scry.io · Principles
Operating principles
Turing-complete internet research: make the public internet computationally traversable.
Published 2026-08-28
The public web should be computable. Not skimmed five links at a time, not rationed into snippets — computable: every row that matches, and the answer worked across the whole matching population. Scry holds large archives of publicly published material and computes over them; what we sell is the computation, never a copy of the work. These principles govern what we hold, what we compute, what we return, and what we will not. They are written to be checked against the deployed surfaces, not admired.
Public is not public domain
Being publicly accessible does not strip a post, paper, or article of its author's rights, and it does not place it outside privacy law. The words remain their authors'. What we build and sell is computation over them — search, aggregation, verification, provenance — a research system by design and by license, not a substitute corpus for the originals.
Maximum computation, minimum substitution
Complete corpora are held internally so questions can be answered completely: the full corpus, the true denominator, every matching row. Output stays query-shaped — results that answer a question, not reconstructions of a source. The terms carry the specific prohibitions, and the service is engineered so wholesale extraction is slow and expensive by construction.
Provenance travels
Rows carry canonical source, source timestamps, observation time, and processing metadata, and query responses report exactly what ran and what it touched. We keep what a source said distinct from what we computed, and we say which fields support each claim.
What is served is documented completely
The live schema registry is the coverage denominator: an enabled relation is available, an omitted one is not, and no unadvertised route exists as a fallback. Acquisition infrastructure is source-specific and not disclosed; the service is described by what it serves.
Archives spare the origins
The point of holding an archive is that answering a million questions lands the load on our hardware, not on the sites where the material first appeared. Cache once, compute locally.
People get agency
Removal, correction, and copyright channels are documented at /legal/removal, and each request is evaluated under its own standard. A verified removal holds in every later snapshot. Archive integrity is also a value we defend: suppression is recorded, never silently rewritten into a past that did not happen.
Hosting is not ownership
Platforms operate systems; they do not own their users' public speech. When publicly posted material becomes enclosed, something real is lost. We defend the legitimacy of reading, searching, and computing over what people published to the world.
Questions
Write to [email protected].