Quoted from Trium:

I'm still struggling to find much time for this study

Please, only continue if you have time and enjoy it. (I will finish analyzing whisperwind777's cities, but it could take a while.) I do appreciate your finishing and updating "Census - Death Rates".

Quoted from Trium:

your Palace Peaks (which has 'excellent' health) creates 8 ghosts 51 years after the save date {ie in 129 AD), 1 after 76 years, another 5 after 98 years and 3 more after 166 years (for a total of 17). The census stabilizes with 1260-1262 births/deaths a year with nobody making it beyond 95.

Again, I think we should make clear that Palace Peaks does not do that--it devolves less than 24 years after the save date. Your population-only projection of Palace Peaks does that.

I thought it might interest you anyway

It does. 8 ghosts (almost half) created 99 years after the rebuilding began--I don't understand this (or why the 3 ghosts were created so late). 5 ghosts created 99.X years after city health improved to average and 1 ghost created 99.Y years after city health improved to excellent--those make sense.

When you (or equi) have both the time and the inclination, it would be nice to have the percentages of workers, school kids, and academy kids for stable large populations at very good, excellent, and almost perfect city health. (We got the numbers for good, average, and below average city health from Trurl aka lemmus.) Your "Palace Peaks" run might give the excellent city health numbers.

Trium,

In #63, you were tempted to ask me something.

The denarius has finally dropped.

Hi!

I have been trying to hide the fact that I've been lurking around for a while, at least until such time that I could perhaps build a city. Alas, I don't have enough time for such a thing.

However, this topic interested me quite a bit. I had a few spare hours this morning and decided to generate some stable demographic curves.

[DISCLAIMER: The data that follows is based on my untested understanding of deaths and births as described in Trium's Census - Death Rates. It may be completely incorrect.]

With the disclaimer out of the way, here's the data.

The data comes in pairs of two lines. Each pair represents a starting curve and a final curve. Each curve in the output is guaranteed, save for the disclaimer, to not produce any more ghosts.

Furthermore, the data comes in two sections. The first section has as input demographics that are almost stable already. The second section takes as input a curve with every single person only just having been born.

The input of the first section was generated by simulating the actual death/birth process for a million using the input of the second section, but without the bug that causes ghosts.

The numbers on each line mean the following:

1. The amount of people we start with.
   
2. The amount of people that are not ghosts.
   
3. City health, ranging from 0 to 10 with 0 being Appalling and 10 being Perfect. Only City health between 5 and 9 was tested.
   
4. % of school children
   
5. % of academics
   
6. % of workers
   

The rest of the numbers are the amount of people of various ages, ranging from 0 to 100.

Note that the percentages are not averages.

I hope this is of some use to someone out there.

[This message has been edited by Shenghi (edited 04-27-2013 @ 06:03 AM).]

There is a problem with my ghost prediction routines It only becomes evident in larger populations and/or better health levels, so my forecasts for ww777's Mediolanum are unaffected. The forecast malfunctions in two ways - the first is possibly me misunderstanding equi's description while the second seems to indicate that we have not fully recognized the nature of the ghost 'bug'.

My model assumed that 100-year-olds were included in the 'sweet-sharing' and so some would die if enough deaths occur in the 90-100 age group to affect those at 100. Not only is that not so, but an extra person is killed off in the 90-99 group for every 100-year-old.

I was testing a modestly-sized Fishopolis population (larger ones need regular festivals to sustain them) under almost perfect health. The population was 63555. An unpredicted ghost formed immediately following this census:

89-99 years:
54 31 0 0 0 0 0 0 0 0 1

Total in group is 86, number of deaths @ 20% is 17. My model expected this to be distributed as 8, 8 and 1 (no ghost). If the 100-year-old is exempt, I'd have expected 9 and 8 (1 ghost). The game kills 9 and 9 making 18 (1 ghost). This also results in one more birth than anticipated.

A few years later, my census was:

89-99 years:
181 220 170 148 24 158 0 15 0 0 39

Total in group is 955, expected deaths 191, or 183 if the last 39 are excluded. Game kills 222 (=183+39) and forms 39 ghosts.

