The content on our servers can be removed. The cryptographic anchors on Bitcoin cannot, not by us, not by the court, not by any successor to either. So if one still holds the copy, the fact of the investigation can't be refuted.

The chain does three things at once: it records what was claimed, who claimed it, and what they cited in claiming it. Every entry on the ledger carries the cryptographic signature of the investigator who entered it and the fingerprint of the source they drew from, a document, a testimony, a photograph, a filing. The reader can see, at minimum, who said what was so, and what they relied on in saying it.

Truth-of-claim, whether the document tells the truth, whether the witness can be believed, remains the reader's work, as it has always been with any record. What the chain restores is the historian's minimum: a known author, a known source, and a date that cannot be moved.

A small one. Founded in 2024 by veterans of intelligence work and academic research in agent systems, in answer to a condition our generation has not yet named clearly: that the institutions on whose record the public's life depended have, one after another, stopped keeping one.

The press has been killed in numbers exceeding every prior conflict combined. Archives have been dissolved by court order in the country with the oldest tradition of resisting such things. Two hundred and forty-one news organizations are actively blocking archival mechanisms. We are building AI systems that can be manipulated to recursively misrepresent reality, and we increasingly rely on them as the ultimate source of knowledge. Some fifty million human beings are at any given moment held in forced labor or forced marriage, the majority of them children, the majority of them entered into no register because none was ever built. Law enforcement is occasionally subjugated to working against its own original ethos.

We sell a paid platform to investigators, journalists, prosecutors, civic researchers, due-diligence firms, and we use the revenue to keep open an evidentiary substrate of the corruption or civic harms that no one else, at this moment, is keeping open.

Palantir is the instrument by which the institutions of state and capital see further into the lives of their subjects. Symphoria is the instrument by which the subject can see, and keep seeing, what those institutions did. The direction of the asymmetry is reversed. Palantir does not publish what it ingests; we publish what we record, by definition, into a decentralized chain we do not control. Whether the substitution is sufficient is for the reader to judge, but the parties served by the two systems are not the same parties.

Where the subject's privacy interests are real, we are bound by the same laws of evidence, the same prohibitions, the same standards of justification as any institution that traffics in the names of human beings. The chain does not exempt us from those duties. It holds us to them, since the record of what we did is itself on the chain.

The twentieth century turned in large part on the question of centralized versus distributed economic planning, and the case for the latter, that free agents holding local knowledge, with the right to acquire and dispose, produce better aggregate outcomes than any central planner, was, by the century's end, demonstrated by history. Whatever one's politics, the epistemic lesson held: distributed systems are better than centralized ones at solving problems no single mind can hold in full.

The question now before us is whether the same logic applies to justice, to intelligence, and to the public's adjudication of what is so. The centralized institutions on which we have relied for those functions, state archives, the official press, intelligence services answering to one capital, have not produced a record durable enough to support the judgments a free people must make. We are building the wager that the same answer works again: that a community's ability to find the truth, matched to the state's through open-source technologies any reader can audit and any peer can run, produces a better civilization than any single authority's monopoly on it.

The freedom and proprietorship that distinguished the economic argument apply with equal force to the epistemic one. The single authoritative archive failed because it was single.

The objection is true and unanswerable as stated. What we are building is not a substitute for action; it is a substrate beneath it. In the five years from 2020 to 2025, more journalists have been killed than in the two-decade span CPJ documented for the Iraq and Afghanistan wars combined. The Gaza war alone, since October 2023, has killed more journalists than any other conflict in CPJ's history — over 260, the majority Palestinian. Some left behind testimony their employers later muted, archives the platforms later pruned. Their work was not failed by their unwillingness to act. It was failed by the disappearance of what they had already produced.

The record is not the labor of the witness; it is the condition under which the witness's labor remains useful after the witness is gone. To say "we lack the will to act" and "we lack the records to act on" are answers to different questions. Both are true. We are responding to the second because we are equipped to.

Every register has been used both ways, and ours will be no exception. The Dutch civil registry, which had recorded the religion of its citizens for centuries with no malign purpose, became the instrument by which the occupying force in 1940 located the country's Jews. The genealogical records of Rwanda became kill lists in 1994. We have read these histories.

Our answer is not to refuse memory, which would foreclose the instrument's use for good, but to design the chain so that allies can verify the signature of the originating party, and verify the truth only if they also hold the copy. Adversaries can see the hash chain on Bitcoin, but without the original copy it is impossible to decode it into meaningful information. We are building a chain of verification and a network for operational security.

Most of what the public remembers about the work of intelligence is what was done in its name to civilians abroad, and a substantial part of that memory is fair. The American services in particular have, by their own internal accounts, presided over more civilian killing than any other set of institutions in the last century. We did not invent that record and we do not justify it.

What we know from inside that world is what it looks like when a record is created the public will never see; what it looks like when a record is destroyed because its contents would change the public's mind about an action taken in their name; what the operational conditions are under which a small group convinces itself the world is better off if certain things are not written down. That knowledge is what the work of Symphoria refuses.

Whether it makes us trustworthy or untrustworthy is the reader's to decide; whether the work itself is necessary does not depend on us.

Because the design refuses what the design refuses. Memorial was a small group with an archive, and the archive could be closed because it sat on premises a court could enter. WikiLeaks was a small group with files, and the files could be hounded because they passed through servers whose providers could be pressured. The ledger we are keeping has neither of those weaknesses by construction, not by promise.

The hash of every record is published into a network operated by no one and verified by everyone; the keys that signed each entry are public; the chain head, once anchored, is as durable as the network that carries it. If our team is shut down tomorrow, if our domains are seized, our servers locked, our directors prosecuted, the entries on Bitcoin remain readable to anyone who has a copy of the bundle, and the work can be resumed by anyone with the will to resume it.

The answer is not "trust us." The answer is "we have made ourselves removable in advance, and the record will outlive our removal."

Journalism

Source verification. Evidence recovery. Stories that can't be silenced.

Legal

Evidentiary standards for a digital age. Chains of custody that hold.

Democratic Memory

Records that outlive those who would erase them.