[Boldened15]

Headline buries the lede imo, should be "Xcode slows down builds by constantly phoning home". Given the walled garden nature of Apple and the app review process it's not really surprising that Xcode would be full of forced telemetry

Also not the worst thing for Apple to measure average build times or whether developers are discovering some new feature they added, that can be actually helpful for improving the product.

[pjerem]

> Also not the worst thing for Apple to measure average build times or whether developers are discovering some new feature they added, that can be actually helpful for improving the product.

That have always been the point of telemetry. The issue is when it’s hidden and /or the collected data is misused.

[3eb7988a1663]

Or even used at all. Feels like a lot of telemetry is getting collected simply because it can. "What if there is value we could unlock?!" Never mind that all decisions will be made on a manager's gut instinct.

[threeseed]

I have worked at plenty of places where analytics data is used to drive decision making.

And surely we can all agree that a data driven approach is better than gut instinct.

[layer8]

A data-driven approach can optimize for the wrong thing, usually because of statistical fallacies or because it is disregarding context. In that case, gut instinct may yield more desirable results. Telemetry also isn’t a replacement for end-user field studies.

[bolognafairy]

If you let the perfect be the enemy of the good like this then I can’t imagine that you ever have or ever will be in charge of, well, anything.

[tomnipotent]

> A data-driven approach can optimize for the wrong thing

So does firing from the hip, so what? And I'd wager firing from the hip has a higher failure rate. At least by using data I have an actual argument for my position, other than "but muh feelings!" My CEO doesn't give two shits about how I "felt" the project would go, and I can't imagine how poorly that discussion will go when we meet to review what happened.

At the end of the day the companies that are succeeding and growing are using data to inform their decisions.

[lucianbr]

> the companies that are succeeding and growing are using data

Is there data to support this conclusion, or is it just your feelings?

[inetknght]

> I have worked at plenty of places where analytics data is used to drive decision making.

Plenty of places mask violating peoples' privacy as "analytics data" to drive decisions about how to manipulate those and other people.

> surely we can all agree that a data driven approach is better than gut instinct.

Generally, sure, but not always. How you get that data matters. Violating someone's privacy is unethical. Violating someone's privacy to make data-driven decisions is still unethical. Telemetry without knowledge or consent to it is violating privacy.

[shalmanese]

OK, but how does XCode conceivably "manipulate people". Apple's privacy promise was always as rooted in its business model as it was its technology. It says we make money primarily from selling you hardware and paid services so our interests are far more aligned with yours than ad supported free services.

[bolognafairy]

What’s your actual point, dude?

> Plenty of places mask violating peoples' privacy as "analytics data" to drive decisions about how to manipulate those and other people.

Are you talking about a different group of places? Or is this your view of places that collect telemetry?

Do you use Xcode? Are you aware of Apple’s telemetry disclosures? How do you feel that Xcode telemetry has ever or could ever be used to manipulate people? How do you feel peoples privacy has been violated? Do you want to re-state your point in a more concrete way instead of saying something that basically amounts to “I’m upset, so it’s bad!”

I hate rent-a-crowd privacy people.

[inetknght]

> What’s your actual point, dude?

My point is that telemetry violates developers' privacy. Whether that's in a consumer-facing app or in a developer-facing Xcode, it's still violating privacy and therefore unethical to use for data-driven decisions.

> is this your view of places that collect telemetry?

Yes

> Do you use Xcode?

No, because I don't use macOS.

> Are you aware of Apple’s telemetry disclosures?

Are you? How many developers using Xcode are aware of those disclosures? How many developers using Xcode are forced to use Xcode by their employer and therefore cannot have reasonably agreed to anything?

> How do you feel that Xcode telemetry has ever or could ever be used to manipulate people? How do you feel peoples privacy has been violated?

Monitor who builds what kinds of applications and then use that information to decide what products Apple should focus on. Monitor who tries to compile. Tie their name to "security" or "authenticity". Monitor how those built applications are used or distributed, and then change compilation processes to suit Apple's walled garden.

> Do you want to re-state your point in a more concrete way instead of saying something that basically amounts to “I’m upset, so it’s bad!”

My point was already plenty enough concrete. You just simply decided to ignore it because you hate privacy-focused people, and summarily dismiss it as "being upset" instead of addressing any of the issues that were brought up.

[droopyEyelids]

The two most interesting problems I know of with data driven decision-making are:

1) The “drunk looking for his keys under the street light“ problem. There is a human inclination to limit the scope of thought to what is visible.

2) The inclination to look for data that supports your conclusions. This is the real intractable problem. For most people, gathering data is a fetch quest for what they want to do anyway. It’s a big problem for academics doing “real science”. In a company where many are trying to launch, get the promotion and move onto another job, (without even the veneer of a check through peer review) it’s an even bigger problem.

This isn’t exactly a replied to your post I know, but I see that everyone else is using your thread to bring up there quibbles with data driven decision-making, so I thought I’d do the same cause I don’t see anyone making these points

[saagarjha]

No, because data driven decision making is often just firing from the hip dressed up as being scientific.

[rtpg]

I think all else being equal, using data to inform the decision is better than not. But the detail there is about whether everything else is equal.

Is "gut instinct" better or worse than "data driven approach with a bunch of implicit data collection"? Is "data driven + implicit data collection" better or worse than "slightly less good data but explicit data collection"? Is that worse or better than "qualitative user interviews to drive decision making despite that only capturing a subset of users"?

Nobody wants their apps to be badly designed, insecure, slow, crash all the time, and hard to use. Nobody wants their app to be bad. So then you're choosing how you make your app good. And you choose how much you care about each axis while working on things.

Nobody wants their app to be bad.

[bayindirh]

> And surely we can all agree that a data driven approach is better than gut instinct.

Then when people do the right thing, or have a good instinct about what to do next, we call them leaders (if they succeed), and praise them. Like Steve Jobs.

If the same people fail, even with the data on their side, we kick them down, set on fire and parade them for their failure.

[threeseed]

a) There is no correlation between a walled garden and telemetry. I use plenty of open source software that asks for analytics and crash reports. In fact I take it as a positive sign as it means they are committed to making a better product.

b) In this case the provisioning profiles are essential to the build process so it makes sense for Apple to check for updates as you are building.

[lapcat]

> Also not the worst thing for Apple to measure average build times

There's no evidence that Apple is measuring average build times. As the screenshot in the article shows, gather provisioning inputs is actually one of the earliest build phases. Moreover, build time is not a useful measure, because it depends crucially on the number of source files, the programming languages, clean vs. incremental builds, run script build phases, and various other factors that vary almost infinitely from project to project.

The article does not even claim that the connections are telemetry. Gather provisioning inputs is without a doubt exactly what it says it is. Nonetheless, it's not necessary for Xcode to gather provisioning inputs on every build, especially not for non-archive builds, and a side effect of doing it on every build is that Apple receives personally identifiable data about developers and their everyday activities, regardless of whether that was Apple's intention.

There appears to be a common assumption that every privacy violation has to be intentional, some kind of conspiracy, but that's not true. A lot of privacy violations are just thoughtlessness, laziness, or incompetence. That doesn't excuse them, however.

[0x1ceb00da]

Why does it have to slow down builds. If it's pure telemetry, it can be done in background right?