A more extreme version of an already-familiar tradeoff
An offline authorization cache stores a previously obtained decision locally for a device or application to rely on when it genuinely cannot reach the authoritative policy source — not a brief network hiccup, but real, sometimes extended disconnection: a field device with no signal, an app used somewhere without connectivity. This is a sharper version of the ordinary caching tradeoff, because there's no live source to check freshness against at all until connectivity eventually returns.
The design question this forces is stark precisely because there's no live check available during the gap: how long should a cached decision remain honored before the device simply refuses to act on it at all, given that the principal's actual authorization could have changed at any point during the disconnection with absolutely no way for the offline device to know.
What happens once connectivity comes back
Reconciliation is what closes this loop: comparing the decisions made locally during disconnection against what the authoritative source would have actually decided, once connectivity returns, and resolving any difference explicitly rather than silently. An action taken offline under a grant that's since been revoked needs a deliberate response once that's discovered — simply refreshing the cache for next time and moving on leaves whatever already happened during the disconnected window unexamined, whether or not it should have been allowed to happen at all.