Yes, but when matching the estimated number of Internet users against servers, then there are still about twice as many users per server in Belgium than there are in France. I guess with number of people (vs. estimated Internet users), that might change, but only slightly.
I think there is a big misunderstanding: As has been discussed at length and in depth a while back, the pool system assigns clients to servers from the same zone only as long as there is at least a single server in the zone. Only if the zone does not have any server in it (considered separately for the two IP address families, obviously) will there be a fall-back to the enclosing zone, i.e., continent. (This is true when any of the “global” names are used. When querying with a specific zone in the name, if the zone has no servers in it, that is how many servers one gets back - none.)
There were various proposals at the time, among them the general idea of what you describe, i.e., fall back to the enclosing zone in some way (various ideas on that) when a zone is considered not having “enough” servers (again various thoughts as to how that could be defined/measured).
Ask himself wrote one or two years back that that is something he plans to do, but has been silent on that front ever since, and various other features have come out since then instead.
How do you determine that? I mean the “than needed” part?
EDIT: I realize that you probably refer to servers being hammered needlessly as ways to prevent that are known.
The speed setting is primarily for sharing among servers within a zone. It doesn’t have anything to do with offloading to enclosing zones. There was an idea a while back to try to make the speed setting more of an absolute setting, rather than relative to the overall sum of netspeeds of all active servers within a zone.
I think the setting is also used for load sharing among servers in a continent zone in some way, but I have yet to figure out exactly how.
There is no clear definition of what some people consider an underserved zone, but one aspect could very well be that a zone may be considered underserved when a server with a typical home Internet connection setup is not able to join the pool even at the current lowest setting of 512 Kbit.
Belgium still is in a relatively good position. Consider what happens when the 512 Kbit setting gets you double-digit Mbit/s traffic, which performance aspects aside (NAT, …) may overwhelm Internet connections in large parts of the world, e.g., DSL lines with only 1 or 2 MBit/s in the uplink direction, or cloud servers with only 1 or a few Mbit/s contractual uplink bitrate, or low traffic volume quotas.
That is why I keep advocating for allowing netspeed values below 512 Kbit. E.g., I have a server in India that is at a 1 Kbit setting (from when that was still possible by tweaking the system), and despite its low contractual traffic volume in this case, it can still contribute to the pool. Not much, but “every server counts”.
Or my server in Malaysia, which has 1 Mbit/s contractual bandwidth for the IPv4 side, but it can still contribute to the pool because of the legacy 1 Kbit setting in this case as well. It isn’t much, either, but again, “every server counts”.