> If blocking a subset of users is such a good idea, why isn't Google rolling it out to everyone today?
There's a bunch of reasons to do an experiment with a small percentage of users first, and to do a gradual rollout.
First, they don't actually know for sure that this works, since users are unpredictable. You need an experiment to see how they react now and in the longer term. Do all the impacted users just immediately switch to a different ad blocker that isn't getting detected? No point in rolling this out further until that is fixed. Do all of those users churn out entirely, rather than some of them disabling the adblocker or subscribing to Premium? Then there's a business decision to be made, and they didn't have data for it earlier.
Second, there might be technical issues that make the implementation harmful. For example maybe the JS code causes crashes for legit users on a browser you didn't test, or something. You'd like to contain the damage to just members of the experimental group, not all users of that browser.
Third, even if they know this will be a beneficial change, they don't know how beneficial. The people who proposed and implemented the project want to show the impact (an increase of X million in ads revenue, extra Y thousand Premium subscribers, Z reduction in serving costs, whatever their KPI is). The people who funded the team want to know whether they actually got a return on the investment, and whether they should continue staffing further work to improve coverage, etc.