This is a great effort. However, if you don’t factor in that your IP is recognized as you by Google, then you aren’t anonymous by any stretch. Google presents results that are biased; maybe in your favor, maybe against. But they are biased results. The only way to do this fairly, is by making the experiment operate from an anonymous point of view. Maybe use a Tor browser or some other method to make yourself an unknown entity. Then do it ten times (or a hundred) with different identities, then do some analysis to see what’s going on. Unless you are anonymous and the data is generated from different places and personas, you won’t have a useful result. I’d love to see this kind of effort taken to that level.