The 2008 break in GDP
Agreed.
The timing matches the network upgrade.
I notice the schema changed in 2015 — the
adjustment_reason field was added — but there are no backfilled values for pre-2015 revisions. This creates a gap: we know the figures were revised, but we don't know why. Should we attempt to infer the reasons from historical documentation, or should we note that pre-2015 revisions are unexplained?Tomas Shienar Need to check the access permissions.
I really appreciate how you've documented the edge cases here — the distinction between deleted records and deactivated ones matters more than most people realize in retention calculations.
The aggregation is missing a crucial filter on the
status field — we're including canceled and suspended accounts in the denominator, which artificially reduces the metrics by 8–10%.Units?