On Ethics in Software with the Authors of “The Pragmatic Programmer”

> Well, exactly. It's kind of funny because it reminds me of the early ages of industrialization, the stories from the factories were pretty horrific. ... A lot of large scale tragedies, because there really wasn't any concept of worker safety, because it was all new. As a civilization, we didn't really quite know how to handle that yet.

You could argue that it was hard enough just to make the machines work without having to also make them safe. You could also argue that workers hadn't yet figured out how to organize (unions) against their bosses, and that consumers hadn't yet figured out how to organize (boycott) against producers. But at some basic level it's because the factory owners just didn't care enough to make the machines safe. All of which are true today as well.

Different point: As technologists we are implementing someone's vision. It might be ours or someone else's, but it's either contributing to society or it's not. It can be building power for the powerless, or lining the pockets of the already rich. We as technologists can choose whether to take this job or that job, or whether we continue working on a project once we know the ethical implications of it. Unfortunately the ethical choice is usually less lucrative.

The excerpt is too small to fully articulate a complete point about ethics, but I agree with the small ideas: We have no idea what we're doing, we aren't good at thinking for ourselves and demanding ethical behavior from our employers, and we don't take the effects of our actions in the world very seriously.

The problem that I see, the difference between software and other engineering domains is this: it is easy to re-purpose software. Macbooks and programs like scikit-learn are not inherently dangerous or bad, but once that's available to everybody it's easy for one individual (not even with engineering background) to make a harmful piece of software. It only take one person with evil intentions to build a racist prediction algorithm for their insurance company and affect millions, where it takes dozens of engineers to build a bomb or a border wall and harm thousands.

The idea that engineers could unionize and refuse to build facial recognition while continuing to develop self-driving car building blocks seems unworkable in practice. Unless we lock all the open-source ML code and only let people access it after taking some kind of pledge?

I agree with the principal of it but I am not sure I can find a company that is not using software as a competitive edge to take poach business away from others (smaller companies) or cut costs through automation. The world we live in is seen as a zero-sum game.

TurboTax has drained money from personal accountants. Cloud computing is shifting responsibilities from "IT admins" to developers. Autonomous personal and commercial transport is just waiting to put a large number of people out of work. I am not sure how you can end up in development and not participate in mass misery of non-developers.