----------------------------------------------------------------------------------------
Ok, cabooses. I use them all the time. They seem to work for me. I've tried running trains with them and without them. I haven't got any hard and fast figures all tabled up (might run some comparisons when I have time) but I always seem to have more reliable trains with them. This may be partly because they aren't as heavy as a loaded freight car. Any reduction in load is going to help reliability a bit, so hauling 7 freight cars and a caboose would be a bit easier on the locomotive than hauling 8 freight cars.
That doesn't seem to be a complete explanation, though. I usually run "any cargo", so it's common to have express and freight mixed, and I still seem to get some benefit from the caboose. I have noticed that the reliability bar shows a marked improvement when the train is loading and the caboose goes on. It seems to be in line with the stated 50% reduction in breakdown probablilty. I have also noticed that if I run a new train with 8 cars and no caboose, its reliability bar deteriorates a lot faster. This is not reversed by later maintenance or changes in consist. It seems that once a loco is stressed to the point of becoming less reliable there is nothing you can do to fix it (apart from replacement).
I'm wondering if some people may be a tad confused by the stated probability of breakdown and the results shown in the game. Tossing a coin is a good analogy. A lot of people think that if you have tossed a coin four times and got heads all four times, the chance of getting tails with the next toss must be greater. It isn't greater, though. It's still exactly the same: 50%. Probability only tells you how many times you should get a certain result if you do an infinite number of trials. Tossing a coin one million times and getting all heads is perfectly possible (although it would be exceedingly rare) and does not change the probability of getting heads or tails for the next toss. It's still 50% for either.
What I'm thinking is that if you have a loco which is showing a 25% chance of breakdown, the natural thing is to expect that it will break down once every four trips. I think the game is using probability the way it should, which instead means that for every trip there is a one in four chance that the train will break down, but this says absolutely nothing about the previous trip or the next one. If someone expects that they can figure out when a train "should" break down and it doesn't, they may think that the stated probability is not correct. The only way of knowing for sure would be to play the exact same scenarios several times each, with and without cabooses, and tally up the crashes and breakdowns after each game finished.
Has anyone ever done this? I haven't yet, but I'm curious.
----------------------------------------------------------------------------------------
Which sorta leads to the next thing: how many crashes and breakdowns are acceptable? I always want to minimise them, which influences how I play. The way I figure it, any time I have a crash it will probably cost me a minimum of 300K. That's at least 100K for the crashed loco, probably 100K on average for the load, and at least another 100K to buy a new loco. If the crashed train significantly delays another train on the same line, you can probably add another 100K or so to that total. If your locos are more than 100K a pop that will hurt more too. I think it's fair to value each crash at between 300 and 600K, which is getting quite significant. Breakdowns I figure cost you roughly half as much as a crash. No capital costs, but the several month's delay in turnover will hurt. Given that the last car in the train will be the lowest valued one, I think there is (in general) a good case for skipping it in favour of better reliability and running a caboose.
----------------------------------------------------------------------------------------
The other thing I'm fussy about is maintenance sheds. I hardly ever follow the AI's standard practice of just plonking them halfway between stations. I might do it very occasionally if the run is long enough, but even then the train will be scheduled to stop at the shed. Just assuming the train will stop when it needs oil has always given me worse results. The problem is that if the train has even 51% oil, it will naturally go straight past the shed without stopping. This can mean that by the time it gets back to the shed on the return run it has been out of oil for ages, and that will definitely trash its reliability. I've seen this many times with AI trains. They will frequently be running low on oil, even if they aren't actually out of it, and their reliability bar will be looking pretty grim.
I've noticed when checking my own locos that whenever the oil gets below 50%, the reliability bar rapidly starts getting worse. 50% seems to be a break point for deterioration. The game may specifically say that being totally out of oil will make things worse, but it doesn't say that running at <50% oil will not make things worse. IOW, I think this is coded in as a sneaky bit.
![Wink ;-)](./images/smilies/icon_wink.gif)
Because of this, my trains are always scheduled to stop at all maintenance sheds and water towers. It's a bit of a nusiance to set up, but you usually only have to do it once for each train, so I think it's worth the effort. Sometimes I will let them go a very little bit under 50% oil on a long run if it means they can get there in one go, but I always keep a close eye on any such trains and replace them earlier than normal. I don't so much mind if a train runs out of water towards the end of a long run, provided it is close to its destination and is on the flat or heading downhill. That doesn't hurt reliability at all, and makes no significant difference to delivery time since it will coast down to a lower speed rather than stopping suddenly.
Since I'm scheduling my trains to stop at all maintenance buildings anyway, I almost always put them on spurs next to a station. I do this even if the scenario doesn't set a speed goal. It just doesn't make sense to me to stop a train that is going full knacker if it will get to the other end in one go. The acceleration time for a fully loaded freight train is abysmal, while if it goes the the sheds after unloading it will take off like a rocket. Result: faster turnaround and less congestion. It might cost me a little bit extra in spur track, but I think this is outweighed (or at the very least equaled) by the better turnover.