Thread · 4 tweets · 15 Jan 2023

Replying to @Log3overLog2 and @vkw, @larakiara

I beg to differ. To begin with, they're designed as one sided: they don't (and can't) measure the loss that the system creates. Then they don't break out the value of different cookie functions. Maybe there is value but all of it is from, say, attribution?
↺ 1
They make a big bag that they call "publishers" but offer no useful details about variability (which is, as far as I can gather, high). Finally, there are way too many factors that they don't control. So the number could be different to the point of changing the sign.
A fix that addresses just a fraction of the use cases could be sufficient. Or a fix that is specific to one industry or product class could be sufficient. Or there may just be too much noise. It's irrelevant to the question of Topics because it doesn't inform a decision.
So "we're going to ignore the consensus of the web community because we have these dodgy numbers" doesn't strike me as a great example of responsible stewardship of the platform, sorry.