Lens demo · multi-foton output

One figure, produced three times

A. Analyst · QC (independent) · External lab — same data, same script, three independent runs

The plot below has a plankton lens. Because these exact bytes were produced by three independent fotons — the analyst's run, QC's re-run, and an external lab's — its badge shows ↻3: the figure isn't just verifiable, it's corroborated. Hover the lens for "reproduced · 3 fotons · 3 signers"; click it to open the kton world. The badge is resolved by hashing the plot and fetching one small index file from a static mirror — the whole registry is never loaded, so this scales to any number of records.

A concentration–time profile for a test compound. Nothing here asks for trust: the bytes re-hash to a recorded output, and three separate parties independently produced the very same bytes. The more independent producers a file has, the harder it is to doubt.

Figure 1. Concentration–time profile. Independently produced by the analyst, QC, and an external lab — identical bytes, so the lens reads ↻3. A figure only one team made would read ✓ with no counter.

A rainbow table of every calculation — what it would cost

Each execution is a file named by its own hash, so the mirror is just static, immutable, auto-deduplicated blobs. Serving one is the same operation as serving five billion — a content-addressed GET. There is nothing to "scale": it is a storage problem, and storage of cacheable blobs is the cheapest thing on the internet. Numbers below are measured from this demo (~1.2 KB per signed record, ~⅓ index overhead) and extrapolated:

1 billion

recorded executions — a foton for a billion scientific calculations

≈ 1.6 TB

total static files: 1.2 TB signed records + 0.4 TB rainbow-table index

≈ $10–24 / mo

storage — Backblaze B2 ~$10, Cloudflare R2 ~$24 with $0 egress

≈ $0

per additional viewer — immutable + content-addressed ⇒ cached at every CDN edge

Serving unlimited people is essentially free: every object is immutable and addressed by its hash, so it caches perfectly at the network edge — the origin serves each object roughly once per edge, not once per person. A figure seen by one scientist or by ten million costs the same, and identical calculations run anywhere deduplicate to a single stored object. So at this scale a public, verifiable record of a billion reproductions is affordable to a single hobbyist — for the price of a coffee subscription. (The lookup index is ordinary infrastructure, not part of the protocol; the one thing that isn't free is knowing the full set to index — completeness — which lives in the ordered nekton layer, not in this rainbow table.)

Rendered with the kton lens · a plot with no badge is simply unknown to the connected planktons. The counter is the number of distinct fotons that output these exact bytes — reproduction made visible.