Comment by AnthonyMouse
19 hours ago
You, an engineer at a major aircraft manufacturer that isn't Boeing, have been working after hours with some of your colleagues on a hobby project to add some modern safety features to an older model of small private plane, because you regard it as unsafe even though it still has a government certification and you got into this field because you want to save lives.
Your "prototype" is a plane from the original manufacturer with no physical modifications but a software patch to use data from sensors the plane already had to prevent the computer from getting confused under high wind conditions in a way that has already caused two fatal crashes.
Now you have to fly somewhere and your options for a plane are the one with the history of fatal crashes or the same one with your modifications, and it's windy today. Which plane are you getting on?
This example is so right. Including the parallel with what happened with those two aircrafts.
Definitely not the untested code I wrote myself!
Are you kidding me? How many times have you unwillingly introduced bugs into a code base you didn’t fully understand? That’s basically table stakes for software engineering.
> Definitely not the untested code I wrote myself!
Nobody said it was untested.
> How many times have you unwillingly introduced bugs into a code base you didn’t fully understand? That’s basically table stakes for software engineering.
Which applies just the same to the people the company hired to do it, and now we're back to "the people with a stronger incentive to get it right are the people who die if it goes wrong".
I can’t tell if you seriously think a random person writing code in their basement is equivalent to a company that has access to API docs, design specs, actual test hardware, the expertise of a ton of engineers that have worked on the project and understand how it can go wrong, not to mention all the regulations and verifications they’re subject to.
But if you do then wow. That really puts in perspective the kind of people that use hacker news. I’m gonna be more selective about who I bother replying to going forward.
1 reply →
Tested how? With 100% "unit test" coverage? I can certainly see how a random person on the internet might be highly motivated and actually talented enough to contribute to these sorts of projects. But they don't have the budget and resources that commercial entities have. They don't have the same due diligence requirements. They don't have the same liability. If I use a commercial device unaltered, it's the company's fault if the device fucks up or is defective and causes harm. If I install random internet software on my medical device and it fucks up and causes harm, it's my fault.
I say this as someone who might modify my own medical devices because I'm so fucking jaded over the capitalist march towards enshitification and maximizing profit over human lives. There is simply no way random folks on the internet can test these types of systems to any reliable degree. It requires rigorous testing across hundreds to thousands of test cases. They at best can give you the recipe that works well for them and the few people that have voluntarily tried their version. That doesn't scale and certainly isn't any safer than corporate solutions.
4 replies →