News:

SimuTranslator
Make Simutrans speak your language.

Challenges in the transport model.

Started by Dutchman on Rails, January 27, 2018, 07:56:55 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Dutchman on Rails

As Simutrans Extended develops more into realism, Several challenges appear or become more marked that I think we should discuss. Thinking it over I came to five.

1. The transport revolution(s): The arrival of the railways, and then of the car, sparked a complete change of model in the way people went about travelling and sending goods. The railways only where they were available, the private car universal. There's a lot of complexity there.

2. Supply and demand: For a long time, the balance in transport was towards the supply side of the model. People would fit their demands to what was available. These days of course with the car, fast transport is available to nearly all, and so demand is dominant. This shift causes a complex model.

3. Scale Variety: We each have our different views on how we want to start. I take a very 'Highlands and Islands' kind of attitude, with small towns and villages. Some like big, metropolitan areas. Others have a preference in between. This causes different challenges and frustrations. In the backwater region, the challenge is for at least some demand to allow a company to start (which in my case is not helped by lowering the starting capital severely). In a metropolitan area, the challenge is the opposite, to avoid the 'Megacity planet' syndrome (virtually every square is populated), and also to avoid so much demand that the micromanagement of the cats cradle of lines becomes like painting the Forth Rail Bridge.

4. Dynamic Growth: Industries sprawl and close again, but they lack the dynamic of expansion and contraction according to demand. We now have the situation where three farms of a combined 24 production sprawl up to eventually cater for a bakery of 2. New bakeries later on will crossconnect and solve that, but in the meantime it's a bit overproduction. Again, a more complex model.

5. The Outside World: All transport simulations operate like the Galapagos Islands, no interaction beyond the map. In reality (some countries more than others), the outside world is often the dominant factor in transport. The Great Western Railway came into existence for the London-New York route. The Liverpool-Manchester Railway catered for the New World-Manchester transport. In my own country, these days the far majority of freight traffic is from the port of Rotterdam to out of the country, and several companies and railway lines came into existence and flourished on the London-Berlin route.

So how to go from here? I think most of these would be headaches to program and quite resource intensive.

jamespetts

I am not sure that I entirely understand what you mean exactly (in reference to Simutrans-Extended, at least) by some of the specific points. I will reply in so far as I am able in correspondingly numbered paragraphs below.

1. I am afraid that I am not able to parse the way in which you think that this is a challenge specifically for Simutrans-Extended. Can you elaborate?

2. Again, I am having trouble understanding what you think to be the actual implications of this in practical terms, or even really precisely what you mean by "the balance" being "towards the supply side". There were and remain constraints on supply of goods and transport services, and people can only acquire such goods and transport services as were/are available and affordable to them. Over the years, the constraints have reduced considerably in many respects, but they still exist (we cannot all have all the goods/transport that we want). The fundamental economic model of the relationship between constraints on supply and consumption is the same: it is just that the constraints, and therefore the consumption, are different in amount. I am not sure that I understand how this relates to Simutrans-Extended, however. Can you elaborate?

3. I am not sure that I fully understand what you mean by this. I imagine that there would not in reality be many transport companies starting in very small villages until such time as most of the rest of the transport market were saturated and the small village presented a niche market so far untapped. In that case, it would be easier to generate demand in any event, since transport from the small village would be to a local town, which would already have transport connexions to everywhere else. In terms of large towns, is your point that large towns grow too quickly? This can be adjusted in simuconf.tab. Additionally, the second highest priority coding project at present is re-writing the town growth algorithm.

4. The recently introduced logistics feature does scale the demand of consumer industries to local passenger demand, and producer industries will only produce such goods as are demanded by consumer industries. Additionally, for a long time now, industry growth has been directly linked to population growth. I am therefore not sure that I understand what you mean when you write that there is no dynamic expansion and contraction of industry by demand. Can you elaborate?

5. International destinations are a specifically planned feature, but there are a number of higher priority features planned to be implemented first. I do suggest that you take a look at this list of planned features.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Dutchman on Rails

I have to admit that I only have faint beginnings of ideas myself. The idea is to just dream out loud and spark a bit of discussion to see where it leads to.

I can elaborate a bit though.

1. It's not exclusively to Simutrans-Extended. Though some of the features (journey time toleration, commuting (and later on possibly migration), classes) cause a complicated interaction that also interacts with town and industry growth. In a way, a lot could be done already with having minimum tariffs for transport, differentiated journey time toleration (for commuting luxury has little or no effect on journey time toleration) and some rules for accepting purposes for transport (coaches and most boats don't take commuters). Over time, the percentage of classes visiting (or maybe demography should slowly start to 'eat away' the lower classes). Other rules like having a very long journey toleration time on foot for the lower classes for migration in competition with what transport is available, could also have an effect here.

