I agree that the geography of Canada makes having a rating system which is consistent difficult....
An anecdote: When I began playing chess at the Hart House Club there were a number of players there who began at the same time. We all played in the CUCC reserves 2006-7 and recieved provis. of between 1000 and 1300. We all almost never played outside the club. We never played against the established players in the club. After a year or two of Friday night blitz we were all much improved--- Anyone of one of us on varying occassion would play an event elsewhere and have an 1700-1900 perf. Then the player who played an event outside would generously donate his new wealth of rating points to the rest of us at the next in house event... since we were all close to the same strength. Now (4 years later) most of us have caught up to the national pool (Though there are still a few guys in the 1500s).

The problem in this case was demographic--- though outsiders said we were underrated, in fact our rating were accurate within the pool we played in. To us it was everyone else who was overrated.

The problem of isolated rating pools is one which I do not think will be corrected by the means suggested in this thread. When I looked at the junior Paul mentioned, I noticed that although they would have occasional perfs. above the their current rating, within their localized pool they still performed at their current rating.

It seems this issue, which stems from demographic/geographic factors should require such similar solution... the issue being that the only way to determine how deflated/inflated a pool is is by having them play against outsiders---
which they don't do consistently enough to provide relavent data. :$