Minimum transfer time

[Carl]

A couple of thoughts about the minimum transfer time feature in the RC. I really like this feature, but I wonder if it needs tweaking. Here are two points that I'm not fully convinced of, but which merit discussion.

First thought. Two stations which differ only in platform length should arguably not differ in minimum transfer time. That is, a 2x2 station and a 2x8 station shouldn't really take any more time to change at. So, arguably, only the smaller of the two dimensions should count for minimum transfer time (so a 3x6 would have a longer transfer time than a 2x6, but a 2x8 wouldn't have a longer time than a 2x6). However, one difficulty with this is that the game has no way to tell whether the 2x8 station is two platforms of 8-tile length, or 8 platforms of two-tile length. And an 8-platform station should certainly have a greater transfer time than a 2-platform station. So this is tricky.

Second thought. Minimum transfer time should not arguably not vary with distance-per-tile, because track width does not respect scale changes. When we build a two-platform station at 250m-per-tile, we are not *really* building a 500m-wide station. Track layouts are not always sensitive to mpt changes in this way. The thought here: there should be a fixed minimum transfer time for 2-platform, 4-tile length stations, and there's no reason that this should be diffferent depending on the mpt setting -- after all, the displayed length of trains doesn't change depending on mpt either.


Any thoughts?

[Junna]

and there's no reason that this should be diffferent depending on the mpt setting -- after all, the displayed length of trains doesn't change depending on mpt either.

How would this be achieved, I mean, with a station that is very wide? Say 20 platforms or so? Does it require a specific additional time to be programmed for all possible widths, or would it be achieved elsewhere? I also had some of the same concerns as you, otherwise.

[Carl]

I was thinking that it could be achieved algorithmically, rather than having to specify values for each possible station size.

[jamespetts]

This is a not entirely straightforward topic. One of the essential aims of this feature was to make it not matter whether any given tile was a station tile or not for the purpose of calculating walking times, so that one cannot magically shrink walking times by making stations larger (as one can, in effect, in the current release version and Standard by concatenating a large number of 'bus stops in a town, making a single stop cover the whole town). If we have a different scale for tiles within a station than tiles outside a station, there will be a perverse incentive to build larger stations in order to reduce passenger walking times.

As to the computation of the transfer time, this is not straightforward, as there many different types of stops, all of which have to share the same algorithm - 'bus stops, coach terminals, airports, docks, railway stations and stops that are some amalgam of two or more of these all have to be considered, and not all will have the same concept of platform length to deal with.

[Carl]

Thanks for the reply. I appreciate that there are difficult issues here, and it may be that the current implementation is the least worst option.

[sdog]

The issue here is i think that stations serve two conceptually different tasks. Collecting passengers and freight, and loading and unloading of convoys. For the former it is very sensible to use the metre-per-tile directly, disregarding platform configurations. Afterall, the player benefits from the 2-dimensional extension of the station.

The main problems arise when passengers are transfering. Here the scale are completely out of proportion. The dimensional extension of the station is not a benefit, only a side-effect of providing enough berths for convoys.

A consequence would be to use different transfer time concepts for different situations. For the first and last leg of a journey: From origin and destination to the convoy a distance based, metre-per-tile scaled, transfer time as an extension of travel time to stations ought to be used. For transfers from one convoy to another, a transfer time based on a quantity to express the stations complexity and extension could be used.

One approach would be to use a different scale factor but the same system. I'd rather suggest to use the station capacity and a reasonable fixed factor as a measure of its complexity and thus transfer times.

Reason for this is: It is, for most passengers, not so much the time one takes to walk, but also to quite some extend the time needed for orientation and information, that determines the time needed  when changing vehicles

[jamespetts]

Hmm - an interesting idea, but an impractical one, I think: it would be quite incomprehensible to most players.  It strikes me also as the sort of thing that can create perverse incentives: if transfers count as less distance than anything else, then there will be insufficient incentive to have few transfers: small stops might be placed very near a large stop with the express intention of attracting people to the large stop by means of going via the small stops and thereby reducing transfer time. There may well be other distorting effects that I have yet to consider, but this sort of internal inconsistency always brings the danger of distortions.

[sdog]

Hmm - an interesting idea, but an impractical one, I think: it would be quite incomprehensible to most players.

Yes, this is a very big disadvantage of that concept.

It strikes me also as the sort of thing that can create perverse incentives: [...] small stops might be placed very near a large stop with the express intention of attracting people to the large stop by means of going via the small stops and thereby reducing transfer time. There may well be other distorting effects that I have yet to consider, but this sort of internal inconsistency always brings the danger of distortions.

I think you can't, and don't need to try to, prevent all types of abuse. Of course it would be possible to have a tiny stop A, take a bus let it run 2 tiles to huge stop B to save on transfer time. But this is so silly and clearly abusive it could be sanctioned in online games. It is not a direct game incentive (as in economic system of the game) but only one that could be gained through exploiting loopholes.

if transfers count as less distance than anything else, then there will be insufficient incentive to have few transfers:

I don't think so. Having more transfers is still disadvantageous enough. There is the limited number of hops, waiting times due to timetables, and of course still transfer time (while greatly reduced).
Large traffic hubs would be at less of a disadvantage as they currently are to be. But that is perhaps what Carl would consider an improvement.

[jamespetts]

Hmm - but a difficulty is that players might unintentionally create a situation as described with small and large stops, and then find it very odd the passengers are all going to the small stops...

[sdog]

Yes, that would lead to unpredicted behaviour. It would even get worse as the world tile scale is increased. Making it very difficult to explain in response to help request or bug reports.

My suggestion is not fit for the task. A few km of walk to change trains would be rather inconvenient too. I'll keep in the discussion, interested if there's a good sollution comming.

(perhaps a wait time factor decaying with distance from origin or destination, with a fixed minium wait. not very elegant and very ad-hoc, which i don't like in simulations)

[jamespetts]

This discussion does rather emphasise the advantages of a larger scale. As you may know, Pak128.Britain-Ex is moving to 125m/tile for the next release, which should help matters.

[ӔO]

This might give players incentives to build efficient hub stations or create schedules with better connections, which reflects real life.

But I find scheduling to be quite tedious and time taking to get right. What's worse, you need to redo them completely when you replace the trains.


From my experience, you should be able to serve, a minimum of, 1 line per platform at the station.