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

What is triald and why is it taking so much disk space?

eclecticlight.co

Post reply on HN
Print view View original

81–90 of 295 posts

Re: What is triald and why is it taking so much disk space?

#81

Post by Ensorceled »

cabirum wrote: post #12
dannyw wrote: post #5

Remember when operating systems didn't have multi-gigabyte binary blobs performing cryptic actions with an extremely chatty online connection?

No, I don't remember a time when users were in control and actually owned Apple-made hard- and software. However, linux is alive and well, despite some flavours attempting to force autoupdating snap stores and the like onto users.

> Apple-made hard- and software.

Not sure why are a singling out Apple here. Windows is worse as are many (most?) hardware vendors; there is a reason everyone REALLY should fresh install vanilla Windows over the vendors OEM version.

Re: What is triald and why is it taking so much disk space?

#82

Post by jtbayly »

sneak wrote: post #75
skygazer wrote: post #9

I trust that Apple isn't doing anything nefarious here, but at the same time, I'm quite grateful that others aren't so trusting. I want the "white hat" suspicious and skeptical rooting through everything. In fact, I don't think I'd trust Apple so much if it weren't for those that don't trust them. Their constructive distrust enables my trust.

Why do you trust that? Apple has specifically admitted to having a nefarious plan to scan your local files without consent and further use your computer (also without consent) to report on you to law enforcement. This may well be that implementation. The latest desktop and mobile OSes are quietly shipping code to enable this, despite the uproar that happened when Apple announced it. They let people believe they were…

Do you have any evidence this is what’s happening?

Why should I trust your declarations?

Re: What is triald and why is it taking so much disk space?

#83

Post by secondcoming »

londons_explore wrote: post #57
BluSyn wrote: post #29

Earlier quoted context omitted.

I checked mine, and I see lots of mention of "Siri" (TEXT TO SPEECH, FIND MY, etc). Seems most likely this is A/B testing for ML models, such as those for Siri and Photos analysis? That's my best guess. Since Siri and Photos now processes more locally (especially on M1's dedicated ML cores), all the training improvements and testing probably happen locally as well. It makes sense it would regularly download updates f…

I think apple is pretty invested in distributed ML training. Ie. your PC will go through your photo gallery/siri audio data, and use the labels and stuff to fine-tune an apple-provided ML model. The fine-tunings are then re-uploaded to apple, added up, and distributed to other users in the next update. The privacy implications of this are dubious. There is as-yet no good research showing how these finetuned models ca…

> The privacy implications of this are dubious.

Really? Apple can essentially be asked by Law Enforcement for answers to questions like "Give me all the IMEIs of people in location X, radius 10 miles, that appear to have pictures of cats"

Re: What is triald and why is it taking so much disk space?

#84

Post by dangerface »

bee_rider wrote: post #68
rosndo wrote: post #65

Earlier quoted context omitted.

And the people who can do so are usually the same people who’d be perfectly capable of analyzing binaries directly. Besides, you’ll never know if the code you’re looking at abuses some compiler quirk without studying the binary.

Somebody should tell the folks designing high-level languages that their code is no easier to analyze than a binary, I guess.

Debugging C is usually done in assembly so I guess they know.

Re: What is triald and why is it taking so much disk space?

#85

Post by Ensorceled »

mschuster91 wrote: post #41
yborg wrote: post #38

Earlier quoted context omitted.

Maybe you should contain your outrage until it's based on something other than uninformed speculation.

The fact that Apple once again delivers bananaware ("ripens at the customer") to their users is outrage-worthy on its own, particularly if the software in question does not even provide any benefit to the user. Bad enough that Spotlight and TimeMachine have had issues for years (which, to Apple's "credit" have now been largely solved), but these at least provide useful functionality. Apple is a filthy rich company, t…

I really think you should stop buying Apple products, you are clearly really upset about it.

Re: What is triald and why is it taking so much disk space?

#86

Post by sneak »

jtbayly wrote: post #82
sneak wrote: post #75

Earlier quoted context omitted.

Why do you trust that? Apple has specifically admitted to having a nefarious plan to scan your local files without consent and further use your computer (also without consent) to report on you to law enforcement. This may well be that implementation. The latest desktop and mobile OSes are quietly shipping code to enable this, despite the uproar that happened when Apple announced it. They let people believe they were…

Do you have any evidence this is what’s happening? Why should I trust your declarations?

https://web.archive.org/web/20210903214920/https://www.apple...

