorporating them all harmoniously into a single design that, to paraphrase goonsquad, separates the sheep from the goats.

*Largely because I don't design too much in advance, leaving lots of space as I go in case I need it later then never using it. A city preplanned on paper can produce a better 'fit' of its components

[This message has been edited by Trium (edited 04-09-2013 @ 01:46 PM).]

Quoted from Brugle:
Is it more important to design a city beforehand or to get it running the best way with trial-and-error?

I don't think you should impose any restrictions on a building process itself (like no deletions or no trial and error) when attempting such city.

However, If I ever try to build something similar, what puzzles me most is what knowledge should I be allowed to use. I'll explain.

In one of my other threads I made up a stupid term "theoretically stable". Now I'll throw in 3 terms instead:

Stable in the sense I
A city that would be stable if random walkers walked randomly all the way before returning home (with no turning back given choice).
It would basically restrict players to building intersection-free loops and not-so-long dead-end roads.

Stable in the sense II
A city that would be stable if random walkers walked randomly the second part of the walk.
i.e. Ambulomancy is allowed.
Some of whisperwind's cities are not stable in the sense II. But some are.

Stable in the sense III
Assuming everything is as it is.

With the goals discussed in this thread, it's clearly the choice between II and III. But I am hesitant to build a stable in the sense III city, because it seems to me to be very borderline in terms of exploiting the game where we shouldn't, so it almost becomes cheating.
I would be very interested in other players' thoughts on it.

-----
edit:

Some of whisperwind's cities are not stable in the sense II.

I shouldn't have said it as confidently. I don't know that for a fact, it just seemed to me that there would (eventually) be some problems with fire protection in some of his cities.

I also should have rephrased everything without making up stupid terms.

edit 2:
This post just shows my inexperience and lack of knowledge. No need for anyone to reply to it.

[This message has been edited by equi (edited 04-10-2013 @ 06:37 AM).]

Thank you both for your comments. There were exactly what I was looking for.

I have some comments but will wait until I can check a few things. I left home yesterday for a while, but I do have a computer here. When I can, I'll download C3 (from GOG) and whisperwind777's cities.

I downloaded C3 and whisperwind777's cities onto this computer and looked at only my favorite career mission, Lugdunum. Many of its houses devolved after about 6 months because the gladiator school was not connected to the amphitheater or colosseum (except through gardens). I know that whisperwind777 was careless about labeling cities "eternal" or "timeless" when they were not, but I didn't expect anything this bad. So I wonder: he was working with an unpatched C3--did C3 version 1.0 allow gladiators to go from their school to a venue through a road-and-garden path when there wasn't also a road connection? I have only the GOG.com version here and can't check for a while. Could a player with an old unpatched C3 test if that's the case?

Eventually I'll look at whisperwind777's cities and will have more comments about making an eternal city attractive.

I loaded his Lugdunum. I have a patched version, but I can see that there are a couple of gladiators going through this garden connection, heading to the amphitheater. But no "new" gladiators are able to go through it. I guess it's because of the version.

Do you remember of any of his cities which are not eternal in the state they are currently in the downloads? I would be interested to take a look at their problems.

Quoted from equi:

I can see that there are a couple of gladiators going through this garden connection, heading to the amphitheater

There are also a few gladiators behind them going to the colosseum.

Quoted from equi:

no "new" gladiators are able to go through it. I guess it's because of the version.

I guess it's not because of the version, but I could be wrong. If someone else can't make a new or second install of version 1.0, then in a week or two I'll be home and will do it myself.

Quoted from equi:

Do you remember of any of his cities which are not eternal in the state they are currently in the downloads?

I made comments about instability in the first Capua, Tarsus, and Massilia, but it appears (from the dates in the Downloads) that all of them have been updated since those comments. I don't know if those problems were fixed or if other problems exist. I haven't recently looked at any of his saved games except Lugdunum.

[This message has been edited by Brugle (edited 04-11-2013 @ 01:12 PM).]

The first theatre across the bridge gets an actor just after the save loads but no more ever go to it. I don't have an unpatched C3, but I think this solves the puzzle of why this 96 Luxury Palace city from Micerin collapses in ruins because only entertainers already on the map go to venues.

Brugle,

I was editing my post when you submitted yours. I replied too quickly and hadn't thought it through.

--
His Lugdunum is puzzling.

Prefect Marty reported that there is no such difference between the versions. In this thread, the relevant part starts from reply 5. Not only I tend to believe him, but it also seems unlikely to me for such a difference to exist.