My current thinking is that this is two groups - those who will be 100 are a group of their own with 100% death rate, but the game applies those deaths to the wrong group! That would explain both of the situations above. If I find that it explains all other situations I'll amend the model and update the forecasts I've made with it, though it's been a long winter and it's far too nice and sunny outside to spend the day on it. In the meantime, please don't place too much reliance on figures given previously for specific populations.

[This message has been edited by Trium (edited 04-27-2013 @ 05:16 AM).]

Quoted from Trium:

There is a problem with my ghost prediction routines It only becomes evident in larger populations and/or better health levels, so my forecasts for ww777's Mediolanum are unaffected. The forecast malfunctions in two ways - the first is possibly me misunderstanding equi's description while the second seems to indicate that we have not fully recognized the nature of the ghost 'bug'.

My model assumed that 100-year-olds were included in the 'sweet-sharing' and so some would die if enough deaths occur in the 90-100 age group to affect those at 100. Not only is that not so, but an extra person is killed off in the 90-99 group for every 100-year-old.

My model assumed the exact same thing. I should probably fix that.

I was going to just edit my data based on Trium's findings described in reply #79 into my reply #78 and walk away.

Then I had a look at it, and found that the results of the set of data that has everyone starting at age 0 is significantly different using this new model.

With the old model, when everyone starts at age 0 the first time a cycle is found, it's a nice curve. However, with the new model such a curve is never formed which makes it a terrible base for any useful data.

The good news is that, given a curve that is already smooth, it appears that any amount of people can produce a cycle that does not lead to ghosts! The challenge may be to find out what kind of curve you need to begin with.

[edit]
I found a bug in my new model. One that prevented ghosts from forming at all. This data is completely useless.

[This message has been edited by Shenghi (edited 04-27-2013 @ 06:35 PM).]

Quoted from Caledonius:

Trium,

In #63, you were tempted to ask me something.

The denarius has finally dropped.

I'm still clueless. Will someone explain?

Quoted from Trium:

There is a problem with my ghost prediction routines

I feel guilty about my reply #67, where I encouraged you to update your "Census - Death Rates" (in contrast to your scientific attitude expressed in reply #43).

Quoted from Trium:

I was testing a modestly-sized Fishopolis population (larger ones need regular festivals to sustain them) under almost perfect health. The population was 63555. An unpredicted ghost formed immediately following this census:

89-99 years:
54 31 0 0 0 0 0 0 0 0 1

Total in group is 86, number of deaths @ 20% is 17. My model expected this to be distributed as 8, 8 and 1 (no ghost). If the 100-year-old is exempt, I'd have expected 9 and 8 (1 ghost). The game kills 9 and 9 making 18 (1 ghost). This also results in one more birth than anticipated.

A few years later, my census was:

89-99 years:
181 220 170 148 24 158 0 15 0 0 39

Total in group is 955, expected deaths 191, or 183 if the last 39 are excluded. Game kills 222 (=183+39) and forms 39 ghosts.

My current thinking is that this is two groups - those who will be 100 are a group of their own with 100% death rate, but the game applies those deaths to the wrong group!

The only simple model I came up with which agreed both with your two censuses above and equi's two examples in reply #42 was that the deaths for 90-99 and the 100% deaths for 100 were added together and applied to 90-99 (but the program would sometimes have to not make all of the expected deaths, as in equi's first example). That model would mean the 222 deaths in your second census would be distributed as 37, 37, 37, 36, 24, 36, 0, 15. Did that occur?

Quoted from Trium:

it's been a long winter and it's far too nice and sunny outside to spend the day on it

I know the feeling. Unlike much of the USA, Seattle didn't have a cold winter, but it was especially dark.

[This message has been edited by Brugle (edited 04-27-2013 @ 01:46 PM).]

That model would mean the 222 deaths in your second census would be distributed as 37, 37, 37, 36, 24, 36, 0, 15. Did that occur?

