Comment by gravescale | Hacker News Reader

Comment by gravescale | Hacker News Reader

Parent Full thread

> If it is physically and financially possible to make your product safer, you do it, without any thought to how much a life is worth.

A cursory glance at how literally any product is designed will tell you that this is not true.

All engineering is fundamentally compromise between utility functions and cost functions (where cost is monetary, or weight, or size, or poor usability, ugliness etc). There's always a balance point somewhere. It may be the cost/benefit functions to the user differ from those of the manufacturer, and again to regulators, so the product appears out-of-balance to one or the other, but a meta-balance was struck somewhere.

If every light switch was a 2-foot, 200kg cube with a titanium shell filled with monitoring electronics running on lockstepped processors, fire suppression and potting compound, and was tested individually for a year before sale, it's still not as safe as one with six grounding points (in case the first 5 fail) and triple-thick gold plating on the contacts. You have to stop somewhere.

This isn't what the parent comment is saying. Functional safety is an entire discipline. It neither results in titanium light switches nor uses "value of a human life" calculations. The scenario described by the parent would be a civil and potentially criminal liability.

I know what functional safety is. It's just not described by "you must always make all possible efforts to make things safer under all possible circumstances".

In the case of a normal light switch, for example, two cable screws rather than one per wire would be safer than one as the cables are less likely to come adrift. This would easily be affordable for the manufacturer and the buyer, and well within the abilities of the manufacturer. And yet no light switch has those.

And every aircraft manufacturer could think of something the costs $20 billion to develop and adds safety but they don't, unless the product is unsafe already The one crash a year is, to be fair, solidly in the unsafe end of the spectrum, but would they spend the same to avoid a once in 10,000 year crash and if not, what's the cut-off? There must be a cut-off or it would be impossible for any manufacturer of anything to make any profit, ever, as all spare cash must go into safety system research. And this is demonstratively not what they do.

Boeing has a slightly different problem in that they actively made something less safe and lied about it. They'd have gotten away with it but they flubbed the execution with a single point of failure, which made it easier to line up the Swiss cheese holes and the rest is history.

> It's just not described by...

Yes, I know - I think I read the parent comment as "If it is physically and financially [reasonable] to make your product safer, you do it"

This is basically what I was taught working in automotive safety.

> ... what's the cut-off? There must be a cut-off...

There's no hard $$ or other cutoff though, it's a soft cutoff that may come down to engineers and/or management arguing about what is really safer.

> Boeing has a slightly different problem in that they actively made something less safe and lied about it.

Yes I agree, for Boeing it is different, reading about their management's intention destruction of positive safety culture is sickening.

Reply on news.ycombinator.com