投稿者 トピック: [nightly] locality_factorのデフォルト値について  (参照数 9547 回)

z9999+

  • 管理人
  • 急行
  • *
  • 投稿: 381
簡単な実験の結果、一般的なマップサイズでは、locality_factorに100以上の値を設定すると、旅客は従来のバージョンよりも遠くに行くという事がわかりました。
これは今まで以上に駅の混雑を引き起こし、また必要以上の収入が得られてしまい、結果としてゲームを面白くする事にはなりません。
デフォルトのsimuconf.tabの設定では、1958年にはすでに100を超え、2004年には1000を超えてしまいます。

もちろんプレイヤーがどんな値を設定しようが、その人の自由ですが、少なくともデフォルトの設定は、旅客の行き先をうまくコントロールすべきです。1980年や2000年から始めたプレイヤーは最初から行き詰ってしまいます。



locality_factor more than 100 are not useful for general size games.
Passengers go further places than old stable versions, and it only causes heavy overcrowded and born too many income. It doesn't make game fun.
It is really far from player's demand.

コード: [選択]
locality_factor[0]=1835,1
locality_factor[1]=2050,10000

On this default setting above, it will reach 100 in 1958, it will reach 500 in 1991, it will reach 1000 in 2004,

YearFactor
1958101
1972201
1991524
20041003

Okey, anyone can use any factor value, but at least default value should control passenger's destination well for regular map size.
A player who start a game in 1980 or in 2000 will stuck from the beginning.

Note: I tested both koord_distance and shortest_distance to make graph, and result was the same. So, I only used koord_distance for graph.
« 最終編集: 2011/08/26 17:56 by z9999+ »

prissi

  • 各駅停車
  • *
  • 投稿: 45
level and size matters. Largest city gets most in any version. One other possibility was discussed: locality increase with average speed.

z9999+

  • 管理人
  • 急行
  • *
  • 投稿: 381
Any locality factor increasing doesn't need. When cities growth, number of passengers increase by itself.
You should not mix possibilities and probabilities.
At least simutrans-experimental treats these two things separately.

In simutrans, as long as possibilities is not zero, people will go there if a player makes a route to there.
But in real life, most of people doesn't go farther places every day. Most of people favor nearer places.
This is true in old days and now-a-days.
If there are two nearly equal cities, most people will go nearer one.

Passenger destination system of simutrans standard is mostly affected by the number of cities. Players want to start a game with many cities because it will not increase automatically in simutrans. And it makes harder to play a game, because passengers don't favor nearer places but large cities regardless of distance.
So, it requires low locality factor or a kind of zoning system which exists in simutrans-experimental to limit influence on the number of cities.

Amount of passengers is very sensitive. If 500 passengers arrived and 500 passengers left in an hour, the station will not be overcrowded, but if 501 passengers arrived, the station will be overcrowded soon or later.
Short distance travellers will reach their destination soon and disappear from game, but long distance travellers occupy stations and vehicles for a long time. If many passengers favor large cities regardless of distance, it inevitably generate overcrowded.
So, a drastic change of locality factor doesn't need.
The game should control it well.

Some players call simutrans an eliminating congestion game.
They want decrease amount of passengers in the late period of the game.
Moderate overcrowd makes game much fun and worth challenging, but excessive overcrowd is only a pain for the players.
« 最終編集: 2011/08/28 17:04 by z9999+ »

z9999+

  • 管理人
  • 急行
  • *
  • 投稿: 381
level and size matters.

I think this is not true statistically.
It is proportional to area of each distance.
So if factor=10000, in square map, many passengers go to middle distance area because it is large.
In slender map, area of each distance are mostly the same, and passenger destination will be longtailed.

Knightly

  • 各駅停車
  • *
  • 投稿: 4
Sorry for digging up this rather old thread, but seems that the default (Year 2050, Locality Factor 10000) has not been changed yet.

I strongly agree with Z9999+ that people, no matter in the past or nowadays, mostly travel to nearer places. This is obvious -- no matter how fast transportation becomes, it costs to travel far, in terms of time and money, unless in the future people can go anywhere by teleportation free-of-charge, but that is apparently unlikely. From an economic point of view, both time and money are scarce resources to a human being, thus if a nearer substitute destination can be found, they will usually prefer nearer places, except tourists which might have a stronger tendency to prefer attractive places and care less about distance.

One important goal of preference for nearer destinations is to avoid insane overcrowding and massive profits in late games. A locality factor of 10000 basically means that distance matters no more -- but can that possibly be true in reality? For large maps with numerous cities, the overcrowding could become out of control -- this is not a challenge, but a headache to players.

If some people prefer such insane overcrowding in late games, fine, they can set locality factor to 10000. But if default is also 10000, this will only scare away the new players.
« 最終編集: 2011/10/31 13:36 by Knightly »

Knightly

  • 各駅停車
  • *
  • 投稿: 4
Prissi,

In the new stable 111.0 , locality_factor[2].year is 0 and locality_factor[2].factor is back to 1 -- I suppose it's just an oversight?

Besides, I still don't quite understand why, in 2100, distance nearly matters no more (given the default 10000 locality factor) for reasonably sized maps (e.g. 512 x 512 or 1024 x 1024). Is that really conceivable, from a scientific point of view, that people will not care about the travel time and monetary costs involved in long distance travels in 2100? Anyway, I will say no more, and leave it to you to make the final judgment.

z9999+

  • 管理人
  • 急行
  • *
  • 投稿: 381
In the new stable 111.0 , locality_factor[2].year is 0 and locality_factor[2].factor is back to 1 -- I suppose it's just an oversight?

locality_factor[2] is missing in simuconf.tab.

\pak\config\simuconf.tab
コード: [選択]
locality_factor[0]=1835,1
locality_factor[1]=1900,5
locality_factor[3]=2100,10000

prissi

  • 各駅停車
  • *
  • 投稿: 45
The train are so fast in 2050 (more than 400) that such distances become playable. It is, as always, a balance between the old (everything same) and the new approach.

Thus I tried to edit the values (and entered also the typo, thanks!)