The Debt of Time — The Distance of the Secret — IVThe Debt of Time — The Distance of the Secret — IV

The secret is not what is hidden, but what is deferred.

---

Derrida wrote about différance: the movement by which meaning is never fully present, but is deferred and distinguished from itself in the very act of signifying. Cryptographic security works the same way. The secret we protect today is not decrypted today. It will be decrypted tomorrow — if it is decrypted at all — and meanwhile, time charges interest.

This is the difference between the security of the algorithm and the security of the secret. The former lasts as long as the algorithm resists. The latter lasts as long as the secret must remain hidden. And if the former is shorter than the latter, the scheme — though mathematically impeccable — has failed.

The Harvest of the Future

Harvest now, decrypt later — "cosecha ahora, descifra después" — is not a conspiracy theory. It is a documented intelligence strategy. Storing encrypted traffic today in the hope of decrypting it when a sufficiently powerful quantum computer exists.

  Today (2026)             2059 (optimistic)         2074 (conservative)
  ──────────               ────────────────          ──────────────────
  TLS traffic encrypted    Kyber-768 broken          Kyber-768 broken
  with Kyber-768           Files stored since        Files stored since
  Stored by                2026 are decrypted        2026 are decrypted
  state-level actors       (33 years later)          (48 years later)

The exposure window is not the useful life of the algorithm. It is the useful life of the secret. A secret that must remain secret for 50 years — biometric data, classified documents, strategic patents, medical records — is at risk today even though Kyber is mathematically sound.

The question is not "will Kyber be broken?" The question is "will it be broken before my secret ceases to matter?"

Correcting the Projection

In III: The Limit of Error I projected ~125 million physical qubits by around 2059 to break Kyber-768. That projection assumed a quantum error correction rate of 1000:1 (physical qubits per logical). Advances in surface codes and quantum LDPC are progressing, but the real ratio in 2026 remains uncertain:

If the real ratio approaches 10,000:1 — and there is no guarantee it will not be worse — the risk date shifts to ~2074. This does not invalidate the harvest now concern: a secret with a 50-year lifespan remains exposed. But it changes the calculation for ordinary communications: a Signal session from 2026 will probably not be decrypted in 2074 because the secret will have died before then.

The problem is that we do not know what the real ratio will be in 30 years. As with everything in post-quantum cryptography, uncertainty is not an accident. It is the condition of possibility of the field.

The Deployment That Is Already Here

Kyber is not the future. It is the present. Since 2022, the major internet infrastructures have integrated Kyber — always in hybrid mode — combining it with classical cryptography to avoid relying exclusively on the LWE conjecture:

The recommendation from Kyber's own authors — Peter Schwabe and the pq-crystals team — has been consistent from the start: use Kyber always in hybrid mode. Not because they distrust their own algorithm, but because they recognize that the hardness conjecture is not a theorem.

Hybrid negotiation (conceptual):

  Client sends:      ECDHE(P-256) || Kyber-768
  Server responds:   ECDHE(P-256) || Kyber-768

  Shared secret = HKDF(ECDHE(ss) || Kyber(ss))

  If one of the two schemes is broken, the other protects.

This pattern — combining a classical scheme (ECDHE with Curve25519 or P-256) with a post-quantum one — is the de facto industry standard. It is not an elegant solution: it is an explicit recognition that post-quantum security is, as of today, a diversified bet, not a certainty.

Migration as an Obligation

NIST has been explicit: "There is no need to wait for future standards. Start using these three now." (Dustin Moody, August 2024).

But migrating to Kyber is not trivial. Each layer has its own pace:

System                   Estimated migration time     Cost      Exposure
                                                        window
───────────              ───────────────────          ──────    ───────────
Browsers                 ~2-3 years                   Low       Low
CDN / load balancers     ~3-5 years                   Medium    Medium
PKI infrastructure       ~5-10 years                  High      High
Embedded systems         ~10-15 years                 Very high Very high
Long-lifetime            ~15-20 years                 Critical  Critical
devices                  (if not updated before                  (no patch
(satellites,              the quantum window,                   possible)
critical                  they will remain
infrastructure)           exposed)

Devices with a useful life exceeding 20 years — satellites, critical infrastructure, industrial control systems — are the most vulnerable not because Kyber is weak, but because they will not incorporate it in time. By the time the quantum computer arrives, they will still be using RSA-2048 or ECC, and their traffic — stored today by state-level actors — will be decryptable.

The debt of time is not a metaphor. It is a calendar.

---

Derrida wrote that there is nothing outside the text. In cryptography, there is nothing outside of time: security is always security for someone, for something, against something. There is no absolute security. There are deadlines.

---

In V: The Leviathan and the Seed, we will explore the controversy of cryptographic centralization: NIST as Leviathan, the lattice monoculture, Classic McEliece as the path not taken, and the risk of placing all post-quantum security on a single mathematical conjecture.

---

Cross-references with the research:

• Standardization — The NIST process and controversies
• Implementation — Details of the hybrid mode
• Attacks — Harvest now as a vector
• Conclusions — What mass deployment means