Exactly that. I did ponder on the fact both sets of deaths do seem to be executed (no pun) in one batch, that is the game does not seem to start over with the youngest when inflicting the second set. I expect your explanation is close.

I feel guilty about my reply #67, where I encouraged you to update your "Census - Death Rates"

Not at all. It's about time I got around to it. I feel guilty that Shenghi posted so much data just as I was typing my announcement that the model on which it is based is wrong.

I'm still working on it ...

Quoted from Trium:

I feel guilty that Shenghi posted so much data just as I was typing my announcement that the model on which it is based is wrong.

You shouldn't. It's the risk of working with brand new research, and I was well aware that equi's findings of handling 100-year-olds could be interpreted in more than one way.

After I read your post I hastily implemented the new model, and as it turns out, that was too hastily and the data presented in reply #81 is therefore completely useless -- not a single ghost was generated. It's funny how it can sometimes be harder to reproduce a bug than to implement something the correct way.

Here's the data again. Once again based on the new understanding as described in reply #79, but this time without the bug that prevented ghost production.

I would like to have some more realistic input data, without having to actually read them from existing cities. The reason for not wanting to use existing cities is that it's much easier to generate lots of input data for different city sizes than having to build or find such cities. Not to mention I don't have a save file reader and don't have the time to write one.

Is there any known data about the age distribution of immigrants?

[This message has been edited by Shenghi (edited 04-27-2013 @ 06:56 PM).]

Quoted from Shenghi:

Is there any known data about the age distribution of immigrants?

I don't know of any. But does it matter? After hundreds of years the data will look fairly similar no matter how it starts.

It's good to hear from you again.

Trium and Brugle,

I feel guilty more than both of you, because I was the one who posted the misleading information. I hope that Trium's guess turns out to be right and we will finally have a correct model.

Quoted from Brugle:

I'd like to know whether a designed (but not yet built) city is likely to lose only a small number of ghosts before becoming stable.

How large does the population need to be before almost any city can lose only a modest number of people before becoming stable?

These are the questions that need answers, but even if my theory were correct, I wouldn't have good answers. I too don't want to lose many people due to ghosting. When I say that some ridiculous number of people is "not a big loss", I mean that compared to losing everybody it doesn't seem too bad. I want to lose as few people as possible in my cities.

Shenghi,

It's nice to see you around. I remember reading many interesting posts by you.

Quoted from Shenghi:
Is there any known data about the age distribution of immigrants?

For my program I used the numbers suggested by Theo in the reply #6 of the thread Immigrants (ages from 0 to 50, random, with higher probability in the middle).

To make this post slightly less uninformative, I'll post two interesting censuses. These are what I assume to be the two "smallest" stable curves with a cycle of 1 year at "almost perfect" health. There is never a change in workforce in these censuses. Read from left to right. The numbers are hexadecimal.

4656 people

ingame picture

3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d
3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d
3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d 3d
3b 39 37 35 33 31 2f 2d 2b 29 27 25 23 21 1f 1d 1c 1b 1a 19
17 15 13 11 0f 0d 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 00

workforce: 1098 (23.58%)
schoolchildren: 854 (18.34%)
academy youth: 427 (9.17%)

4802 people

ingame picture

3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f
3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f
3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f 3f
3c 3a 38 36 34 32 30 2e 2c 2a 28 26 24 22 20 1e 1d 1c 1b 1a
18 16 14 12 10 0e 0c 0b 0a 09 08 07 06 05 04 03 02 01 00 00

workforce: 1134 (23.62%)
schoolchildren: 882 (18.37%)
academy youth: 441 (9.18%)

Trium,

Could you please tell me the offset for census in Pharaoh saves? I know nothing about them.

EDIT:

I found all stable curves that always have a straight line from 0 to 60 at "almost perfect" health (and that means no change in workforce, schoolchildren or academy youth). There is a fair amount of those in population range 4656–6247, then there is a gap, and then starting from 10282 they occur very frequently. My guess would be that no matter what the exact nature of the ghost bug is, if there is somewhat more than 10282 people in a city, then at "almost perfect" health the city won't lose many people before becoming stable.