(linked from the above-linked wikipedia article as citation for the clientside scanning feature documentation on wikipedia)

https://web.archive.org/web/20210903214920/https://www.apple...

Apple said they were doing it.

Apple never subsequently said they were not doing it. (They did say some things that implied that they MIGHT not do it, designed to make you stop being mad at them ("Based on feedback from customers, advocacy groups, researchers, and others, we have decided to take additional time over the coming months to collect input and make improvements before releasing these critically important child safety features.").)

They did, however, reiterate that this was going to be released.

What more evidence do you want than the vendor's direct affirmative statements?

When people plainly tell you who they are: believe them.

Re: What is triald and why is it taking so much disk space?

#87

Post by ratww »

klohto wrote: post #70
dan1234 wrote: post #40

Earlier quoted context omitted.

I've also disabled Siri and my ~/Library/Trial directory is only 3.6M (with no SIRI_DIALOG_ASSETS dir - so no valentines for me). Maybe you should check that Siri is still disabled on your Mac?

Interesting, I have it off but still have the folder. Did you do any "extra" steps to disable Siri?

My guess is the data was put there before disabling, and disabling doesn't automatically remove the data.

Re: What is triald and why is it taking so much disk space?

#88

Post by londons_explore »

secondcoming wrote: post #83
londons_explore wrote: post #57

Earlier quoted context omitted.

I think apple is pretty invested in distributed ML training. Ie. your PC will go through your photo gallery/siri audio data, and use the labels and stuff to fine-tune an apple-provided ML model. The fine-tunings are then re-uploaded to apple, added up, and distributed to other users in the next update. The privacy implications of this are dubious. There is as-yet no good research showing how these finetuned models ca…

> The privacy implications of this are dubious. Really? Apple can essentially be asked by Law Enforcement for answers to questions like "Give me all the IMEIs of people in location X, radius 10 miles, that appear to have pictures of cats"

The model fine tunings can't immediately be used to say "this person has pictures of cats". They effectively are just a big diff of the ML weights. It's a big set of floating point numbers which is hard to get any meaning out of.

But experts believe that techniques might be developed in the future to figure out based on the numbers what data must have been in the training set (and hence your photos). Nobodies managed to do that yet.

Re: What is triald and why is it taking so much disk space?

#89

Post by npteljes »

AussieWog93 wrote: post #61
caslon wrote: post #53

Earlier quoted context omitted.

"One distribution using one package manager is bad when piloted by an incompetent YouTuber who ignores multiple warnings. Therefore, all of free software is bad."

It's more than just Linus's experience with Pop OS. So many (consumer-focused) FOSS products have huge, obtuse quirks in them that prevent their widespread adoption, and uber-tech-literate people tend to overlook this because they really want to believe that FOSS can succeed. In reality, a lot of these quirks come about because of the non-commercial nature of the projects[1]. When a business produces software, the ai…

No. Widespread adoption is hindered of the status quo that Linux is not the default OS on the PC platform. If governments used Linux internally, mandated that schools teach Linux and Libreoffice, and the Windows bundling thing hadn't happen, and especially if all this happened 15 years ago, the landscape would be vastly different. Linux wouldn't have fewer quirks, maybe it would have more, but nevertheless people would use it because that would make the most sense to them. This is the only thing that matters, not user experience, not the goal of the software, but things that are orthogonal to the software: its integration into society.

Re: What is triald and why is it taking so much disk space?

#90

Post by ben-schaaf »

dangerface wrote: post #77
sdfjkl wrote: post #6

Open source is where you don't have to apply guesswork to find out what a suspicious looking piece of software is doing on your computer.

Linux package managers are great I haven't had to build a program from source in years. But how do I know the source code I check is the binary my machine runs? Even if I build from source I could have a malicious gcc that takes clean source and outputs a malicious binary. Unless you are running jit or something you can't really know what your computer is running even if you use open source.

You can check whether the binaries you have are produced by the source if you have reproducible builds, see https://wiki.debian.org/ReproducibleBuilds.
Post reply on HN