I’ve been working through a question that I think is genuinely open, and I’d like feedback from this community on whether the framing is useful.
Question: Does Bitcoin’s fee market correctly price the lifetime cost of permanent data storage?
Look, BIP-141 (SegWit) created a 4× weight discount for witness data. This was designed to fix transaction malleability, not to price state growth. The discount made inscriptions (Ordinals, Runes, BRC-20) economically viable at scale. This was an unintended side effect of a different design goal, and nobody has seriously asked whether the weight formula appropriately prices the cost of adding permanent state to the network.
I built a UTXO cost model using publicly available data. Source code and full verification appendix.
The fee market prices congestion, the cost of getting into the next block. This works well: during high demand, fees rise, and the most valuable transactions get confirmed first.
But storage cost is different:
- Fee: One-time payment → Storage: Recurring annual cost forever
- Fee: Payer chooses → Storage: All future node operators bear it involuntarily
- Fee: ~10 min horizon → Storage: Indefinite horizon
The SegWit discount amplifies this: inscription data in witness pays 1/4 the weight of non-witness data, making it cheaper to add permanent state to the UTXO set than to send a standard financial transaction.
The open question Is the “data permanence externality” economically significant enough to warrant a protocol-level response?
Arguments for “yes”:
- Unpriced externalities lead to overconsumption (the quantity of permanent-state transactions exceeds the socially optimal level)
- At scale, UTXO growth increases node operation costs
- The SegWit weight formula was never designed as a state-pricing mechanism
Arguments for “no”:
- Most full nodes are pruned, storage is cheap and getting cheaper
- The fee market already caps inscription volume during congestion (when blocks are full, inscriptions must compete with financial transactions)
- Node operators choose to run nodes voluntarily; the cost is visible and accepted
- ~$9K/yr across ~50K reachable nodes is ~$0.18/node/yr, that is negligible
What this is NOT
- Not a proposal. Not a BIP draft. Not claiming the problem is urgent.
- Just a question about whether there’s a gap between what the fee market prices and what node operators collectively bear.
Would love feedback from people who have thought more deeply about this — particularly around:
1. The 10-year UTXO lifetime assumption is there better data?
2. The pruned vs archival node ratio most estimates I’ve seen suggest ~30% of nodes are archival. If that’s wrong the externality changes significantly.
3. Whether the SegWit weight formula was ever intended to be revisited, or if it’s considered settled architecture.
Research repo with full model, verification appendix, and sensitivity analysis:
[ bitcoinsahi.com ]