Passenger Generation Anomalies

[Carl]

In short: the ratio of passengers generated to population of a town seems to vary wildly and unpredictably. This is summed up in the shot below:



I take it that the "Passengers" value here is not dependent on the town's network connections. It shows a positive and constant value even in towns which are not connected to the network, and does not change upon connection to the network.* As such one would expect this value to co-vary with the population of a town. But in fact, Kál (population 528) is producing almost as many passengers as Gyongyos (population 3714). Something seems wrong here.

Here's the savegame and pak folder:
https://dl.dropbox.com/u/61716/PassGenAnom.rar

All the stations in the shot above can be found nearby on the same line.

*The downward trend in Gyongos and Mezokovesd's passengers line is a result of a recent change in passenger factor. Kál has only just been built, which may - or may not - be relevant to this bug.

[greenling]

Carl gives in Kál,Gyöngyös,Mezokövesd buildings that tourist like?

[Carl]

No, there are no tourist attractions in any of those towns, so that can't be the problem.

[jamespetts]

This is interesting, since I have not changed the basic code that relates the city's population to passenger generation from Standard, although a number of passenger generation things are different from Standard that might cause this result. However, it is very possible that this same effect exists in Standard, especially since, as recorded here the level of a building has a somewhat capricious relationship with the number of passengers generated by it owing to a somewhat crude formula used to reduce the overall numbers of passengers generated introduced many years ago when the concept of return passengers was introduced.

I am looking into passenger generation more generally, but, in the meantime, can you check to see whether you can find similar anomalies in Standard? That would be most helpful.

[Carl]

Thanks for the reply, James. Frustratingly, I cannot even reproduce this in another auto-generated Experimental game: I created a couple of large maps and the cities in them had perfectly linear passenger generation distributions.

This reminds me of one key fact I forgot to mention about the map in which this error occurred. The map was generated with no cities at all, and all cities and buildings have been subsequently placed by hand. I don't know how this could have such a strange effect, but it seems to be the main relevant difference between the auto-generated maps and the problematic map.

I've done some experimenting with another empty map. Cities placed initially after map creation will usually respect a linear distribution -- but those placed in later months will not always do so. That is, later placed cities will often have odd passenger generation distributions. (This chimes with the fact that, in the above map, Kál is a very recently placed town). So perhaps new city placement is the culprit here. Furthermore, I seem to be unable to reproduce this latter behaviour in Standard, where newly added cities will respect the linear distribution.


Note that I can only reproduce this on maps which were initially generated with zero cities.

[Carl]

On reflection I have had a thought: perhaps this is related to an old bug where some cities would display over 100% of passsengers transported:
http://forum.simutrans.com/index.php?topic=7781.0

That particular bug doesn't occur any more, but a couple of the cities in old maps where I used to observe that phenomena are now generating disproportionately high levels of passengers for their population size.