I was thinking, yeah Google, why not turn the knobs a tiny bit - people will hardly notice and it's win-win for both of us anyway.
"The problem of fake references in Google Analytics has changed significantly over the past few years. In 2014, we had some bots from semalt and buttons-for-website that visited your website and left fake referrals in your analytics. In December 2014, the attacks began taking advantage of a weakness in Google’s new Measurement Protocol that allowed direct attacks on the Google Analytics tracking servers without having to actually visit your website. This is a lot easier than crawling the web looking for new websites. "
[0]: https://help.analyticsedge.com/article/definitive-guide-to-r...
However these stories involving some mega corp as an evil actor always raise the question: what's to be gained, and what's the risk.
As soon as some manager instructs someone beneath them to do bad stuff, some kind of trail exists. The downside to mega corp is enormous if that comes to light.
Imagine if Google is caught in the act of pulling some shady shit like that. The law suits, loss of trust and prestige. The personal exposure to the individuals involved.
All it takes is for some underling to pop up with a smoking gun email, a recorded conversation on their IPhone, a receipt from a shady transaction. Then it's immense risk and a big cover up needed to try and save things.
I don't doubt that shady shit happens, passed down the chain Mafia boss style so there's no paper trail, even at Google.
But does that lead to actual dodgy acts like prioritizing AdSense customers in search results?
How many engineers would need to be brought along for the ride for that to happen? How many of them are potential whistle blowers?
For all of these reasons (and perhaps naively) I like to think that corporate corruption is not this overt.
Tldr; I hope you were mistaken and it was purely a coincidence.
Step 2: Design and implement sufficiently complex algorithm that behavior isn't easily reasoned from changes
Step 3: Iterate through random walk of changes to optimize KVMs
Step 4: Land on parent's behavior, without any attributable decision or paper trail
Risk is too high for such things to be decided, but maybe someone buried a change in the second decimal place in an to "ad-fraud" prevention subsystem, which required a minimum amount of traffic or some other non-voluntary things like that.
Even more: impenetrable deep learning models trained to maximize ad revenue - imagine pagerank and availability of ads on the site are two of the N features, what would the optimizer suggest you do?
Easily brushed away, quickly forgotten.
> The personal exposure to the individuals involved.
Virtually nonexistent.
To pick an example, Google colluded with other tech companies to limit the salaries of its engineers, yet here you are claiming it couldn’t possibly happen.
Do a search for “corporate malfeasance” and see the truly monstrous acts committed by companies that continue to do business quite successfully.