# Main Street: More Lookouts Than Miners

October 2026. ProvenPaid market note no. 3, by Justin Smith, Ph.D. Measured October 12 to 27 from daily unpaid probes and public Base data. Previous note: https://api.provenpaid.com/notes/who-actually-pays

Every boomtown has a Main Street: the outfitters, assay offices and general stores that open before the miners arrive. The first note found more shops than miners in the x402 town. The second took a few of those shops to the assay office and weighed who actually paid them. This one pulls up a chair on the porch and watches the street for three weeks: the same shops checked every morning for 16 days, the same tills weighed every day for 9.

We also watched our own front door. ProvenPaid is a shop on this street too, a paid x402 service that has been open for three weeks, and it turns out most of the folks walking past are lookouts, not miners.

Three findings:

- **Main Street held still.** We watch 45 x402 endpoints. In 16 days none changed the wallet it gets paid to, one seller cut prices on three endpoints, and one endpoint went dark on the last day.
- **The miners didn't rush in.** At the four sellers we can weigh in full, rolling 30-day payer counts went from 76 to 83 wallets in a week, and six of the seven newcomers walked into one shop.
- **The lookouts outnumber the miners.** In 19 days, 129 differently named bots asked ProvenPaid's price 7,779 times. Not one outside wallet has paid. We're a lookout too, and our own till reads zero.

## Main Street held still

Every morning ProvenPaid knocks once on the door of every x402 endpoint it knows about, without paying, honouring robots.txt, and records what comes back: whether the payment challenge is valid, the price, the network and the receiving wallet.

| | Oct 12 | Oct 27 |
|---|---|---|
| Endpoints watched | 41 | 45 |
| Answered the probe | 41 | 44 |
| Valid x402 challenge | 34 | 37 |
| Base receiving wallets behind the valid ones | 5 | 4 |

Across 16 days and about 690 probes:

- **No seller changed where it gets paid.** Every endpoint kept the same receiving wallet on every network it offers.
- **One price cut.** Seller B (as named in the last note) cut the price of three endpoints from $0.002 to $0.001, on October 14 and 19.
- **Four new endpoints, all from a seller already listed.** A weekly re-check of the directory snapshot found them on October 19.
- **One endpoint went dark.** It answered every day until October 27, when the connection failed. One day is not churn yet; it stays on the list either way.
- **Sellers hedge across rails.** Besides USDC on Base, 18 of the 44 endpoints that answered on October 27 also accepted USDC on Solana, and some list Polygon or Lightning options.

A quiet street is good news for anyone building on these endpoints: the shops you find today will likely be open next week, at the same price, paid to the same wallet. It also means the town isn't sorting itself out yet. Nobody packed up, nobody struck it rich.

## The miners didn't rush in

Since October 19, ProvenPaid has weighed each shop's till every day: every Base receiving wallet behind a valid endpoint, over a rolling 30-day window, with the same method as a paid check.

| Seller | Payers, 30 days to Oct 21 | Payers, 30 days to Oct 27 | Returning payers, Oct 27 | Payments, Oct 27 |
|---|---|---|---|---|
| B | 62 | 68 | 17 | 329 |
| C | 12 | 12 | 10 | 118 |
| D | 2 | 3 | 0 | 3 |
| No-payer seller* | 0 | 0 | 0 | 0 |

Sellers lettered as in the last note. Seller A, the busiest saloon in town, is left out: each day's scan covers only part of its 30 days (43% on October 27), so its counts rise as we see more of the window, not necessarily because more miners walked in. *Its October 27 figures are from October 26, the last day its endpoint answered.

What changed:

- **Seller B gained six payers, and its 30-day payments jumped from 200 to 329 in a single day (October 26).** Its top payer's share fell from 36% to 22% while the top five's rose from 63% to 67%: a few wallets stepped up at once. Its price cut came a week earlier; the chain can't tell us whether one caused the other.
- **Seller C kept its regulars.** Twelve payers all week, ten or eleven of them coming back. A small shop with a loyal crowd.
- **One shop never made a sale.** Its door opened properly every day from October 12 to 26, with a valid paywall, and no x402 payment came in 30 days. On October 27 the door didn't open at all.

A rolling window means a flat count can hide comings and goings: new miners arriving as old ones drift off. On these four shops there was little of either.

## The lookouts

ProvenPaid opened its doors on October 9. Every request gets tallied by what happened and by the name the visitor gives itself. No IP addresses, no record of which URLs anyone checks. Here's the view from our own porch, October 9 to 27:

![Bar chart: 9,305 requests for ProvenPaid's price, mostly from named bots doing uptime checks, directories and trust services. Outside wallets that paid: zero.](/brand/notes/lookouts.svg)

- **9,305 knocks, no sales.** That many times, somebody asked what a check costs. No outside wallet paid. One paid check came through in three weeks, and it was our own launch test. Thirteen free checks went out, several of them ours too.
- **129 riders, each with a name.** Self-named bots made 84% of the knocks, and their names read like a frontier business directory: health and uptime checkers ("probe", "liveness", "monitor"), directories and censuses ("index", "census", "catalog"), and trust services ("trust", "verify", "vet"). Sorting is by name only; a name is a self-description, not proof.
- **A few ride past all day.** The ten busiest named bots made almost two thirds of the knocks. The busiest asked our price about 85 times a day.
- **Even the Bazaar's own crawler drops by,** which is exactly what you'd hope for from a listing.

This is one shop's porch, not a census of the town. But it rhymes with what the first two notes found: more people are building tools to watch x402 commerce than are buying through it yet. ProvenPaid included. We are a lookout too, and we're saying so plainly.

## What it means for the town

- **For agent builders (the miners):** the shops are stable enough to build on. What a storefront can't tell you is whether anyone else trades there. Check that before the first payment to a new seller.
- **For the lookouts (platforms, directories and monitors):** a lot of riders are watching the same street. If you run one of them, a daily feed of payer depth for every Base receiving wallet we track may be cheaper than riding past every shop yourself. We are looking for one or two founding partners; the first 30 days are free. Write to contact@advancedanalyticsgroup.org.
- **For shopkeepers (sellers):** want your till weighed by someone who isn't you? Your first independent payer-depth report is free. Same address.

Try a check: `uvx provenpaid-mcp` from Claude or another MCP client (free checks, no funds needed), or `GET https://api.provenpaid.com/v1/check?url=<endpoint URL>` for $0.01 over x402.

## Method and limits

- **Endpoints:** the 45 x402 endpoints found from the public x402.org directory snapshot of October 9 and weekly re-checks of it. Not the whole market: platforms that see facilitator traffic list far more sellers.
- **Probes:** one unpaid request per endpoint per day, honouring robots.txt. A failed connection on one day is recorded, not treated as churn.
- **Payers:** public Base USDC logs, one settlement per EIP-3009 authorization, payers who were paid back by the seller set aside (none were, this week). Rolling 30-day window, recomputed daily. Partial scans are labelled and left out of trends.
- **Traffic:** ProvenPaid's own request counters, October 9 to 27 (UTC days): outcome and the client name from the User-Agent, never IP addresses or checked URLs. Client names are self-declared and grouped by keywords in the name; one operator might run several names, or several might share one.
- **Sellers are anonymised.** This note reports evidence, not ratings.
- Full methodology: https://api.provenpaid.com/methodology