4 ms·
How does a moral compass help the engineer whose spec says "If DEBUG_SEND_CRASH_INPUT is defined at build time, send the content of all files opened by the cras
by yorwba 10y ago
How does a moral compass help the engineer whose spec says "If DEBUG_SEND_CRASH_INPUT is defined at build time, send the content of all files opened by the crashing process to the telemetry server."? How are they supposed to know that the flag will be enabled in the release build, causing their code to exfiltrate user documents to Microsoft?
I believe this is what nebabyte means by
> A moral compass can't work without a magnetic field of context.
- Sir_Cmpwn 10y agoThat makes sense. The only thing for it is to start the uproar now, then, and hopefully it'll be pulled in the near future.
- nebabyte 10y agoExactly. It's not like it's even going to be 'a devil in the details' kind of thing; once the mechanisms are built, they could be used securely or maliciously depending on any number of factors such as presets. The engineer isn't necessarily going to get a play-by-play of the big picture - just what they need to know to get their part built - so expecting some moral stand against the final product they probably weren't part of the meeting on sounds absurd.
- pdkl95 10y agoUnless you are working in a fully compartmentalized organization (e.g. the NSA), engineering doesn't happen entirely in isolation. I agree that the complexity of computers can obscure what any piece of the whole will eventually responsible for, but engineers should still be considering the ethical implications that are visible, especially the global nature of the project itself. Regarding "telemetry" specifically, it should be obvious that designing something that automatically exfiltrates random data is a potentially serious security issue. It shouldn't be transmitting anything without explicit permission. It is irresponsible to assume that a preset represents intent and the informed consent of the user. Even when permission has been given, care should be taken to limit collection to only what is needed and log entries about what was sent should be created. Yes, even on debug builds. > The engineer isn't necessarily going to get a play-by-play of the big picture Maybe you should ask for that before taking the job. You don't need to know every detail, but it's negligent to work on a project without at least some understanding of what the big picture goals are. -- My point is that you do know some things about the projects you are working on. You need to consider not only what your designs do, but also how it may be used, which may not be what you originally intended. You no longer have the luxury of pretending engineering work is entirely isolated from political and social consequences. Those consequences will vary greatly in magnitude and are probably good in many situations, but it's irresponsible to ignore them entirely. The entire concept of "disruptive" startups relies on the existence of those consequences and how strongly they impact society. I encourage everyone to watch this[1] talk from 30c3 (or read the transcript[2]), which discusses this topic. [1] https://media.ccc.de/v/30C3_-_5491_-_en_-_saal_1_-_201312272300_-_no_neutral_ground_in_a_burning_world_-_quinn_norton_-_eleanor_saitta https://media.ccc.de/v/30C3_-_5491_-_en_-_saal_1_-_201312272... [2] http://opentranscripts.org/transcript/no-neutral-ground-burning-world/ http://opentranscripts.org/transcript/no-neutral-ground-burn...
- xorblurb 10y agoHu, you seriously believe Windows is developed like this, with an army of entry level programmers writing code from an ultra-detailed spec for a product they don't know? (and then what, they use Linux and Mac at home, probably?) At least your theory would explain why some parts of the OS are so shitty :p But I'm not buying it anyway.