However, I don't believe a mistake like this happening in a whisperwind's city either. I am no Sherlock Holmes, but I changed the speed to 10%, reloaded the city, and immediately paused the game. I noticed the gladiator near the school. He was headed to the Colosseum and he had walked only 3 tiles from his spawning square. I also noted that there are not one, but two garden connections on his way. It makes the time required for replacing previously put roads to be longer. All combined, it would have required quite a good timing to replace roads with gardens and save the game.

Whisperwind didn't post much on this forum, but there is a thread he started on his Lugdunum. Here. He mentions cancelling the landslides and running the city at 100% speed. He also mentions a strange version of the game.

Either way, you have to take a look at his other cities. I don't remember any of them having a problem like this. It's unfortunate that there is such a strange thing about his Lugdunum. Not only it's a very interesting mission, but also, speaking of aesthetics, I find his Lugdunum to be very appealing. I guess partially for the fact that no same culture buildings are located right next to each other.

EDIT:

Looking at the population history, it's a straight line. Looking at the census curve, it's stable. Looking at the messages, landslides are turned off.
I don't know the answer for this riddle.

[This message has been edited by equi (edited 04-11-2013 @ 02:24 PM).]

Quoted from Trium:

I don't have an unpatched C3

I thought one of you (or someone else reading this) might have an old CD which could install version 1.0. Oh, well...

Quoted from Trium:

I think this solves the puzzle of why this 96 Luxury Palace city from Micerin collapses in ruins because only entertainers already on the map go to venues

Yes, it might. However, notice that the description of the file says "Need special technique to maintain this city." Hmm...

Quoted from equi:

Prefect Marty reported that there is no such difference between the versions.

True. Still, I'd like to test it again.

Quoted from equi:

it also seems unlikely to me for such a difference to exist

Neither do I, which is why I guessed that it wasn't the version.

Quoted from equi:

it would have required quite a good timing to replace roads with gardens and save the game

No, it would only take 3 or 4 ticks--the gladiator would move a fraction of a tile.

Quoted from equi:

there is a thread he started on his Lugdunum

Yes, I read that (along with the other threads he wrote and comments on his downloadable cities) to try to answer your question about his non-eternal cities.

Quoted from equi:

He mentions cancelling the landslides

Yes, he learned how to cancel flooded clay pits and other problems (such as landslides) from Trium in Tools to turn off random clay-pit collapse in .sav files? just a few days before.

Quoted from equi:

He also mentions a strange version of the game.

He didn't say "strange", but he did say it was "neither for British nor North America" and he "cannot find any patch available". There are various reasons why a system might not take the patch--perhaps it had a trainer. We could speculate how a system that has entertainers behaving strangely could get into the world, but I'd rather wait until someone tests version 1.0.

Quoted from equi:

you have to take a look at his other cities

What do you think "Eventually I'll look at whisperwind777's cities" means?

Quoted from equi:

I find his Lugdunum to be very appealing. I guess partially for the fact that no same culture buildings are located right next to each other.

Yes, I noticed that. It would be included when I "will have more comments about making an eternal city attractive".

Quoted from equi:

Looking at the population history, it's a straight line. Looking at the census curve, it's stable. Looking at the messages, landslides are turned off.
I don't know the answer for this riddle.

As far as I can tell, the only riddle is the entertainers. Of course the population history (for roughly the last 33 years) is a straight line. Of course the landslides are turned off--whisperwind777 said he cancelled them. The census curve is not stable and probably never will be stable, but it's pretty good since the city has been running for over 220 years.

[This message has been edited by Brugle (edited 04-11-2013 @ 08:24 PM).]

A straight line in population history shows that the population was stable for 30-something years, the (relative) stability of the curve makes it likely that the population was stable for a lot longer. It would be an extremely tedious process to constantly replace gardens with roads and back to maintain this stability, if it can maintain it at all. I think that the probability that whisperwind was doing something like this is close to zero.

Quoted from Brugle:
I thought that increasing the population by 20% (to 14400) would be enough to eventually stop making ghosts

How likely do you think it is for an eternal city of this population, with culture 100, to hit the walker sprite limit? If it's fairly likely, could you (or anybody) give some suggestions on how to deal with it?

The save file keeps monthly populatiom records for only the last 200 years (!). The earliest record shows a population of 4524 (2 ghosts). Apart from a third ghost forming 180 years ago there has been no change. However, in a city of mostly Grand Insulae it is possible for some houses to briefly devolve one level, perhaps frequently, without displacing population.

Correction - looking more closely there were periods of instability from time to time. The last change was a ghost forming 12 years ago. Apparent devolution last occurred a little over 40 years ago.