[This message has been edited by equi (edited 04-27-2013 @ 11:55 PM).]

Could you please tell me the offset for census in Pharaoh saves?

The main problem lies in decompressing the file. It's compressed in the same way as a C3 file using the same 37 sections, but some 16-bit data is now 32-bit and the size of the sections is different. C3GameExplorer cannot be used and although Pecunia wrote a decompressor for her minimap series she uses it internally and provides no way to export the output.

My own decompressor is written in Smalltalk and has no GUI. I generally don't release my utilities because there has been very little interest and the effort of making them user-friendly isn't worthwhile, but if you give me a day or two I'll have something usable.

Quoted from Shenghi:

After I read your post I hastily implemented the new model

Me too, but I'm testing it thoroughly. I wish I had my old Pentium 4 - this dual core runs the game at only 60 years per hour and its slow going. I'm trying to find a census where the number of 100-year-olds exceeds the number of surviving 90-odds. Will the excess survive as ghosts? Will the game start killing 100-year-olds? Will it do something odd? I know what I think will happen, but I want to see it in-game.

I may have to manually edit a census, but there are so many other parts of the game file tied in with it and I don't know which of those parts get their data independently (and so need synchronizing with the census) and which will simply update themselves from any new census. We'll see ...

[This message has been edited by Trium (edited 04-28-2013 @ 02:11 PM).]

Well, the answer to that question appears to be that the population will display wrongly (unless you edit that too) until it corrects itself at year-end.

I had a Fishopolis with 154 99-year-olds (the last from before the health ramp-up) and no-one in the 90s. In accordance with equi's model, all these are all killed (and so reborn). I added 20 people at 96. The game kills all those, generates 4 extra births and the total population increases by 4. Since we added 20, we seem to have lost 16.

This is what I expected since it explains equi's observations. It seems that the game calculates the number of deaths (if any) among the nineties, adds the 100-year-olds at 100% then applies those deaths to the nineties only until it runs out of people, then applies any remainder to the 100s. With ths model I'd expect rather more ghosts to form in any population than I previously thought.

Quoted from Shenghi:

It's funny how it can sometimes be harder to reproduce a bug than to implement something the correct way.

Remarkably, I was saying that to a friend yesterday. However, the new model has a couple less ifs and buts so I've been able to simplify my program.

Quoted from Brugle:

I'm still clueless. Will someone explain?

I was simply pulling his leg about why he wants the data restored when the thread doesn't interest him

[This message has been edited by Trium (edited 04-28-2013 @ 09:34 AM).]

Quoted from equi:

I found all stable curves that always have a straight line from 0 to 60 at "almost perfect" health (and that means no change in workforce, schoolchildren or academy youth).

Assuming that a "straight line" means that the numbers from 0 to 60 are exactly the same, this shows some of the stable curves but probably not all of of them. (I doubt if anyone cares whether an eternal city has an additional worker or child in some years.)

Quoted from equi:

My guess would be that no matter what the exact nature of the ghost bug is, if there is somewhat more than 10282 people in a city, then at "almost perfect" health the city won't lose many people before becoming stable.

That is reasonable, but it would help us if you described how you got that conclusion. Also, it would be nice to get an idea of "somewhat more".

Since a real eternal city might not have that "straight line", your conclusion may be more limiting than necessary.

Quoted from Trium:

the population will display wrongly (unless you edit that too) until it corrects itself at year-end

I don't remember reading that the population will correct itself (by reading the census) at year-end. Is it really at year-end or is it at the next census (a few moments later)?

Quoted from Trium:

then applies any remainder to the 100s

Since the 100s aren't displayed (or saved?), the game doesn't actually need to do this--it could just add "any remainder" to births. (It's a little easier for me to imagine the program doing that.)
[Edit--While out walking, I realized that I was making the process too complicated. The planned deaths at each stage (which are the 90's fraction plus the hundreds in one stage) simply need to be added together (to become the births), so there is nothing to do with "any remainder".]

