Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Brew commands send data to Google Analytics

tobiastom.name

Post reply on HN
Print view View original

291–300 of 342 posts

Re: Brew commands send data to Google Analytics

#291

Post by TeMPOraL »

CJefferson wrote: post #210
codedokode wrote: post #197

Earlier quoted context omitted.

Opt-in can be implemented other way without questions. For example, user could set an environment variable or type a command.

Then we hit the problem that if we don't push that request at users, probably a tiny fraction will turn it on. Worse, that fraction will be the statistically unusual people who bother reading and finding such options, meaning we can't derive any statistically useful results about the user base from them!

Frankly, that's your problem. I don't see even a true desire to make the software better as a legitimate reason to exfiltrate data from people without asking.

Re: Brew commands send data to Google Analytics

#292

Post by TeMPOraL »

CJefferson wrote: post #191
majewsky wrote: post #154

Earlier quoted context omitted.

I can totally see the benefits of telemetry. But it would be way less phishy if it were opt-in, or at least opt-out with a very visible information message.

The problem with this is, do you want every program you ever use, to start prompting you with a series of questions about various opt-in / opt-out questions? I get annoyed enough that gnu parallel keeps asking me about citing it, and that's one program. Bash would like to record analyitics. y/n/more information > y > ls *.c ls would like to record information about how you use it. y/n/more information. > y file1.c fi…

If I saw something like that, I'd format my drive and install another OS.

The very idea of your shell or file listing program sending analytics to the mothership is ridiculous. They shouldn't be even talking to a network directly. They have a set of well-defined tasks, and spying on you is not one of them.

Re: Brew commands send data to Google Analytics

#293

Post by tunap »

CJefferson wrote: post #191
majewsky wrote: post #154

Earlier quoted context omitted.

I can totally see the benefits of telemetry. But it would be way less phishy if it were opt-in, or at least opt-out with a very visible information message.

The problem with this is, do you want every program you ever use, to start prompting you with a series of questions about various opt-in / opt-out questions? I get annoyed enough that gnu parallel keeps asking me about citing it, and that's one program. Bash would like to record analyitics. y/n/more information > y > ls *.c ls would like to record information about how you use it. y/n/more information. > y file1.c fi…

"The problem with this is, do you want every program you ever use, to start prompting you with a series of questions about various opt-in / opt-out questions?"

Simple answer, Yes.

It would lead to conversations about the elephant in the room. It also alleviates the assumption I trust all providers(comprehend the EULA/ToS) and puts their actions under scrutiny. If more people realized what data is collected and by whom for whom they could make an informed, cost/benefit choice. Other useful results are, what is the applicability of that data to the project, what characteristics define the 'Trusted Partners' that data is shared with and, possibly, inject some restraint into the collectors' decisions of what to collect.

edit: removed unintended paste fragment

Re: Brew commands send data to Google Analytics

#294

Post by TeMPOraL »

codedokode wrote: post #180
majewsky wrote: post #154

Earlier quoted context omitted.

I can totally see the benefits of telemetry. But it would be way less phishy if it were opt-in, or at least opt-out with a very visible information message.

Opt-out is what commecrial companies do because they do not respect their users' privacy. Free open source software should do the opposite. By the way I think that browser history should be disabled by default too.

Browser history is useful to user (but maybe the access to it could be restricted by means of authentication). I'm fine with off-line browser history that's accessible only to the user (and not JS running on a website). But sending any of it over the wire should definitely be opt-in.

Re: Brew commands send data to Google Analytics

#295

Post by TeMPOraL »

rubber_duck wrote: post #189
blub wrote: post #102

Earlier quoted context omitted.

It's not paranoia, it's simple understanding of history and human nature.

Freaking out about disclosed anonymized analytics about package usage in an OSS project is paranoia in my book

May be, but you didn't use the word "paranoia" to describe just this in your previous comment; you described also things like medical information.

Re: Brew commands send data to Google Analytics

#296

Post by TeMPOraL »

rubber_duck wrote: post #83
tomfluff wrote: post #68

Earlier quoted context omitted.

Organ donation?