I suppose you've tried replacing the gardens with road tiles? (some houses run out of furniture in May 59)

Quoted from equi:

How likely do you think it is for an eternal city of this population, with culture 100, to hit the walker sprite limit? If it's fairly likely, could you (or anybody) give some suggestions on how to deal with it?

With half-decent design, not very likely. If the walker limit did pose a problem there are many ways of dealing with it, including providing only a single road tile against culture buildings whose service walker is not required so that walkers just 'blink' every 50 ticks and disappear.

[This message has been edited by Trium (edited 04-12-2013 @ 06:43 PM).]

Quoted from equi:

Quoted from Brugle:
I thought that increasing the population by 20% (to 14400) would be enough to eventually stop making ghosts

How likely do you think it is for an eternal city of this population, with culture 100, to hit the walker sprite limit? If it's fairly likely, could you (or anybody) give some suggestions on how to deal with it?

I don't have a good answer. I thought that the probability of hitting the walker sprite limit would be small for a city of 12000 people or I wouldn't have started Timeless Tarraco. For a larger city I'd have to check some things (perhaps getting a utility from Trium). Since Pharaoh has higher sprite limits, the next timeless city I build might be an Improved Immortal Iunet.

It is possible that walkers would be created in building order, so if all key buildings in an eternal city were early in the building list then occasionally hitting the walker sprite limit might not be a problem. (It would be very nice to check that!) Aside from that, the only suggestion I can come up with now is to try to stagger walkers, so that about the same number of school kids are active at any give time. Trium has a good suggestion if you don't connect all roads.

Note that my guess that a city of 14400 people would eventually not have ghosts is my opinion but I wouldn't put too high a probability on it. (I'm fairly certain that it would eventually have a ghost generation rate of well under 1 per century.)

Quoted from Trium:

I suppose you've tried replacing the gardens with road tiles? (some houses run out of furniture in May 59)

If that's addressed to me, then yes I did. (I meant to mention doing it in reply #12 but forgot.)

From the description of the patch in the downloads section:

What the update fixes

A bug was fixed where a gladiator's school was attempting to send gladiators to an amphitheater that was not road accessible.

This could be the answer for the mystery.

Quoted from Trium:
However, in a city of mostly Grand Insulae it is possible for some houses to briefly devolve one level, perhaps frequently, without displacing population.

Only the population stability is mentioned in my post, but speaking of grand/large insulae devolution—is there any way to check if it happens in a city? I am always paranoid that it happens in part of my city I am not looking at

Quoted from Trium:
providing only a single road tile against culture buildings whose service walker is not required

This is a very good suggestion, but it would be against the interconnection rule.

Quoted from Trium:

How likely do you think it is...to hit the walker sprite limit?

not very likely

I did some estimations and it indeed seems unlikely to hit the sprite limit at 14400.

Quoted from equi:

speaking of grand/large insulae devolution—is there any way to check if it happens in a city?

Nothing much springs to mind. If I wanted to detect it I could probably write a script to compare houses in two saves and check if second food consumption is as expected (you can do this manually if you want ). Other than that, I'd keep a close eye on the society census - if LIs exists at month roll-over they will continue to show on the census until the following month. Neither of these methods would detect a brief devolution unless it was conveniently timed.

Quoted from Brugle:

It is possible that walkers would be created in building order, so if all key buildings in an eternal city were early in the building list then occasionally hitting the walker sprite limit might not be a problem. (It would be very nice to check that!)

I think that's very likely. I just did a quick test placing random buildings in an empty city then placing a tent (citizens spawn simultaneously as soon as there is a population). The citizens can be individually identified in the walker table by their coordinates (and those of their home building). They are all listed in the same order as their home buildings show in the buildings table (though after seabirds and sheep, of course)

[This message has been edited by Trium (edited 04-13-2013 @ 08:26 PM).]

Quoted from Trium:

Correction - looking more closely there were periods of instability from time to time. The last change was a ghost forming 12 years ago. Apparent devolution last occurred a little over 40 years ago.

I had planned to discuss devolution, but more detail would be nice. How often and how much (each time) did the population recover (after dropping)? Did you check any other of whisperwind777's cities, perhaps Lutetia?

Quoted from Brugle:

more detail would be nice

You got it Check your email.

I realised in the process of collecting this data that when the 2400-entry history array gets filled up it starts over-writing the first entries rather than tagging new data on the end (fairly logical, really), so the details I gave in reply #19 are wrong. The city has been stable for 68 years, bar ghost formation 40 and 8 years ago.

Regarding cities ran for hundreds of years, assuming no housing devolution, what does the labour force fluctuation tend to be?

Furthermore, do Ghosts pay taxes?