Consider this: Most humans want trains to be safe (A). Some humans are lazy fucks and will risk hurting people or deliberately hurt them to save time and make

Consider this: Most humans want trains to be safe (A). Some humans are lazy fucks and will risk hurting people or deliberately hurt them to save time and make money (B). Some humans know a lot about how trains work (C). Most humans don't know much about how trains work (D). Some humans don't know anything about how trains work (E). A train is an object that operates at normal scales, in multiple senses. It is very natural for the state of the safety of a train to be observable to its users. They would know if its rusty and unkempt, if the doors weren't put on properly, if the engine is making unusual sounds, if the conductor looks shady and may be there to scam them. Group D may not know exactly how the train is going to kill them, but you don't need to be in Group C to recognise most design failures. The safety of trains is ensured in part by the size of Group D, operators (Some C, Some B) won't make enough money if Group D refuse to ride the train so they keep standards high. Now software isn't quite the same as trains. Software has nearly everyone, like yourself, nearest to Group E. Group D is not insignificant - these are those who work with software at a design level, and can spot design faults. Group C are minuscule. Closed source software means that the only ones checking Group B are the members of Group C who have, for whatever reason, gone the extra mile. Group D are as oblivious as E. Open source software allows Group D to get involved in checking that the operators are more of Group C than Group B. --- Your busses vs trains scenario is interesting. Trains are controlled by widely available and auditable schedules, making them more open than the untracked antics of the completely closed source reasoning of bus and car drivers.