The receipts, and how to check every claim in the post above. The rules, the engine and the sound are on chain in the clear; the record itself is on chain as a hash until a claim carries its bytes. Nothing here requires trusting me, which is the point.

The inscriptions:

AUDIO    cec29fd094fed5214f2ff191e0bfa52b466581a29a6a741c53c705b13e8c522f
ENGINE   1d4a7785e0ac9e4584722212398e0e676ddc65ada3cb1f1432ff44ab3dce8273
PARAMS   77aaaa6b0eeb76b4387036bfda1c88a8ad07bb35ab9b8c9ba3d4d99ecae8c681
PAYLOAD  sha256 a054f358e1949abfbfbdc1f08200e3e3ea78c26319663da3317641b8562f025a
GENESIS  a3776fb529a1d45b29998649af91384d3232b387d24ceb0c660e04b78f70b1e5

The ordering proof — the part I care most about. The edition's geometry (which seat gets which bytes) is computed from the payload. So a fair edition has to prove the rules weren't tuned to the payload after the fact. Arithmetic can't prove that. Order can. PARAMS — the rules, in plain text — had to confirm below a block height it itself declared (961,803). It confirmed at 961,786. The payload is bounded by a timestamp cut-off, the art by a manifest hash, and the assembly is deterministic: same inputs, same bytes, rebuildable by anyone. The rules were on the chain before the thing they govern existed.

Seat assignment. regionLen = floor(payloadLen / 256) — derived, no free parameter. A seat's offset is computed from hash256(payload) and the seat's index; your shard's individual form then descends from your claim's own txid. Could you target a favourable seat? Partly — and I'd rather say so than sell you an adjective. You cannot grind one: the position is seeded from the commit frozen at deploy and the covenant's own seat index, so nothing you supply touches the draw. But every seat's slice is computable from genesis, so you can wait for one you like — and a losing fee conflict never confirms, so trying to take a marked seat off someone costs nothing. The covenant does not prevent that. What I did instead was measure what it is worth: at this edition's size the best seat's edge is worth about half of what chasing it costs. At a nearly-empty edition that margin narrows and can tip positive — but you cannot know at claim time whether two people or two hundred will follow, and the entire seat-quality gradient across the whole edition is worth roughly two US cents. The rational move is to just claim. The derivation rule, the collision fallback, and its fixed search bound are all stated in PARAMS — inscribed before the payload existed, which is the only moment stating them proves anything.

The ticket → shard chain. Your claim pays 260,000 into the pool and writes your ticket at vout 1 — a 1-satoshi ordinal output bound to your key; spending it is the mint that produces your shard, and only your key can. Be clear about who enforces that. The claim is signed SIGHASH_SINGLE, so it commits output 0 and cannot see output 1 at all. The covenant does not require a ticket — the claim page builds it and refuses to broadcast a claim without one. So this is an edition of 256 seats and up to 256 shards: a claim hand-built without a ticket would still take its seat and produce nothing. I'd rather tell you that here than let you find it in the script. The mint covenant bakes the ENGINE's txid and its subresource-integrity hash into the shard — so the artwork your browser runs is the exact 33,813 bytes on chain, or it refuses to run at all. The audio rides the same way: one recording, one engine, every shard reaches them by txid. Any gateway can serve it. None can alter it. None of them is me.

The economics, plainly. The pool = my lock + every claimed seat's 260,000. From the deadline (height 964,082) the settle becomes broadcastable — by anyone, not just me — and the edition closes the moment it executes. I'll settle at the deadline; the point is that nobody has to trust me to. The settle disperses the whole pool across the seats on a proximity gradient: each seat's weight falls with its distance from the work's centre, so near seats draw multiples of a far seat's share and the far edge draws down to the dust floor. No raffle, no single winner — the gradient is deterministic from the deployed geometry, and the weight arithmetic is public the moment the edition exists. My lock pays the settle's own mining fee. Nothing in the pool flows to me — and my no-seat posture is stated in PARAMS §5, dated before the first claim, checkable against every claimant address forever. The real price of a seat is the carrier: your claim hauls the full record onto the chain, ~257,242 sats of mining fee, every copy a redundant public witness of the whole work.

Check it yourself (PARAMS carries the full list): confirm PARAMS' height < its fence · recompute floor(payloadLen/256) · recompute the art manifest hash · read the resolution transcript · read §5 and judge it the way promises are judged. If any of these fail, the piece is off-spec and you can prove it without my help. That is the entire point.

Questions welcome — especially the adversarial kind. 🌱