it could be related to the fact that there's more people leaving in cities, not that many cities, and those cities have data centers close-by. bigger concentration of population near data centers should go a long way.
Yeah, that stat is based on performance tested by speed.cloudflare.com, so it's actually a mix of overall capability and local speed to Cloudflare data centers. It may also be skewed by relative unpopularity of Cloudflare's speed test site compared to more well known alternatives.
Their speed test service isn't very accurate either, its always over reporting by a factor of 100Mbit/s for me (vnstat running at the time of the test shows this). The JavaScript tests in general are not very useful IMO.
The continent graphs are actually based on measurements towards a set of providers, only one of which is Cloudflare. This is mentioned in the blog post.
Yeah, the IQI measurements are aggregated, but much lower than the continent map, which is only to Cloudflare. So the high bandwidth figure noted by parent is only for Cloudflare.
The map shows numbers from speed tests, which is only to Cloudflare, yes. But IQI's bandwidth numbers are lower because it displays average utilization (i.e. shorter requests, like web browsing, mostly capped by latency) while the map displays connection capacity.
I believe the parent was referring to the lines for Oceania being surprisingly good considering we usually think of it as being far away from content (which is not true when a large chunk of it can be served from local datacenters).
The entirety of Oceania has roughly the population of California.