Proof of Existence: How Cryptographic Timestamping Protects Research Before Publication

There is a period in the life of every research finding during which it exists but cannot be spoken about. The result is solid, the protocol works, the dataset is clean — and the manuscript is in review, where it will remain for some time. Journal publication cycles routinely run between twelve and thirty-six months from submission to public availability. For the group that produced the work, that interval is not an administrative pause: it is a period in which the most valuable thing they own is the thing they are least able to use.

This article sets out in mechanical terms what cryptographic timestamping does about that problem: what a hash written to a public ledger establishes, why the underlying data never leaves the institution, and — with equal weight, because credibility depends on it — what a timestamp does not establish. A proof of existence is a narrow, well-defined piece of evidence. It is not a patent, and no enthusiasm about the technology makes it one.

The secrecy trap

The logic that produces the embargo gap is rational at the level of the individual group. A protocol that took four years to optimise can be reproduced by a well-resourced competitor in months once its details are known. Priority in science is established by publication, and publication is slow. So until the paper is out, you say nothing specific to anyone who is not a co-author.

The cost of that silence is rarely accounted for. During the embargo gap, a group forgoes:

• Conversations with industrial partners, who need technical detail to assess fit and will not commit on a vague description
• Investor discussions, where diligence requires disclosure long before any term sheet exists
• Conference presentations, which are how most collaborations and most recruitment begin
• Licensing discussions for methods, reagents or early-stage protocol access that could fund the next phase of the programme
• Funding calls whose deadlines do not wait for an editor’s decision

Silence is treated as the free option. It is not. It is an insurance premium paid in foregone partnerships, every month, for up to three years. The trap is structural: to have a useful conversation you must disclose enough to be interesting, and to disclose safely you need a dated, tamper-evident record that predates the conversation. Most institutions have nothing beyond an internal notebook, an email chain or a file server timestamp — all held by the party asserting the claim, and all therefore contestable.

What a hash actually is

A cryptographic hash function reads a file of any size and produces a fixed-length output. SHA-256, the function used here, produces 256 bits, conventionally written as sixty-four hexadecimal characters. Four properties make that operation useful as evidence.

Deterministic

The same input always produces the same digest, on any machine, in any year. This is what makes verification possible at all: a reviewer in five years’ time recomputes the digest and gets the same sixty-four characters.

One-way

The function cannot be run backwards: given a digest, there is no feasible method of recovering the file that produced it. The digest of a 400-page protocol and that of a sequencing run are identical in form and reveal nothing about their contents — not the subject matter, not the file type, not the size.

Avalanche behaviour

Changing a single character — one decimal place in a concentration, one swapped figure, one edited sentence — produces a completely different digest, bearing no visible relationship to the original. There is no such thing as a nearly matching hash.

Collision-resistant

Finding two different files that produce the same SHA-256 digest is not achievable with any known technique. This prevents anyone constructing an alternative document that matches an existing proof.

One honest caveat: because hashing is deterministic, anyone who can guess the exact input can confirm the guess by hashing it — a real limit for short, predictable strings. For a protocol document or a dataset of any substance, the space of possible inputs is far too large for guessing to mean anything.

Claim	Established?	Why
A specific file existed at a specific time	Yes	Its digest is recorded in a block with a known timestamp
The file has not been altered since	Yes	Any change produces a completely different digest
Who authored the underlying idea	No	A hash binds a file to a time, not an idea to a person
A granted patent right	No	Priority is a matter for patent offices and your technology transfer office
Freedom to operate	No	Unrelated to what a timestamp records

The ledger as notary

A digest on its own is just a number. Its evidentiary value comes from being fixed in time in a way the person asserting the claim cannot later influence. The traditional analogues — a sealed envelope posted to oneself, a countersigned notebook page, a notarial deposit — work on the same principle, and each depends on a custodian whose continued existence is assumed and whose records are, in practice, alterable.

Writing the digest into a public blockchain transaction removes the custodian from the argument. The transaction inherits the block’s timestamp, is replicated across independent nodes, and cannot be retroactively reordered or amended by whoever submitted it. DimenChain records these proofs on Arbitrum, an Ethereum Layer-2 network. That choice is practical rather than ideological: the per-proof cost on a Layer-2 suits an ordinary departmental budget, so a group can timestamp every version of a protocol, every intermediate dataset and every milestone rather than rationing proofs to whichever documents seem important at the time. Every transaction is publicly inspectable on Arbiscan.

Verification is deliberately unremarkable: whoever holds the original file recomputes its SHA-256 digest and compares it with the value recorded on-chain. If the two match, the file existed in exactly that form at or before that block; if it has been altered in any respect, the digests differ and the claim fails. The proofs are WIPO-compliant Proof-of-Existence records, and the protocol underlying them is itself patent-protected under WIPO PCT/IB2024/058791.