[This message has been edited by Brugle (edited 04-28-2013 @ 09:31 PM).]

Quoted from Trium:
I generally don't release my utilities because there has been very little interest

I am certainly interested in your programs. Not only in a Pharaoh saves decompressor, but also in your program that predicts census, since it's apparently a lot better than what I can write (I don't know how to make a program read census directly from a save).

Quoted from Brugle:

Assuming that a "straight line" means that the numbers from 0 to 60 are exactly the same, this shows some of the stable curves but probably not all of of them.

Since a real eternal city might not have that "straight line", your conclusion may be more limiting than necessary.

One thing to add is that I looked only for censuses that end before 99 (that would be stable under the original model and that are independent on the exact handling of 100 year olds). However, under Trium's new model it's impossible for a curve that ends in the 90-99 group to stretch to 99 and be stable.

Quoted from Brugle:
I doubt if anyone cares whether an eternal city has an additional worker or child in some years.

I don't care about that either. It's just another way to generate stable curves. It didn't occur to me before, that I can make my program generate steady number of births and then check if the acquired curve is stable. At first I also checked if the census would always maintain the same number of newborns, but then I realized that it doesn't matter.

Quoted from Brugle:
it would help us if you described how you got that conclusion

I wouldn't call it a conclusion. It's a guess, based on another guess. The more levels of stability are right underneath, the more likely it is to stop at one of them. This is the output of the program (population, cycle).

Quoted from Brugle:
it would be nice to get an idea of "somewhat more"

I had in mind a number around a thousand, but again, these are only guesses. And the more I look at my different programs' data, the less I want to post any further of my guesses.

____________________________________________________

EDIT:

The results of the "steady births" method for 5 different health levels.
A comparison picture.

I already posted the two smallest censuses at "almost perfect" health. Now the smallest census in the table above:

4438 people, "good" health, 1 year cycle

ingame picture

53 52 51 50 4f 4e 4d 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c
4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c 4c
4a 48 46 45 44 43 42 41 40 3f 3c 39 36 33 30 2d 2a 27 24 22
1f 1c 19 16 13 11 f d b 9 8 7 6 5 4 3 2 1 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

workforce: 1318
schoolchildren: 1092
academy youth: 532

The most interesting health levels to me are "almost perfect", "good" and "average". In my future eternal cities I'll probably choose between those.
I think that "excellent" is the worst—I failed to find any stable curves for populations from 500 to 12k at this level.
I found several at "very good", but overall it doesn't seem that there is much chance for a not very large city to stop somewhere.

____________________________________________________

EDIT 2:

And the more I look at my different programs' data, the less I want to post any further of my guesses.

Who am I kidding The whole thing is too interesting for me to keep all thoughts to myself.

My model (which was proven wrong by Trium) favored the "almost perfect" health level. The last group was bigger and the curves could stretch to 99. Now it's a bit different.

"Almost perfect" health is still one of the best health levels though, but probably not the best for most cases. The overall winner seems to be the "good" health level. It's also a bit more relaxing to not see your city's census end right before 100 And there is always the worker issue—you can never have too much of those and "almost perfect" is especially unforgiving in that regard. "Good" health should work great with populations of 6k-something (like whisperwind's Mediolanum) and it rivals "average" and "almost perfect" at all other populations. "Average" might be most reliable for populations around 10700. Cities of 7k-9k don't seem to have much hope though.

[This message has been edited by equi (edited 04-29-2013 @ 11:49 PM).]

equi - I have that program but FileDen has been down a couple of days. I don't know if it's gone for good or just a temporary thing but I can't upload this thing at the moment. If you (or anyone else who's interested) drops me a line (dmiller55 live com - you know where the at and the dot go) I'll mail it to you.

Email sent

Quoted from Trium:
ww777's Mediolanum...the next 4 ghosts in 162, 189, 204 and 215 AD...produces a total 304 ghosts (plus 8 already existing) before stopping after around 13,000 years.