2. I'm dreaming of a set of rules where town generation, town growth and industry generation and growth happen based for a large part on the availability and desirability of (potential) transport links (plus the likelihood that such a transport link will be created), and in part on the unemployment, available jobs and some other factors (like a negative modifier for having a bustling rich town right next door, which could be offset by very high desirable links for commuting services). A town completely devoid of rail links somewhere landlocked (unless we've reached the 1960's and it finds itself on a nice 50mph road free of traffic jams) being at a severe disadvantage to one with excellent rail links and on the coast. Lots of modifiers to make it work (for example there need to be dynamic correcting modifiers to prevent ghost towns (in case of negative growth) or runaway megacities). I see by the way that you're dreaming in the same direction.

Another thing I'm dreaming of is that passengers (already looking for alternative destinations) go a step further and first look for available transport options (including private ones) and have a high chance of going somewhere (people just visiting a church/market that happens to be in reach rather than a specific one, though that happens too).

3. This is a balancing thing. Basically I think it's one of the simplest to solve. If we find out that a certain setting causes negligible transport on maps with a very low population density, rather than increasing the number of passengers/mail overall, a dynamic modifier increase the number of visitors and mail somewhat in relation to the low population density so that you have something to transport. Not much, just enough to help the startups. Over time, as income grows, this modifier is gradually and proportionally reduced until it reaches 0 (OK, we've given you a jumpstart, now you're on your own). On the other extreme, a dynamic modifier to slow down town growth (or perhaps increase renovation percentage) in very high population density areas could help prevent the effect of getting a megacity map. Also, on very high population density maps, a dynamic modifier could reduce passenger and mail generation by a percentage (to avoid having such a wild number of passengers around that you get crazy adapting the network around it).

4. What I'm dreaming is two steps more:
- Industries spawn on locations based on the desirability of (potential) transport links, but also the number of available labor (and in case of farms/mines landscape features), and in turn the existence of industries has a positive effect on town growth (lowering unemployment).
- Industries can not just spawn and close, but also increase and decrease production based on the demand for goods. When an industry spawns, it connects to a consumer and spawns with enough production to meet the contract. When another consumer for a good that this industry can deliver spawns, three things can happen:
a. The new consumer crossconnects with the existing industry. This chance is based on desirability of (potential) transport links, distance in relation to perishability of goods and the potential for cooling and other methods to protect against perishing of goods, and the total production the industry already has in relation to the optimal production. When this happens, the supplier increases production accordingly, which in turn causes demand for workers and goods to also increase. Likewise the closing of a consumer causes an industry to decrease production accordingly.
b. The industry crossconnects with a portal, same modifiers except the total production (there is no limit to optimal production).
c. A totally new supplier spawns.

5. I've read the part on portals. We dream in the same direction. Portals will be very interesting, being special towns, special industries and special stops, all in one.

jamespetts

Quote from: Dutchman on Rails on January 27, 2018, 09:09:45 PM
I have to admit that I only have faint beginnings of ideas myself. The idea is to just dream out loud and spark a bit of discussion to see where it leads to.

I can elaborate a bit though.

1. It's not exclusively to Simutrans-Extended. Though some of the features (journey time toleration, commuting (and later on possibly migration), classes) cause a complicated interaction that also interacts with town and industry growth. In a way, a lot could be done already with having minimum tariffs for transport, differentiated journey time toleration (for commuting luxury has little or no effect on journey time toleration) and some rules for accepting purposes for transport (coaches and most boats don't take commuters). Over time, the percentage of classes visiting (or maybe demography should slowly start to 'eat away' the lower classes). Other rules like having a very long journey toleration time on foot for the lower classes for migration in competition with what transport is available, could also have an effect here.

I am still having real trouble in understanding what you mean. Are you suggesting a new feature? Are you referring to a balance problem with the game as it stands? I am afraid that I am not at all clear.

Quote2. I'm dreaming of a set of rules where town generation, town growth and industry generation and growth happen based for a large part on the availability and desirability of (potential) transport links (plus the likelihood that such a transport link will be created), and in part on the unemployment, available jobs and some other factors (like a negative modifier for having a bustling rich town right next door, which could be offset by very high desirable links for commuting services). A town completely devoid of rail links somewhere landlocked (unless we've reached the 1960's and it finds itself on a nice 50mph road free of traffic jams) being at a severe disadvantage to one with excellent rail links and on the coast. Lots of modifiers to make it work (for example there need to be dynamic correcting modifiers to prevent ghost towns (in case of negative growth) or runaway megacities). I see by the way that you're dreaming in the same direction.

Another thing I'm dreaming of is that passengers (already looking for alternative destinations) go a step further and first look for available transport options (including private ones) and have a high chance of going somewhere (people just visiting a church/market that happens to be in reach rather than a specific one, though that happens too).

