I think I get what he is trying to say and there is some substance to it - but it would have been a better article if he had included some concrete examples of his methodology.
The analogy that came to my mind was that of a forest fire: You can spend lots if time finding out the exact cause of a wildfire (was it an out-of control campfire? An electrical fault? Arson?? etc etc) but it will not help you much preventing future fires. In the worst case, you'd spend an endless amount of resources playing whack-a-mole against increasingly random and hard to avoid causes: No campfires, no smoking, no transparent water bottles (?), no glasses (??), no power lines (???) and even then you get random lightning strikes that no human was in any way involved with.
A more productive question would be how those banal and mostly random events have the ability to balloon into something catastrophic like a wildfire - or rather under which circumstances they do and under which they don't.
This brings you to questions such as the general health of your forest, accumulating underbrush, dryness and eventually even climate. It also would change the objective from avoiding sources of fire to changing the forest in such a way that individual fires cannot spread.
Peteragain•8mo ago
Good example. Adding fire breaks is another.
A key point here perhaps is that RCA - looking for causes - is often used to do blame (re) allocation. The practice of, when an accident happens, those responsible for the system look for someone to blame rather than making the system more fault tolerant. The HORIZON IT scandal might have had IT problems, but the scandal was the management response to the problem. Indeed, as the IT guy at the inquest said, the IT issues were not even that rare or large. Rather than dealing with it, management blamed the post masters ... and then the IT team ... and when that failed, tried crying.
wduquette•8mo ago
Seems to me that the OP’s discussion presumes that the obvious managerial and institutional root causes have already been dealt with. I think he’s right about latent faults and fault tolerance; but an organization that doesn’t do good unit testing and remove technical debt is going to have lots more overt faults than one that does. Seems to me that a both/and approach is needed.
xg15•8mo ago
The analogy that came to my mind was that of a forest fire: You can spend lots if time finding out the exact cause of a wildfire (was it an out-of control campfire? An electrical fault? Arson?? etc etc) but it will not help you much preventing future fires. In the worst case, you'd spend an endless amount of resources playing whack-a-mole against increasingly random and hard to avoid causes: No campfires, no smoking, no transparent water bottles (?), no glasses (??), no power lines (???) and even then you get random lightning strikes that no human was in any way involved with.
A more productive question would be how those banal and mostly random events have the ability to balloon into something catastrophic like a wildfire - or rather under which circumstances they do and under which they don't.
This brings you to questions such as the general health of your forest, accumulating underbrush, dryness and eventually even climate. It also would change the objective from avoiding sources of fire to changing the forest in such a way that individual fires cannot spread.
Peteragain•8mo ago