Does Bitcoin's fee market price permanence, or just congestion?

The debate regarding Bitcoin's fee market has surfaced a critical question: Does it accurately reflect the lifelong cost of permanent data storage on its network? This arises particularly in light of the SegWit update, introduced by BIP-141, which significantly reduced the weight of witness data. Originally aimed at resolving transaction malleability, this change inadvertently made large-scale use of inscriptions like Ordinals and Runes more economically feasible. The core issue here is whether the revised weight formula effectively prices the addition of permanent state to the blockchain.

A detailed UTXO cost model was developed to study this scenario, utilizing publicly available data. The findings from this model show that the annual operational cost for a full node is approximately $925, breaking down into hardware, bandwidth, and electricity costs. Meanwhile, the average inscription data amounts to around 400 bytes, translating to an estimated lifetime storage cost of about $0.008 per inscription, assuming a 10-year lifespan. However, this introduces unpriced externalities which, at a projection of 100,000 inscriptions per month, would result in around $9,200 per year spread across all nodes. The prevailing fee ranges from approximately $0.06 to $50, varying with network activity levels. These calculations are detailed further with source code and a verification appendix available at github.com/prateekposwal/bitcoin-priority-oracle.

The current fee structure primarily addresses congestion—pricing immediate access to the next block rather than ongoing storage costs. This results in a mismatch: fees are one-time payments chosen by the payer, while storage costs are recurrent and involuntarily borne by future node operators over an indefinite period. The SegWit discount exacerbates this by making the addition of data to the UTXO set cheaper compared to standard transactions, thereby questioning if the economic impact of 'data permanence' warrants protocol-level reconsideration.

Arguments supporting protocol changes point out that unpriced externalities typically lead to overconsumption beyond what is socially optimal. This, in turn, could increase node operation costs as the UTXO set grows. On the other hand, counterarguments highlight the declining cost of storage, the voluntary nature of node operations, and the relatively minor cost impact per node annually, which is roughly $0.18 when spread across approximately 50,000 nodes.

This discussion does not aim to draft a proposal but rather seeks to identify whether there exists a significant gap between what the fee market prices and the costs node operators collectively bear. Feedback is particularly sought on the accuracy of the 10-year UTXO lifespan assumption, the proportion of pruned versus archival nodes, and whether the SegWit weight formula should be revisited as a mechanism for pricing state growth. Further research and analysis can be found at the provided repository link, which includes a full model, verification appendix, and sensitivity analysis.