As you note, there are already some very detailed plans for town growth based on local transport, and there are entire threads dedicated to this topic that go into it in very great detail. I do suggest that you look at those (here for instance) to see precisely what is planned. There are, of course, a number of very specific limitations to what can be achieved based on computational limitations (e.g., it is not feasible with current technology to calculate whether any given road route between towns is congested with private car traffic, at least not without a fantastically complex system that may take many years to implement).

Quote3. This is a balancing thing. Basically I think it's one of the simplest to solve. If we find out that a certain setting causes negligible transport on maps with a very low population density, rather than increasing the number of passengers/mail overall, a dynamic modifier increase the number of visitors and mail somewhat in relation to the low population density so that you have something to transport. Not much, just enough to help the startups. Over time, as income grows, this modifier is gradually and proportionally reduced until it reaches 0 (OK, we've given you a jumpstart, now you're on your own). On the other extreme, a dynamic modifier to slow down town growth (or perhaps increase renovation percentage) in very high population density areas could help prevent the effect of getting a megacity map. Also, on very high population density maps, a dynamic modifier could reduce passenger and mail generation by a percentage (to avoid having such a wild number of passengers around that you get crazy adapting the network around it).

I do not understand this issue. Is there a problem with the balance of passenger generation? What real world dynamic would this be simulating? I do not understand the idea of increasing passenger numbers beyond realistic levels just to help players to make a profit - in what is intended to be a realistic simulation of real life transport, this makes no sense: a route that would be unprofitable in reality should also be unprofitable in Simutrans-Extended.

Again, I do not understand what a "megacity map" is. Can you elaborate? Do you mean that town growth for larger towns is too great? As noted in the previous post, this can be adjusted.

Quote4. What I'm dreaming is two steps more:
- Industries spawn on locations based on the desirability of (potential) transport links, but also the number of available labor (and in case of farms/mines landscape features), and in turn the existence of industries has a positive effect on town growth (lowering unemployment).
- Industries can not just spawn and close, but also increase and decrease production based on the demand for goods. When an industry spawns, it connects to a consumer and spawns with enough production to meet the contract. When another consumer for a good that this industry can deliver spawns, three things can happen:
a. The new consumer crossconnects with the existing industry. This chance is based on desirability of (potential) transport links, distance in relation to perishability of goods and the potential for cooling and other methods to protect against perishing of goods, and the total production the industry already has in relation to the optimal production. When this happens, the supplier increases production accordingly, which in turn causes demand for workers and goods to also increase. Likewise the closing of a consumer causes an industry to decrease production accordingly.
b. The industry crossconnects with a portal, same modifiers except the total production (there is no limit to optimal production).
c. A totally new supplier spawns.

As noted, it is already planned to have industries and city buildings spawn based on local transport quality (what you refer to as unemployment would be represented as a low commuting passenger success percentage, and these figures are already stored and displayed in the game).

Quote5. I've read the part on portals. We dream in the same direction. Portals will be very interesting, being special towns, special industries and special stops, all in one.

Given the other priorities, this may be some time away yet, but it will be a good thing to have.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Dutchman on Rails

1. I'm dreaming of a whole range of features, new or existing:
- Minimum tariffs for some vehicles.
- Commuters do not benefit from comfort, while visitors do.
- Negative population growth for towns in some circumstances.
- Population growth can spawn passengers (migration).
- Percentages of commuters/visitors/migrants can vary by class.
- Percentages of commuters/visitors/migrants can vary over time.
- Highest class having a minumum comfort level.

2. We'll see how it goes then. Two things:
a. Not just local transport, but also and perhaps especially interlocal and international transport.
b. Not just real transport, but also potential (in the 18th century, access to a sea that contains a portal, or any navigable river flowing into a sea like that would be a great boon, in the 19th/20th century only the sea itself or a major river would still be a big boon, and smaller navigable rivers just a smaller one).

3. Try creating a 1500x600 map with some 40 towns and villages, but none more than 600 and most in the 100-400 range. Start in 1750, and with a starting capital of no more than 5000. Now you will see one of two things happening. Either the initial start is so difficult you never get off the ground due to insufficient transport, or this is doable, but you get overcrowding and unrealistically high numbers of passengers and mail once you've covered more than about 30-40% of the towns. So it's this initial start, in extreme conditions, that I'm after.
Come to think of it, portals may be the cure for this, as it brings an influx of supply/demand from the outside world that is likely to be profitable even for a small market/mining town in a backwater region.

On the other extreme, a Megacity map is where the map is covered 90+% by towns and industry. Given that towns never stop growing, if playing long enough, this point is inevitably reached.

4. Here we have a miscommunication. I know of the conditional industry spawning. However, once an industry spawned, its production is fixed, right? I'm dreaming of it being flexible, and based on the number of crossconnecting contracts it gets.

jamespetts

Thank you for clarifying. There is a huge range of topics in this thread, each of which could in theory be discussed in very great detail. However, there are other priorities at present, so I will have to keep my responses brief.

