As I mentioned earlier, we do thousands of tests to continue to evolve our products. Often, those tests are to help us in how we take and address misinformation or how we address our understanding of what our users see when they come to our sites. We do them constantly to improve what we do.
It is always important with such tests that we practise sound research methodology—for instance, randomizing the samples of people who see the tests. That is in line with what we also did in Canada. Our efforts in Canada were to understand the behaviours of our users with regard to the inclusion of sites that are referenced, as best we could tell, with the drafting of Bill C-18. For reasons of security, we don't release the results of such tests.
I can tell you, however, that at a very high level the tests confirmed several things. First of all, news queries are very small percentages to Google—less than 2% is typically the case—and there was no impact on our users with regard to non-news inquiries, whether that was seeking, for instance, information from academia, information from the government or, for that matter, how to find a local seafood restaurant.