I doubt this is something we’ll want to consider in the next 50 years, but just for academic interest, this has been investigated before:
Sakura, long term UTXO recycling mechanism for Bitcoin:
Abstract: The very long term viability, centuries ahead, of Bitcoin depends on preventing the runaway growth of the UTXO set (unspent transaction outputs). Also, the ever shrinking monetary mass of Bitcoin is a complication which hinders economic agents from fully relying on perfectly predictable monetary conditions. Here, we propose Sakura, a long term recycling mechanism to prune “dead” UTXO entries, defined as entries that have remained untouched for 80 years (defined as 4,200,000 blocks). It allows those “dead” entries to re-enter the pool of mining rewards, on top of the normal halving mechanism which normally occurs every 210,000 blocks. Sakura comes with a trigger condition that “dead” UTXO entries should represent more than 50% of the UTXO set. This condition prevents a premature activation of the change of consensus if there is not enough economic gains to justify the change. Sakura proposes an exponential decay mechanism associated to a half-period of roughly 20 years. The paper also presents a discussion to justify why those seemingly arbitrary choices are made.
It proposes a much longer “recycling period” (80 years).
My fundamental challenge would be: this assumes many entities will continue willingly wasting money by leaving unused UTXOs around. I’ll be surprised if in 20 years UTXOs are so easily lost as they were in Bitcoin’s early years. If software continues improving, I think it’s more likely that the future has fewer “dead UTXOs” than today.
If my great-grandchildren find I was wrong on this prediction, maybe that would be a good time to introduce a recycling mechanism for the long-term.