Technology

If I understand this properly, I believe you're missing the point. I'll explain my reasoning so you can point out any flaws you perceive in my reading of your comment or my argument.

You've focused too much on how uBlock could theoretically (or not) block outgoing DAP calls and JS code execution. This is way past the point where UBo would've done its job. You need to consider the order in which these events may happen and how they depend on one another.

From the explainer:

At impression time, information about an advertisement is saved by the browser in a write-only store. This includes an identifier for the ad and whether this was an ad view or an ad click.

A site can register ad impressions, either when the ad is shown or when the ad is clicked, at their discretion.

If the ad is never downloaded, something UBo is great at guaranteeing using filter lists, the user could never reach impression time. The JS code is likely never downloaded. An impression is never generated. There is no point in generating impressions for nonexistent, unseen ads. That would be garbage data, which is actually worse for advertisers. No impression data is ever generated, thus there's nothing to send to the aggregate either.

The user does not participate in the system, at all, because it depends on actually engaging with its components, and UBo users have freed themselves from this system completely long ago.

Remember, this is not a privacy enhancer targeted at people who use UBo, but at people who don't, which is still most people, sadly.

There is very limited ability to surgically remove such things.

There is no need to do so. UBo removes ads with prejudice.

Regardless none of that changes the fact that this should have been opt-in from the start instead of opt-out.

I'm still on the fence about this. Currently, the way I see it, Mozilla's biggest sin is being awful at effective communication. Worse than Google, but Google has intent to deceive, while Mozilla seems like they're actually trying to do it properly and just... not getting it right. Spectacularly. Multiple times in a row.

Assuming user consent really stinks, though.

but that alone suggests there might be privacy problems with this entire thing.

I'm not sure if this is a good argument. This is by design, aggregate anonymization works with quantity. I don't think that means it's necessarily a bad design. We use lots of faulty, problematic tools everyday—so long as this one is better than what it's trying to replace, I believe it deserves a chance.

This wouldn't be the first time that a supposedly anonymized data set could be at least partially de-anonymized.

Yes, that's true. I'm choosing to both hope all these experts make it work, while also keeping a careful eye on the project, to the extent of my ability. Maybe you could call it a lazier version of trust, but verify.