Re:You can't trust Amazon Underground
by Jach on Tuesday January 05, 2016 @05:57AM
(#10842567)
Feedback alone is probably not a good idea, but I'd say background metrics alone isn't either. Though I think we're in agreement. (I don't care if your components' average EPTs dropped 60% since last release if your app is still crap... You made it better but not in a way I care about, and tell me again why you need access to my list of contacts?)
Production telemetry is a godsend for developers to debug their crappy apps (I'd wish there were no buggy, crappy apps in the first place, but...) but I think there's an acceptable way and a questionable way of getting it. The acceptable way is simply requiring user action. Many programs let you email or just 'report' logs and other info if you or the program think there's a problem you want someone to look into. Some make it easy to see what the data actually consists of which can help you tell if they put effort trying to know as little as necessary, avoiding things not their business. There's also an expectation that the data will be used once for helping you and that's all. The questionable way of getting the data is to just hoover up whatever behind the scenes.
The interesting cases (which apply to other data too) are in between. On one hand it's acceptable for LinkedIn to ask for your email password so they can get a list of people you might know and want to connect with, but you have little assurance they don't (or won't in the future) questionably hoover up all your email contents for vague reason x, no reason yet/just in case, or inadvertently for really unfortunate reason y when they get compromised, and that's enough for at least a creepy feeling if not fear in some cases.
Production telemetry is a godsend for developers to debug their crappy apps (I'd wish there were no buggy, crappy apps in the first place, but...) but I think there's an acceptable way and a questionable way of getting it. The acceptable way is simply requiring user action. Many programs let you email or just 'report' logs and other info if you or the program think there's a problem you want someone to look into. Some make it easy to see what the data actually consists of which can help you tell if they put effort trying to know as little as necessary, avoiding things not their business. There's also an expectation that the data will be used once for helping you and that's all. The questionable way of getting the data is to just hoover up whatever behind the scenes.
The interesting cases (which apply to other data too) are in between. On one hand it's acceptable for LinkedIn to ask for your email password so they can get a list of people you might know and want to connect with, but you have little assurance they don't (or won't in the future) questionably hoover up all your email contents for vague reason x, no reason yet/just in case, or inadvertently for really unfortunate reason y when they get compromised, and that's enough for at least a creepy feeling if not fear in some cases.