Nothing Happened, and That Was the Point

Going into midnight on 1 January 2000, my partner, our friends and I skipped the big celebration on the National Mall in Washington, DC. Part of it was simple: it was one enormous crowd with limited ways out, and I have never liked being somewhere I could not get out of if everything went wrong. But there was a quieter reason. I was a sailor then, serving under Don't Ask, Don't Tell. The man I was with had been processed out of the service six months earlier under that same policy, for being gay. I was still living inside the exposure; he was the proof of what it cost. So we went to Badlands instead, a club that was ours, got there early, and had a completely uneventful night.

Which turned out to be the story of Y2K everywhere. Nothing happened. And the way most people remember that nothing is exactly backwards, in a way that is about to cost us.

The popular memory is that Y2K was a panic that fizzled. A media apocalypse, a lot of consultants getting rich, and then the date passed and nothing broke. The lesson most people took away is that technologists cry wolf.

That reading is backwards, and it is worth correcting now, because we are walking into the same shape of problem again.

The millennium bug was real. For decades, to save scarce and expensive memory, systems stored years as two digits and assumed the "19" in front. The flaw sat dormant inside critical infrastructure, harmless, until a specific date threatened to activate it everywhere at once. The reason almost nothing broke on 1 January 2000 is not that the threat was imaginary. It is that the world spent years and a great deal of money doing the tedious, unglamorous work of finding and fixing it first. The non-event was the achievement. We engineered a quiet midnight on purpose.

Quantum decryption has the same architecture. A dormant flaw, a future trigger, and expensive remediation that has to happen before the trigger arrives, paid for by people who would rather not touch systems that currently work fine.

The flaw is this. Most of what protects data today rests on mathematical problems that are easy to set and very hard to reverse, at least for classical computers. A sufficiently capable quantum computer changes that. And the threat does not wait for the machine to exist. It is already operating, under a model called "harvest now, decrypt later": adversaries copy encrypted data today, store it, and wait. They are aggressively targeting data with multi-decade confidentiality lifespans, such as classified intelligence, genomic databases, and long-term financial records, because the value of that information mathematically outlives the algorithms currently protecting it. The day decryption becomes feasible, years of harvested traffic becomes readable at once. The breach already happened. The decryption is just the delayed detonation.

This is where the Y2K analogy stops being reassuring. Y2K had a fixed, known deadline, and the fix was forward-looking: patch the system before midnight and the data stays safe. The quantum timeline is a moving, debated target, somewhere across the next decade, and the remediation runs in reverse. Data already harvested is already lost the day the capability arrives. You cannot retroactively re-encrypt a copy sitting on someone else's server. The clock started before the alarm was rung, and it keeps running for every sensitive thing in transit right now.

So the stakes do not arrive in one dramatic moment. They compound, quietly, every day.

The standard response is a cryptographic arms race: migrate the whole world to post-quantum algorithms before the trigger. This is necessary, and it is genuinely hard. The new algorithms are not drop-in replacements. They carry larger keys and heavier signatures, they strain constrained systems, and most organisations do not even have a full map of where cryptography lives in their own infrastructure. It will take years, and it has to be done.

But here is the option the Y2K engineers never had. They could only fix the systems they had. They could not choose to not have the flaw.

We can.

Data that was never collected has no time bomb in it. The safest ciphertext is the one that was never created. An adversary with a mature quantum computer cannot decrypt a file that does not exist, and cannot mine behavioural intelligence from a system built not to log it in the first place.

It is the same instinct that kept me out of a crowd with no easy way out on the last night of 1999. You do not have to predict whether the bad case comes. You can decline to be standing in it. I learned that the hard way, long before I built anything: when the wrong fact about you can be collected and used against you, the safest position is the one where the fact was never there to find. That was survival once. It turns out to be architecture now. For two decades the default has been to collect everything, on the assumption that data is always an asset. Harvest now, decrypt later turns a large share of that stored data from an asset into a dated liability.

Which reframes data minimisation entirely. It stops being a compliance cost or a privacy nicety and becomes a durability strategy. It is the one posture that gets stronger over time rather than weaker, because it does not depend on staying ahead in a race. For the data you decline to collect, you have simply opted out of the harvesting model.

This is also, quietly, the sharper version of the digital-sovereignty argument Europe keeps having. Estonia is already treating the post-quantum transition as urgent rather than theoretical, because a digital state has the most to lose if its encryption has an expiry date. They are actively quantum-proofing their national Population Register precisely because citizens' data requires lifelong protection against future decryption. The serious work there is not only about owning more infrastructure. It is about designing systems that need to hold less sensitive data to function at all. Sovereignty as architecture, not procurement.

The real lesson of Y2K is not that the next warning is overblown. It is that quiet, early, structural work is what turns a catastrophe into an unremarkable Tuesday.

For the quantum threat, the earliest and most durable version of that work is a decision available to us right now, before any of the hard cryptographic migration even begins. It is the same decision I made in a corridor in Washington twenty-six years ago, scaled up to the systems we now live inside:

Deciding what not to collect.