A California appeals court voided an arbitration clause because the company couldn't prove who actually clicked "sign" in DocuSign. They had the audit log. They had the Certificate of Completion. The court said: not enough. That was 2019. Courts are still saying it.

The e-signature industry has sold the world on a convenient conflation: that "legally valid" and "cryptographically provable" are the same thing. They are not. One is a statute from 2000 that says clicking a button counts as a signature. The other is math that no one can dispute. We've been using the first and calling it the second.

I've spent thirty years watching "enterprise-grade" turn out to mean "enterprise-priced." At MSNBC in the late nineties we had audit logs for every content change in the Workbench CMS. They were database rows. We could modify them. Any sufficiently motivated insider could have and probably nobody would have noticed. "Immutable audit log" is a marketing claim, not a technical property, unless someone has made it hard to change through external anchoring and cryptographic proof that lives outside your systems. Most e-signature platforms haven't.

The E-Sign Act Is Not What You Think

The Electronic Signatures in Global and National Commerce Act, passed in 2000, did one thing brilliantly: it established that electronic signatures are legally valid. Full stop. A typed name, a drawn squiggle, a checkbox — all legally equivalent to a wet signature for most documents.

What the E-Sign Act conspicuously did not do: establish any security standards. No authentication requirements. No audit trail specifications. No mandatory cryptographic evidence. The law says electronic signatures are valid; it does not say how you have to prove one was made.

Compare this to the European Union's eIDAS regulation, which creates three tiers: simple, advanced, and qualified. A qualified electronic signature requires a cryptographic certificate issued by a government-accredited trust service provider. It is independently verifiable without calling the vendor. The US equivalent for "qualified" is: nothing. There is no equivalent.

So the E-Sign Act gave us legal validity on the cheap — any click-wrap "I agree" carries the same legal weight as a PKI-signed certificate. For 99% of agreements, that's fine. For the 1% where someone actually disputes the signature in court, it creates a very specific problem.

Your Audit Trail Is a Database Record

When DocuSign sends you a Certificate of Completion, you get a PDF. It contains IP addresses, email addresses, timestamps, the authentication method used, and an image of the signature. DocuSign generated this PDF. DocuSign maintains the underlying database. DocuSign's internal controls prevent anyone at DocuSign from modifying the underlying records.

You are trusting DocuSign.

This isn't about DocuSign's honesty. It's about a structural fact: any database on someone else's server is mutable by the people who run it. "Mathematically verifiable as untampered" requires proof that exists outside the vendor's infrastructure. Right now, it doesn't. And courts are starting to ask why.

  You claim document was signed
         ↓
  DocuSign Certificate of Completion
         ↓
  DocuSign's database records (mutable)
         ↓
  DocuSign's internal access controls
         ↓
  "Trust us, we didn't modify anything"
         ↓
  Court: "Can you prove that independently?"
         ↓
  ⚠ Silence

Compare this to what a hash-chained audit trail with RFC 3161 timestamps actually gives you. Every event in the log includes a cryptographic hash of the previous event. Each entry is time-stamped by an independent Timestamp Authority — a third party that signs the hash with their private key. The document hash is anchored to external ledgers: GitHub commits, Bitcoin's blockchain via OpenTimestamps, transparency logs like Sigstore Rekor that are append-only by design.

You can hand someone a sealed PDF and a manifest of SHA-256 hashes. They can verify every step without calling anyone. The math either checks out or it doesn't. No trust required.

Courts Are Starting to Ask Hard Questions

California courts have been sending warning shots for years. The evidentiary pressure is not new — it is accelerating.

Fabian v. Renovate America (2019) is the earlier shot: an appellate court found that a DocuSign 15-digit verification code plus an "ID Verification Complete" marker was insufficient proof that Rosa Fabian had actually signed the document. The company had the DocuSign records. They still lost on authentication. Then in Iyere v. Wise Auto Group (2023), the employer actually won — but only because the employees had signed by hand. In dicta, the court warned that authenticating an e-signature "can be quite daunting," noting that an employee can deny an electronic signature from a printout in ways not available with a handwritten signature. Read that as the appellate bench telling everyone who relies on e-signatures what's coming.

The losing party in Fabian didn't lose because DocuSign failed. They lost because they couldn't answer the follow-up question: how did that signature get there? Who had access to the link? Was there any independent verification that the person who received the email was the person who signed?

This is the evidentiary gap. Standard e-signature platforms hand you a receipt. Courts are increasingly requiring a proof.

