I think this study is quite good but it does come with one important caveat:
From page 5, "This is before the introduction of Apple's new opt-in mechanism for tracking in 2021. Our dataset therefore reflects privacy in the app ecosystem shortly before this policy change."
Personally, I would prefer to see the difference between apps monetized by customers paying directly and apps monetized by advertising. I take for granted advertiser will track, and I assume paid apps track less but I wonder if that's actually true in practice. It very well might not be.
It should be assumed that any app that won't allow public review and independent builds of their source code can and will take actions for profit you may not approve of at any time, be it intentionally or out of negligence.
Letting a third party run code on your system they don't permit you two or someone you trust to see is a bit like being asked to adopt a legal contract that has some amount of power over you you are not permitted to read.
If you want apps willing to prove you can trust them to respect your privacy and freedom, the reasonably strict process of apps in the F-Droid store are what you are likely looking for.
There is an accountable and privacy preserving alternative for almost everything.
Thanks for pointing me to that. It's hard not to agree with the Transparency Matters conclusion but the 13% drop in Tracking Attempts is worth calling out so I'm glad the Oxford researchers did it.
This is not my field but one thing that I don't get when looking over the Transparency Matters report [1] is why do some apps have 9 trackers and others have 300 requests? Are they really so different? And are they able to get information that Starbucks didn't with only 3 trackers and 21 requests?
I guess in the end the magnitudes aren't so important but I found the large differences pretty amazing. If nothing else I would think they might want to reduce the amount of data they need to retain, reducing their costs. It seems like there has to be diminishing returns there.
It also shows how much data Google has as they show up on 8 of the 10 apps tracker lists and one can only assume they keep a copy of all that as well.
So perhaps the iOS/Andriod comparison is meaningless anyway as Google knows all about you either way.
This article shows a fundamental misunderstanding of the functionality of "App Tracking Transparency", and I will admit the text "ask app not to track" is confusing.
ATT, when a user selects "no", blocks the app from accessing your device ID. This makes it so that even though the app can still track what you do in their app, they can't connect that to the data collected from other apps through your device ID, and therefore build a profile of you as a person.
I analysed[0] this early in 2021 based on the self repeated “nutrition labels” that Apple started requiring. There’s definitely a strong correlation between and collects significant data.
From page 5, "This is before the introduction of Apple's new opt-in mechanism for tracking in 2021. Our dataset therefore reflects privacy in the app ecosystem shortly before this policy change."
Personally, I would prefer to see the difference between apps monetized by customers paying directly and apps monetized by advertising. I take for granted advertiser will track, and I assume paid apps track less but I wonder if that's actually true in practice. It very well might not be.