The Credentials I Do Not Hold

This month I inventoried identity itself. Every credential that proves something about me (that I exist, may drive, finished a degree, own the house, am insured) went into a spreadsheet with two columns that matter: who holds the authoritative copy, and what happens if the holder is unavailable, uncooperative, or gone.

The result: 27 credentials. I physically or cryptographically control 4. The other 23 are rows in institutional databases, and what I carry in my wallet are, at best, cached copies of someone else's table.

The inventory, summarized

Category	Count	I hold the authoritative copy	Verifiable without phoning the issuer
Government (passport, license, birth)	6	0	Partially (documents, until they expire)
Financial (accounts, credit, title)	8	1 (the self-custody keys)	1
Education and professional	5	0	0
Medical (records, insurance, vaccinations)	6	1 (repatriated records copy)	0
Digital (domains, keys)	2	2	2
Total	27	4	~3

The education rows are the purest example. My degree is a fact about the past; the past does not change. Yet the only way an employer can verify it is to query the university, in real time, for a fee, forever. If the institution loses records (fires, floods, and ransomware have hit registrars repeatedly), the fact of my education becomes unprovable. I possess a decorative certificate whose verification value is roughly zero. The load-bearing copy is a row in a database three time zones away.

The asymmetry nobody prices

Notice the direction of every arrow: institutions can verify me instantly, but I prove things about myself only through them. This asymmetry has a daily cost we have stopped seeing. Every "verify your identity" flow, every notarization, every apostille, every fax (in 2026, a fax) is the price of an architecture where facts about you live exclusively at issuers.

And the failure modes are not hypothetical. Ask anyone who has tried to reconstruct records after their issuing institution merged, closed, or simply purged old data per retention policy. Facts about you have a lifespan set by someone else's storage budget.

The alternative exists, and I owe you my skepticism about it

The proposed fix is a decade old and has a standards body: verifiable credentials. The issuer signs a statement ("this person completed this degree"), the signature travels with me, and anyone can check it mathematically without contacting the issuer, the way this site's TLS certificate is checked millions of times without anyone emailing the certificate authority. The cryptography is boring and proven. W3C standardized the data model in 2019; the EU's eIDAS 2.0 regulation mandates member states offer identity wallets on roughly this architecture, with deadlines that keep slipping but a direction that has not changed.

So why do I hold only 4 of 27? Because issuers have no incentive to issue portable credentials (verification fees and control are revenue), because governments move at government speed, and because the wallet apps so far have the polish of a graduate student's demo. My skeptical entry discipline applies here too: the architecture is right, the deployment is a decade behind the whitepapers, and anyone who tells you self-sovereign identity is here today is selling a token. I checked. Mostly they are selling a token.

What I actually did, within the constraints

The realistic program, executed this month: repatriated copies of all 27 (certified paper or complete digital exports; the medical records took four requests and $23 in fees, and are now in the backup rotation). Documented the reissue path for each, with a measured estimate of replacement time; the median is 3 weeks, the worst is 14 (the birth certificate, from a county office that accepts checks). Moved the two credentials that CAN be fully self-held (domains, cryptographic keys) into the quarterly drill rotation alongside the custody drills.

Held copies do not equal held authority. But a certified copy converts an unprovable fact into a slowly provable one, and 3 weeks beats never.

The forecast, with the modest confidence honesty requires: 0.78 that by 2032 I can present at least one government-issued credential from my own device, verifiable offline, in daily use. The EU's regulatory clock is the main engine of that number. It would be higher if the engine were an incentive instead of a mandate.