My program suggests that under the new model this city produces 2 less ghosts (stopping at the population 6070 instead of 6068) and the last ghost is formed after 16,506 years instead of 13,093 years. Could you confirm whether this is correct/incorrect? The 4 ghosts are predicted correctly.

To anyone who plans to use good health: there is a very big gap between "islands". A city of population between 6100 and 10500 will very likely drop to around 6070 before stopping. I suggest don't use "good" health, use "almost perfect" instead. It doesn't have such gaps

It was so only because I (incorrectly) extended the last group to 100 years. It doesn't hold true anymore—both "good" and "almost perfect" have this gap, with "good" being a little better.

Coming to think of it, ww's Mediolanum has a pretty lucky population number and health level. I hope nothing keeps it from being eternal, it's such a beautiful irony.

equi - my email keeps bouncing back undelivered, possibly because of the size of the attachment (it's 12 MB - sorry about that). It is temporarily available here. This includes various utilities including the census forecaster, so you can test for yourself whether your own program agrees

Actually, when I my asserted that ww777's Mediolanum was unaffected by the change in the model I was referring to the 115-year test run I did in-game. I should have been more specific. You are indeed correct that the new model forecasts 302 ghosts in addition to 8 already present at the save date.

Edit - link now points to my CX account. This version is improved and includes Brugle's enhancements which are mentioned in replies #96, #97 and #98.

[This message has been edited by Trium (edited 08-08-2013 @ 06:07 PM).]

Trium,

I've downloaded your stuff. Thanks very much.

When I posted that I would comment on the quality of whisperwind777's cities (and eternal cities in general) in reply #5, I didn't expect this thread to grow into its current form (with an emphasis on censuses). So when I do make those comments, I'll start a new thread.

Many thanks to Trium. Not only the programs are great, but also there is a nice description.

I'll later post a guide on what populations at what health levels are likely to become stable without losing many citizens. As I can't double post, it will probably go to this post, without a notice.

For large stable populations, the fraction of school age, academy age, and workers (assuming that there are no patricians) should be about:
at below average health, school 35.5%, academy 13.8%, workers 28.3%
at average health, school 27.2%, academy 12.7%, workers 30.6%
at good health, school 24.8%, academy 12.0%, workers 29.8%
at very good health, school 21.8%, academy 10.9%, workers 28.1%
at excellent health, school 20.5%, academy 10.2%, workers 26.3%
at almost perfect health, school 18.4%, academy 9.2%, workers 23.6%

I don't think that a large population can be stable at perfect health or at poor or worse health.

[Edit: I used a programming system from Trium, modified, to get this data.]

[This message has been edited by Brugle (edited 07-21-2013 @ 02:52 PM).]

At a given health, what city populations would form a stable cycle without making many ghosts?

I took a city, set its health, resized its census to various populations, and ran each of those until they formed a cycle (except see the next paragraph). I tested cities from 1000 to 40000 people, looking for populations that are highly likely to form cycles while making less than either 4% or 1% of their people ghosts. In some cases I repeated the test starting with a different city.

Early in my tests, I saw a cycle of 583695950 years, which formed after making only 4 ghosts (the last after 264 years). Since that single cycle took hours to find, I changed the program to stop looking for a cycle if there were over 1000000 years since the last ghost. I assumed that such a city would not make many additional ghosts.

I tried to avoid populations which might have a small chance (a few percent would be enough) of making too many ghosts, but my tests were not very precise. For example, comparing my results with equi's in reply #53, I did not count populations which are likely to land on the first 5 "islands". However, I did count some populations that are likely to stop on the 6th "island", even though some might fall through to the 5th "island", since the chance of falling through both "islands" appears to be very small.

Since my city might have achieved a ghost population quicker than another, I added 10 the the bottom end of each range. But this might not be enough (especially at better healths), so a player may want to start an "eternal" city with a population that is somewhat above the bottom of a range.