A verifier needs nothing from us — only the file, a hash function and a block explorer. A proof that depends on the continued cooperation of the company that issued it is not a proof.

What the proof establishes, and what it does not

A timestamped hash supports four assertions: that a particular file existed no later than a particular moment; that it has not been modified since; that versions and milestones fall in a given sequence relative to one another; and that the record is associated with the cryptographic identity that submitted it.

It does not do the following, and any vendor implying otherwise deserves scepticism:

• It is not a patent. It confers no exclusive right and no protection against a third party filing for and obtaining a patent on similar subject matter.
• It is not proof of authorship of an idea. It proves a file existed and that a particular key submitted it. Conception, inventorship and authorship are separate questions decided under separate rules.
• It says nothing about whether the content is correct, novel or reproducible. A rigorously timestamped error remains an error.
• It is no defence against genuine independent invention. A group that arrives at the same result without reference to yours is unaffected by your record.
• It is not a substitute for legal advice. It is evidence, and how a tribunal, examiner or counterparty weighs evidence depends on jurisdiction and circumstance.

This is why the technology transfer office belongs in the conversation from the beginning rather than after the fact. The rules governing disclosure, priority and grace periods differ substantially between jurisdictions, and speaking publicly about unfiled subject matter has consequences no timestamp reverses. What a proof of existence gives your research office is a firmer basis for that decision, and a dated record to point at if the question is contested.

One point is worth raising with patent counsel explicitly, because it is frequently misunderstood. Recording a hash is not a disclosure of the underlying work: the digest reveals nothing, so notarising does not make the subject matter available to the public in the sense patent examination is concerned with. That is a favourable property, but it remains a legal question in every jurisdiction where you intend to file, and should be confirmed rather than assumed.

Why raw data never leaves the institution

The hash is computed on the file where the file already lives. What is written to the ledger is the digest, encrypted metadata, and the events emitted by the governing smart contract. Sequencing output, patient-derived data, instrument logs and analysis code remain in institutional storage under the access controls that already govern them. Adopting a timestamping layer changes nothing about who can read your data.

This is not incidental to the design. The immutability that makes a ledger useful as evidence is in direct tension with the rights of erasure and rectification European research institutions must honour; writing personal data to a permanent public record would create a problem no later process could resolve. Keeping content off-chain avoids that conflict rather than managing it. DimenChain is operated from the Netherlands and built to align with the GDPR, and the architecture is why that alignment is achievable. Your data protection officer should still review the deployment; the point is that the review is short when the honest answer to “what leaves our network” is a sixty-four-character number.

What a team can do while the manuscript is still in review

The practical effect of a defensible record is that the disclosure calculation changes. Consider — as an illustration, not a case study — a group with clean data and a manuscript in writing that wants an industrial partner now rather than in two years. Once the protocol, the dataset and the analysis pipeline carry timestamped proofs, that group can go into a meeting and describe what it has in enough detail to be assessed. If a counterparty later produces something conspicuously similar, there is a dated, independently verifiable record predating the meeting. That guarantees no outcome, but it turns a dispute over recollection into an argument about evidence.

The same record supports presenting at a conference before the paper appears, opening licensing discussions for early-stage protocol access rather than waiting for publication to make the method free to all, and structuring multi-site collaborations where each party’s contribution is documented as it is made rather than reconstructed from memory. It also makes it possible to place value on work that never becomes a paper — negative results, abandoned routes, protocols that did not deliver for their original purpose but retain value to somebody else. That material is normally lost. With a proof of existence attached, it can be licensed.

Attribution within the collaboration

Priority against outsiders is one problem; credit inside a consortium is another. Contributions recorded through the platform are recognised with Soulbound Tokens — non-transferable records held by the contributor’s identity, which cannot be sold, traded or reassigned. The non-transferability is the substantive part: a reputation record that can be bought is not a reputation record. This does not override institutional authorship policy and is not intended to. It gives that conversation something factual to work from.

Where this belongs in existing practice

Cryptographic timestamping does not replace the electronic lab notebook, the institutional repository, the disclosure process or the technology transfer office. It sits underneath them as an evidentiary layer, answering one narrow question — did this file exist at this time — without requiring anyone to take the claimant’s word for it.

The embargo gap is a structural feature of scientific publishing and shows no sign of shortening. What can change is what it costs to sit inside it. When a proof is cheap and takes seconds, the reasonable default is to timestamp early, timestamp every revision, and stop treating silence as free. Then talk to your research office about what you can now say out loud.