AI keeps replacing its engines

[letsdance]

i just noticed, that after a while (70+ years) in my map, the AI companies replace all their engines in every january or february. i can see their profit dropping from 0 (or close) to -5 mio in an instant,and it's shown in their balance sheet as engine maintenance. i checked by doing mergers, then reloading and doing the merger at another point of time. right at the start of each year, the age of all their trains is 1 year, but after i see the drop in profit, the age of all their trains is brand new.

is that a known problem or does anyone know how to prevent it? it looks like they are not issuing bonds for that, and they stop when they are out of money AND have negative profit. but as soon as they make some profit, they keep replacing their engines again, even if they are already in debts (money negative). this way they are killing themselves.

it seems not all companies are affected, only those who expanded. i can just reduce AI engine cost as a workaround, but it's not satisfying. i set my start date to 1960 (instead of 1830) and after a few years it seems the same was happening also in this game (but i didn't check). i don't think i changed any settings that should encourage such a behaviour.

[letsdance]

maybe the AI is replacing engines if they are inefficient? (maintenance + fuel > revenues) i improved AI revenues and it seems they stopped replacing engines every year in the last test game. before, they lived from interest mostly, while their trains were not profitable (in the later game).

Edited August 7, 2016 by letsdance

[jeffryfisher]

This is a long-standing bug in the game. The AI is testing train age instead of engine age. When it replaces the engine, the train continues getting older, so the engine gets replaced again and again. This bug is listed in the wish-list thread.

[letsdance]

if the problem with engine replacement is train age, should it be possible to retire all AI trains every 10 years to prevent that bug? maybe i'll try that.

[MaglevForever]

Good idea. 25 years seems to be the frequency needed.

 

This is what Jeffry said in the "Open RT2 wish lists thread"

On 20/03/2012 at 0:04 PM, jeffryfisher said:

* AI mistakenly uses train age instead of engine age when deciding to replace its trans' engines. The result is that after about 25 years, every AI company will attempt to replace every engine every year until the company enters a debt spiral to bankruptcy.

 

PS. I believe I tried an early version of your map. Then I wasn't sure how much was changed with the successive versions. From the list you made, it looks like you've been busy. Going to play the latest version soon. Will let you know how it goes.

[letsdance]

yes v6 has had more fixes and changes than expected. the problem with retiring AI trains is that the editor does not support mass-retiring and i don't want to add 100 events for that. but it could be possible to hex-edit the event in the map to achieve the desired result. gonna look into that, sometimes.

[MaglevForever]

Did you try making the engine price cheap for the AI after a certain number of years has passed? Because engine price affects engine maintenance cost an adjustment needs to be made for that. I didn't try it, but for -90% engine cost, it may be as simple as +90% on engine maintenance cost.

 

-90% isn't a recommendation, just an example. I didn't test it. I thought of replacement every ten years. This is very generous because the AI engines always have brand new running costs.

 

Would try it in my game as replacements have started in 1864, but I think that having more profitable AI is going to make the PNW target easier.

[letsdance]

yes the AI gets engine cost reductions later on to keep them from going bankrupt, or at least from going bankrupt too fast  sometimes an AI survives til the end (in test games), but usually you merge the big ones anyways. i did not counterbalance the reduced cost with higher upkeep. the yearly engine replacements work as upkeep anyways. but it's an unsolved issue in v6. it's not breaking the map, because in the late game you play mostly against the car shipping and value goals. but it means the AI companies usually struggle for survival in the 20th century. in my current game i seem to get dividends from them, even though they are struggling for survival. i usually don't pay much attention to that because i don't need the money anyways and i expect to get the PNW from shares value, assuming that the company reaches the book value goal.

[letsdance]

unfortunately, even with hex editing, it is not possible to retire multiple trains with a single event. i'll solve it by further lowering engine purchase price for the AI.

[Raleigh98]

I have a nice easy solution that is slightly cheating but it does solve the problem. Use the cheat Casey Jones every 25 years, or when you see this is happening. The AI's trains will die, and they should get new ones that don't replace.

Of Course it is cheating, but there isn't much else you can do without editing the hex codes. Maybe that cheat code was purely designed for this problem.

 

CASEY JONES

[jeffryfisher]

Hmmm... The lesser of evils. Better to lose the hardware (and freight) once each 24 years than to lose the hardware every year. The sweep would also unclog the rails, especially on mountainsides.

[immpy]

....always try to think "outside the box"....I just KNEW there could be a good use for a cheat code....sharp observation.....

Let's think of some uses for the other cheat codes (to improve gameplay of course).

[jeffryfisher]

The primary purpose of cheat codes is to test the game and its maps. I used them from time to time when modding the EXE, fiddling with the file of text snippets, proving undocumented rules for editor's events and settings, and developing my US History map. In other words, if you're not actually playing then it's not cheating to short-cut to a condition that you need to test (and that's why virtually all games have those codes -- because the game maker needed them, and one never knows when the last patch or the last add-on scenario will need them too).