[gxs]

I’ve been asked at work to build less than savory stuff, here are some general observations, none of which are admittedly an excuse:

* you get caught up in the moment, hell bent on solving the problem you don’t really think twice

* you don’t want to get that stink on you, you don’t want to be that guy that brings this type of stuff up

* you are mindful of the fact that you are being very well compensated to build it and you don’t want to lose your job

* you know it’s going to fall on deaf ears - maybe they will pay lip service, maybe they won’t but either way nothing will happen

* in the back of your mind you figure someone else is fighting the good fight

On and on, so many different things can go through your mind, who knows which it’ll be on any given day, on any given project

[miki123211]

And sometimes, you don't even know what the feature will even be used for.

Today it's an automatic subtitle generator for people with hearing difficulties. Tomorrow it'll be an AI training data generator. In a year, the NSA will re-purpose it into a mass surveillance tool.

[david-gpu]

> And sometimes, you don't even know what the feature will even be used for.

I did some work in the early 2010s that we expected to be used for computational photography, gaming, and little else. Years later, after I had already left the company, its primary use case became image stabilization for quadcopter drones, something that had not crossed our minds at all when we were building that stuff.

Cue in all the drone footage from the Russian invasion of Ukraine. FUCK. FUCK FUCK.

[gxs]

Exactly

Kind of crazy that I’m being downvoted for just expressing some basic, reasonable feelings

[vasco]

Maybe you're finding they aren't so reasonable.

[grues-dinner]

This is all true, and I suppose I participated in a signed update mechanism that I knew the (corporate) end user probably wasn't going to be given the keys to. But, I think there's a difference between this and deliberately going to work on a system that's clearly just top-down designed for something low.

For example, I don't think there's anyone in the (large!) fixed-odds betting terminal industry that can honestly say their work is a good thing for the end users.