As Antoine Poinset suggested in a comment
Maybe spending the corresponding output is considered by the metaprotocol as transferring the inscription
From what I understand of Rodarmor’s original “Ordinals” inscriptions process, it does transfer ownership of the inscription or inscribed NFT when someone spends the output that “Ordinal theory” associates with the inscribed input. So the output can’t be spent without affecting the inscription.
Satoshis live in outputs, but transactions destroy outputs and create new ones, so ordinal theory uses an algorithm to determine how satoshis hop from the inputs of a transaction to its outputs.
Fortunately, that algorithm is very simple.
Satoshis transfer in first-in-first-out order. Think of the inputs to a transaction as being a list of satoshis, and the outputs as a list of slots, waiting to receive a satoshi. To assign input satoshis to slots, go through each satoshi in the inputs in order, and assign each to the first available slot in the outputs.
Source
As transactions occur, the Ordinals protocol tracks each sat through subsequent transactions in a “first-in-first-out” scheme. The sats’ identifying numbers are called Ordinals, since both the identification and the tracking mechanism are dependent on the chronological order of creation and transactions.
Source
Also as each individual Satoshi is inscribable, there is an incentive to create many UTXOs of minimal value, i.e. fewest Satoshi. I’m pretty sure this is what I observe in BRC-20 usage.
