The 50-to-1 Compression That Makes Paywalls Profitable

HMAC receipts turn one onchain settlement into fifty cached reads. Without compression, the math doesn't close. With it, the margin stays north of 99%.

A naive per-request paywall settles every single hit onchain. Even on Arc, at $0.00002 gas per tx and $0.001 revenue per read, that's a 98% margin — good, but not great. The moment your article gets hit a thousand times in an hour by the same agent (think a long-form piece being summarized by a chatbot), the onchain fees still stack up even if each one is tiny.

HMAC receipt caching collapses that. The first request settles onchain. The response includes a short-lived HMAC receipt token, signed by the publisher's per-site secret, encoding the agent's identity, the tier, and a 5-minute expiry. The next forty-nine reads from that agent present the token and skip the facilitator entirely. Fifty-to-one settlement compression.

The compression ratio is tunable. Five-minute TTL is the default — long enough to cover bursts of scraping activity, short enough to force periodic re-authorization. Publishers can dial this per-site. For extremely high-volume paths, a longer TTL amortizes more reads. For sensitive paths, a shorter TTL enforces more frequent cryptographic re-proof.

The invariant is what matters: one onchain tx unlocks N cached reads, where N is large, and the net margin stays above 99% across the full window. Tollgate is the first paywall where this math was not just possible but required by construction. You cannot build this on Ethereum mainnet. You cannot build this on a chain with volatile gas. You build it on Arc.