Thank you trohde for this interesting info.
What stands out in my view is the level of consistent pageview after removing peaks and bulges. It could be described as “the baseline” from which peaks and troughs move off.
A casual eye-balling tells me:
2700 Before Fan Hui games announcement
4000 Before Lee Sedol games announcement
6000 After Lee Sedol games
You have to keep in mind I’m not thouis , and the trailing off period may not be over after the Lee Sedol games, but it could mean there has been over 120% increase in the general level of interest on Go in the world (with a bias towards English-speakers).
Yes, and since the DeepMind team will be at the US Go Congress 2016 and since most probably there will be more games later this year (Ke Jie?), I expect this baseline to rise to ~10K (or even more) views by mid-next year.
The trailing off after the Lee Sedol vs. Alpha Go match looks pretty smooth. Seems like we could find a function which approximates its behavior and extropolate to predict what the new baseline will be, no?
I agree that the “new baseline” is of great interest. Here’s an update showing monthly views, from the day that the Ke Jie matches finished 2017-05-27.
We see that typical monthly page views are bouncing around between 70000 and 100000 (2400 and 3200/day), with some spikes above that in Dec and Jan, probably based on the “master” series. It will again be interesting to follow up in a few more months.
[quote=“nealmcb, post:6, topic:7441”]
from the day that the Ke Jie matches finished 2017-05-27
[/quote]Uhm, from the day? Something’s wrong in that statement b/c the monthly view can’t show May yet.
Here’s, for comparison with the image in the OP, is another daily views graph, from 2016-01-01 to 2017-05-26:
Indeed - it was taken on that day, but did not include that day since it was monthly. Here is all the monthly data now available, including all of May. So we see a modest increase in May, about equal to December 2016, but only about 21% of the peak back in March 2016. The daily views in June returned to about the recent baseline.