Populations very likely to form cycles with < 4% ghosts

Below Average health
4291 - 4804
5040 - 5255
5297 - 5722
9657 - 13091
13252 - 40000

Average health
10474 - 10949
11703 - 12164
12273 - 40000

Good health
4770 - 4918
4939 - 5094
5127 - 5300
5558 - 5796
5852 - 6270
10653 - 14401
14608 - 40000

Very Good health
17676 - 18305
19632 - 22550
23550 - 24429
37793 - 40000

Excellent health
23931 - 24805
25679 - 40000

Almost Perfect health
12290 - 12412
12660 - 12761
16414 - 40000

Populations very likely to form cycles with < 1% ghosts

Below Average health
4291 - 4333
4449 - 4473
4569 - 4594
4631 - 4658
5040 - 5096
5297 - 5324
5414 - 5447
5463 - 5496
5523 - 5549
9657 - 9741
9977 - 11306
11514 - 11722
11944 - 12154
12204 - 12489
12586 - 12695
13252 - 13364
13684 - 14247
14378 - 17939
18070 - 18361
18385 - 40000

Average health
10474 - 10517
10537 - 10560
11703 - 11796
12273 - 13245
13297 - 13504
13651 - 40000

Good health
4770 - 4776
5127 - 5139
5558 - 5620
5852 - 5878
5997 - 6020
6050 - 6080
10653 - 10728
10982 - 11105
11139 - 12442
12688 - 12897
13154 - 13257
13281 - 13378
13436 - 13729
13858 - 13965
14608 - 14723
15078 - 15324
15362 - 15689
15829 - 19452
19480 - 19740
19900 - 40000

Very Good health
17676 - 17750
19632 - 20024
20471 - 20725
21246 - 21588
21749 - 21795
23550 - 23689
37793 - 38208
39123 - 39899

Excellent health
23931 - 24053
25679 - 25770
25933 - 26272
26400 - 27439
27835 - 28047
28217 - 28402
28915 - 29337
29407 - 40000

Almost Perfect health (none)

I used a programming system from Trium, modified, to get this data.

Quoted from Brugle:

I used a programming system from Trium, modified, to get this data.

The modified version, wherein Brugle has added some cycle detection and other enhancements (as well as improving and optimizing some of my original code), is now available from the link in reply #93. Among other things, the program can now be used to decompress/recompress Pharaoh saves and provides some basic editing functionality so that, for example, claypit flooding can be turned off without having to use one utility to decompress the file, another to hex-edit and then recompress.

[This message has been edited by Trium (edited 07-30-2013 @ 03:44 PM).]

This is a really interesting thread. Does anyone know what the "longest running" or "oldest city" anyone has attempted in any of the city builder games?

There's many players of the games so maybe there has been more than one such save game over the years. Another criteria would be longest running or oldest "sustainable city" that managed to smoothly sustain itself for as long as possible, as some cities may have fallen apart over time or gone through various phases.

I read this article here about a guy who played the same Civilization II game for 10 years, so I wonder if someone out there has played a city builder game for a long time too, and what they observed in the game?

I did a search on the Heavengames, Tilted Mill and City Builder games forums in the general and downloads areas and managed to only find one save game for Caesar 3 here:

Brundisium - The oldest and wealthiest city (1,400,000 denari - 500 years old)

...and a comment there by goonsquad: "...I've also heard of C3 cities running until the 21st century, so you don't have the oldest or the richest."

Wonder where the other save games are or if anyone can still upload if they still have them?

[This message has been edited by alincarpetman (edited 09-30-2013 @ 05:00 PM).]

Quoted from alincarpetman:

what the "longest running" or "oldest city" anyone has attempted in any of the city builder games?

I built Iunet (a Pharaoh "family history" mission) and ran it for 3000 years (most of it without intervention) in Immortal Iunet Inequality. Unfortunately, twice in that time it failed because a bazaar did not send out its buyer when it ran low on non-food goods, but each time I backed up to a saved game and it did not fail again (at the same time). During the 3000 years it created 29 ghosts. (I am considering building another immortal Iunet with no poor houses, but with duplicate bazaars!)

goonsquad is probably describing Taracco in the year 2000.