Also somewhat related to this - recently I've realized that this privacy paranoia is going to slow down medical advances coming out from big data so much. For example your wearables get to collect so much biometric info, if that data can be connected to detecting conditions early it would provide a lot of value down the road. At some point we will have the option to collect data about what you ate, what you did, wher…

> anonymized

Remember; there's no such thing as "anonymized data", there's only "not enough other data to correlate out identities from it".

Re: Brew commands send data to Google Analytics

#297

Post by TeMPOraL »

majewsky wrote: post #199
lispm wrote: post #78

In German we have a word called 'Datensparsamkeit': http://martinfowler.com/bliki/Datensparsamkeit.html It's a principle for good software design. This principle is also written down in German law: https://de.wikipedia.org/wiki/Datenvermeidung_und_Datenspars... >Damit gilt in Deutschland der Grundsatz, dass die Erhebung, Verarbeitung und Nutzung personenbezogener Daten und die Auswahl und Gestaltung von Datenverarbei…

For example, data about installed software versions can be used to automatically find systems that are vulnerable to certain exploits.

Data about installed software can also be used to profile you for headhunters or marketers in general.

Re: Brew commands send data to Google Analytics

#298

Post by Nursie »

AsyncAwait wrote: post #250
Nursie wrote: post #211

Earlier quoted context omitted.

Opt Out is scammy. Command line tools do not usually do this at all, particularly FOSS command line tools. This is a very worrying development and reduces my trust in FOSS. --edit-- to whoever downvoted me, can you explain why? I love FOSS but I also lovemy privacy. I have come to generally have a default level of trust for popular FOSS projects. Things like this make me question that trust. Why is this wrong?

> This is a very worrying development Agreed. > reduces my trust in FOSS Please don't distrust a whole category because of a single tool. Most FOSS tools don't do this, especially if they're "Free Software" as opposed to just being "open-source", since "Free Software" indicates a moral, not just practical stance on software.

Well perhaps my assumptions about well-known and widely used open source stuff were just too naive.

I guess I'm not really thinking "distrust them like they actively include malware" just that it looks like I need to at least keep a look out for stuff like this.

I love FOSS. I believe in it, I use it all the time, I've made a few minor contributions here and there too. It's not like I'm saying "OMG FOSS is teh evil!", just that maybe my trust level was calibrated a little high :)

Re: Brew commands send data to Google Analytics

#299

Post by rubber_duck »

TeMPOraL wrote: post #295
rubber_duck wrote: post #189

Earlier quoted context omitted.

Freaking out about disclosed anonymized analytics about package usage in an OSS project is paranoia in my book

May be, but you didn't use the word "paranoia" to describe just this in your previous comment; you described also things like medical information.

I mean the general sentiment. I'm not saying people should have access to your full medical history, personal info, etc. on demand.

I am saying is that these benign things are opt-out not because most people wouldn't want to do them if they weigh the prons and cons but because they don't want to put in the effort of doing so and will just be conservative - which is logical from an individual perspective - but will cause us to lose out on opportunities as a whole.

Also this data is getting collected weather you want it or not, even intelligence people are just taping over their webcams as a security measure - the attitude that we must protect every bit of privacy by default will lead to the future where hidden data collection is the only way to access data - people will be making money off it, it won't be available to general public (for eg. public research) and there will be no transparency about it. And if you think the government will protect you - well they are the biggest transgressor here.

So instead of fighting a lost battle with trying to keep absolute privacy why not just make most of that data public and available and focus on protecting the really sensitive stuff.

Re: Brew commands send data to Google Analytics

#300

Post by codedokode »

TeMPOraL wrote: post #294
codedokode wrote: post #180

Earlier quoted context omitted.

Opt-out is what commecrial companies do because they do not respect their users' privacy. Free open source software should do the opposite. By the way I think that browser history should be disabled by default too.

Browser history is useful to user (but maybe the access to it could be restricted by means of authentication). I'm fine with off-line browser history that's accessible only to the user (and not JS running on a website). But sending any of it over the wire should definitely be opt-in.

An unexperienced user might not know that the history is recorded and later can be accidently seen for example by other member of the family. A good software would not allow that.

I remember some IM app for linux had history turned off by default. I was really surprised then.

Post reply on HN