The International Simutrans Forum

 

Author Topic: Year 1750 - passengers won't travel - route always too slow  (Read 12345 times)

0 Members and 1 Guest are viewing this topic.

Offline Jando

  • *
  • Posts: 505
Re: Year 1750 - passengers won't travel - route always too slow
« Reply #35 on: August 22, 2013, 02:30:13 PM »
Thanks, James.

If I may, I have a request when talking about a possible design issue with these spikes. In another post you wrote that waiting times are averaged - and I assume that the waiting times during a spike are part of that average. I reckon that explains why I see waiting times in the detail screen of a stop that are much, much longer than the factual waiting time, up to 5 times as long from what I've seen - because the very long waiting time immediately after a route is started counts to calculate the averaged time.

I think the stated waiting time would be much more correct if no average would be calculated at all (or maybe only over the last, say, 3 or 4 journeys) instead of calculating an average over several months. Maybe you can consider the effect of averaged waiting times for your next major release.

Thank you in advance. 

Offline jamespetts gb

  • Simutrans-Extended project coordinator
  • Moderator
  • *
  • Posts: 18745
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Year 1750 - passengers won't travel - route always too slow
« Reply #36 on: August 23, 2013, 12:09:18 AM »
The difficulty is what to do other than average waiting times. Averaging over a much shorter period as you suggest also has its problems. The paths are only calculated occasionally, so the last three or four waiting times might be completely unrepresentative of the time since the last calculation (they might be during a spike, for example). Calculating the paths more often would make the game run too slowly. Even if the paths were calculated more often, having the waiting times based on the last 3-4 waiting times only would lead to erratic behaviour. Further, I do not think that using the last 3-4 waiting times would actually affect the spikes that you see at all: the behaviour would be just the same.

Initially, the waiting time is presumed to be about 2 minutes as a default. If this plus the journey time gives an appealing travelling time for passengers, many passengers will travel. If this results in high waiting times because capacity is inadequate, the waiting time will go up and the journey will, on the next recalculation, be less appealing to passengers, so fewer people will travel. The averages are reset whenever the paths are recalculated. It thus does not matter whether we take the average of all waiting times or the last 3-4 for the purposes of the spikes: what causes the spikes is that the number of people waiting to travel now is based on the total journey time of the journey when the paths were last recalculated some time ago.

The only thing that I can think will help, short of the sadly inadmisslbe solution of much more frequent recalculation of paths, is having a greater range of journey time tolerances for any given journey distance (by abolishing distance categories), which should help passengers to reach more of an equilibrium.