Quote from: Dutchman on Rails on January 28, 2018, 07:24:58 AM
1. I'm dreaming of a whole range of features, new or existing:
- Minimum tariffs for some vehicles.

I am not sure that I understand what you mean by this. Can you elaborate?

Quote
- Commuters do not benefit from comfort, while visitors do.

I do not understand the purpose of this suggestion. What exactly do you mean by "benefit from" here? Why should the purpose of the trip make such a binary difference to the relevance of comfort?

Quote
- Negative population growth for towns in some circumstances.

This is planned for the town growth overhaul.

Quote
- Population growth can spawn passengers (migration).

This is under consideration for the town growth overhaul and/or international destinations, but might be too difficult to implement for a relatively modest benefit.

Quote
- Percentages of commuters/visitors/migrants can vary by class.
- Percentages of commuters/visitors/migrants can vary over time.

What exactly would this be simulating? I suspect that the intention of this feature is to act as a proxy for something that is already simulated directly (viz. that passengers of lower wealth have fewer transport options), and so simulating it indirectly like this would make the simulation less coherent and realistic.

Quote
- Highest class having a minumum comfort level.

I do not understand what this would be simulating: if the only transport options that any given person faces are an uncomfortable journey or not making the journey at all, why would a person with more wealth refuse to travel at all while a person of less wealth would make the trip? This issue has been discussed before.

Also, this would be impossible to calibrate: stagecoaches of old (and likewise, the private horse coaches of the very wealthy) were much less comfortable than third class rail travel of the Edwardian era or even trams of the 1930s. It makes no sense to have any specific relationship between any given level of absolute comfort and wealth.

Quote
2. We'll see how it goes then. Two things:
a. Not just local transport, but also and perhaps especially interlocal and international transport.

My apologies, I think that what I meant by "local transport quality" was unclear: I meant not the quality of transport to local destinations, but rather the local quality of transport generally, including to remote locations. In other words, the town growth algorithm will use statistics from nearby buildings as to what proportion of passengers generated by those buildings (if residential) have been able to travel to their intended destinations, and what proportion of the visitor and commuter demand of those buildings (if non-residential) have been fulfilled in recent times.

Quote
b. Not just real transport, but also potential (in the 18th century, access to a sea that contains a portal, or any navigable river flowing into a sea like that would be a great boon, in the 19th/20th century only the sea itself or a major river would still be a big boon, and smaller navigable rivers just a smaller one).

I am afraid that I cannot see any sensible way of predicting what the transport quality in the future might be, nor do I think that this is how people behave in reality, except when there is a specific transport project already planned or under construction.

Quote
3. Try creating a 1500x600 map with some 40 towns and villages, but none more than 600 and most in the 100-400 range. Start in 1750, and with a starting capital of no more than 5000. Now you will see one of two things happening. Either the initial start is so difficult you never get off the ground due to insufficient transport, or this is doable, but you get overcrowding and unrealistically high numbers of passengers and mail once you've covered more than about 30-40% of the towns. So it's this initial start, in extreme conditions, that I'm after.
Come to think of it, portals may be the cure for this, as it brings an influx of supply/demand from the outside world that is likely to be profitable even for a small market/mining town in a backwater region.

If you are getting an unrealistically high number of passengers in the early era, I suspect that you are setting your classes too low. If you use the default classes with the stagecoaches, you will have a more realistic number of passengers (as those are the more realistic classes/prices for stagecoach travel).

QuoteOn the other extreme, a Megacity map is where the map is covered 90+% by towns and industry. Given that towns never stop growing, if playing long enough, this point is inevitably reached.

This suggests that town growth is at too high a rate. You can adjust town growth in simuconf.tab. The rate of town growth will be looked into when the town growth code is overhauled.

Quote
4. Here we have a miscommunication. I know of the conditional industry spawning. However, once an industry spawned, its production is fixed, right? I'm dreaming of it being flexible, and based on the number of crossconnecting contracts it gets.

We already have the industry boost system for industries, so its production is not fixed once spawned. This may well be modified somewhat when the town growth system is overhauled, but I have not considered the specifics of this yet.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Kirill Stepanoff

Quote from: Dutchman on Rails on January 27, 2018, 07:56:55 AM

4. Dynamic Growth: Industries sprawl and close again, but they lack the dynamic of expansion and contraction according to demand. We now have the situation where three farms of a combined 24 production sprawl up to eventually cater for a bakery of 2. New bakeries later on will crossconnect and solve that, but in the meantime it's a bit overproduction. Again, a more complex model.

I have just got more than 80 (yes, EIGHTY) grainfarms supplying 1 grainmill for 1 bakery  :o They had littered about half of 256*256 map.