This pattern is familiar. We saw it with test coverage — a number that signals rigor without guaranteeing it. We saw it with observability dashboards that look like monitoring but tell you nothing about what actually broke. The industry commoditizes the appearance of the thing rather than the thing itself.

The Verification Problem Nobody Talks About

Try to verify a DocuSign document independently. Not "log into DocuSign and look at the envelope." Independently verify that the document hasn't been modified since signing, without using DocuSign's servers.

You can't. The certificate chain for PAdES signatures on DocuSign documents traces back to DocuSign's certificate authority. DocuSign's PKI infrastructure is the root of trust. If DocuSign's CA were compromised, or if DocuSign needed to reissue certificates, or — in the nightmare scenario — if an insider wanted to modify records, the verification chain ends at their infrastructure.

RFC 3161 trusted timestamps break this dependency. The timestamp is generated by a third-party Timestamp Authority (TSA) that signs the document hash with their private key. The TSA has no ongoing relationship with your document after that. They can't modify the timestamp retroactively. Their private key either validates the timestamp or it doesn't. Sectigo, SSL.com, DigiCert: these are independent TSAs with their own audit histories.

Add hash anchoring to a public ledger — Bitcoin via OpenTimestamps, or a transparency log like Sigstore Rekor — and you've got proof that existed at a specific point in time, anchored to an append-only ledger that no single party controls. Short of reorganizing Bitcoin's blockchain, which would require more compute than exists on Earth, that timestamp is immutable.

Who Actually Needs This

Most agreements don't need cryptographic proof. Your gym membership, your SaaS terms of service, your apartment lease renewal — these aren't going to be disputed in court. DocuSign's audit trail is entirely sufficient for the overwhelming majority of what gets signed.

The minority of agreements that end up disputed is where all the money lives.

The arbitration clause that Renovate America lost in Fabian? That wasn't a signature problem. It was an evidence problem. Voided arbitration clause means starting over in court: potentially six figures in legal fees plus the original dispute. Because they couldn't prove who clicked a link.

A standard e-signature is a payday loan: perfect for the signings that never get challenged. In the small fraction that end up in dispute, the evidence package you collected at signing is the only evidence you'll ever have. The signing moment is a one-way door. You can't go back.

The market has never priced that asymmetry correctly. DocuSign charges by seat. The pricing model doesn't correlate to document risk — a $25/month plan processes a media release the same way it processes a $5 million software license.

That's the real market gap. Not price. Not UX. The gap is that nobody in the mainstream has asked: what does it actually cost to make a document irrefutably provable, and what are you doing differently for high-stakes documents versus low-stakes ones?

The answer right now is: nothing. Every DocuSign envelope gets the same evidence package regardless of what's in it. Whether you're signing a media release or a $2 million software license, you get a PDF with some IP addresses in it.

That should bother anyone who has ever tried to enforce a contract.

What Actually Works Instead

The e-signature market is not going to be disrupted on price or convenience. DocuSign and Adobe Sign are deeply embedded in enterprise workflows. Switching costs are real. Nobody is switching their 200-person company off DocuSign because a competitor is $5 cheaper per user.

But the trust architecture is a different conversation. Trust hierarchies matter in ways that become viscerally obvious when something goes wrong — exactly like how productivity metrics that look good on paper don't survive contact with a production incident.

There is a category of user — solo counsel, boutique M&A advisors, healthcare compliance officers, startup founders signing anything with IP implications — who needs something the market doesn't currently offer as a first-class product: a platform where the evidence is the point, not the convenience.

Not cheaper than DocuSign. Not prettier than Adobe Sign. More provable than both — meaning: hash-chained audit trail, dual RFC 3161 timestamps from independent TSAs, document hash anchored to a public ledger. Evidence that exists outside the vendor's infrastructure, period.

There's also a question nobody asks until their platform gets acquired: what happens to your five-year-old contract evidence when the TSA rotates its signing certificates or the vendor shuts down? With vendor-controlled proof, that's their problem and it becomes yours. With externally anchored proof, the Bitcoin blockchain doesn't go offline when a startup gets acqui-hired.

The California courts handed whoever builds the provable version a marketing brief. "Courts have begun requiring more than a vendor audit log" is a very compelling pitch to a general counsel whose job is to make contracts stick.

Legal validity and cryptographic proof are different things. One is a statute. The other is math. Only one of them works when someone actually disputes the signature.