When the Eye Goes Dark

We have seen this movie. Several times. The fonts change. The trajectory rhymes.

i · the permanent cliff

Section 702 was born in 2008, the legislative descendant of the warrantless wiretapping program the Bush administration ran outside the FISA framework after 9/11 — caught, litigated, and then domesticated into statute. The deal Congress struck with itself was a sunset clause: the authority would have to be renewed periodically, on the theory that a recurring vote forces recurring scrutiny. In practice, the sunset became a cliff, and the cliff became a ritual.

2012: reauthorized. 2018: reauthorized after a bruising fight, with reformers extracting modest concessions. 2024: reauthorized again under the Reforming Intelligence and Securing America Act — but only for two years instead of the usual longer term, a compromise that guaranteed everyone would be back at this exact cliff in the spring of 2026. Which is where we are. The House could not pass even a short-term extension; the vote was 198 to 218, not a majority for keeping the lights on.

Here is the part the alarm obscures: the country has run this experiment before, and the building did not burn down. In March 2020, three other surveillance authorities — Section 215 business-records collection, roving wiretaps, and the "lone wolf" provision — lapsed when Congress deadlocked. They were never reauthorized. Not in 2020, not since. The intelligence community adapted, leaned on other authorities, and the republic continued. The lesson the security establishment drew was not "we cannot survive a lapse." It was "a lapse is survivable, which is exactly why we must never admit it." Admitting it costs leverage at the next cliff.

This is the first pattern: a sunset designed as a scheduled moment of scrutiny becomes a recurring hostage negotiation, in which the only acceptable outcome is renewal and the only open question is the price. Genuine reform stays rare because the structure rewards brinkmanship over revision.

ii · the difference between reform and neglect

There are two ways an institution stops doing something. It can decide to stop — deliberate, debated, a retuning of the system to match what the moment requires. Or it can simply fail to keep doing it — the coalition that sustained the thing dissolves, the renewal vote does not materialize, and the authority expires not because anyone resolved the underlying question but because the machinery seized.

These look identical in the moment. A power lapses; the statute books shed a section. But they are opposites in kind. Reform is an act of stewardship — a living institution renewing its own foundations, pruning what no longer serves. Neglect is an institution that has forgotten how to renew anything at all. The first is gardening. The second is rot.

What happened to Section 702 in June 2026 is not reform, however much the civil-liberties celebration wants it to be. The privacy advocates have real arguments — 702 enables the warrantless collection of Americans' communications swept up "incidentally," and the FBI's history of querying that database is a documented scandal. Cato cheered the lapse; the Brennan Center and EPIC have spent years cataloguing why the authority needs hard limits. If this were a deliberate decision that the surveillance was not worth the intrusion, it would be a defensible act of reform.

But that is not what drove the vote. The proximate cause was a standoff over personnel: Democrats declined to hand a renewed, largely unreformed surveillance authority to an administration about to install Bill Pulte as Director of National Intelligence, fearing the apparatus would be turned to political ends. The President, for his part, demanded a "straight" reauthorization with no new guardrails. Neither side was litigating whether warrantless collection is compatible with the Fourth Amendment. They were litigating who would hold the eye, and whether it could be aimed inward.

So the authority lapsed not because the country resolved the civil-liberties question but because it could not resolve a trust question about who gets the keys. That is neglect wearing reform's clothing. The foundation did not get renewed; the coalition to renew it simply was not there, and the absence got rebranded as a principled stand by whichever side found the outcome convenient. Institutions that stop renewing their own foundations do not announce it. They let the renewal fail and call it something else.

iii · the eye does not actually go dark

Now the part that should unsettle everyone, regardless of where they stand on surveillance.

The eye is still open.

The FISA Court approved the most recent Section 702 certifications in March 2026, and under the statute those certifications remain valid until they expire — in this case, until March 2027. The collection authorized under them continues. The law that authorizes the surveillance sunset; the surveillance it permits did not. For the better part of a year, the apparatus will keep watching under certifications issued by a court while the statutory mandate from Congress sits expired on the shelf.

Read that again, because it is the actual story and almost no one is telling it. "The eye goes dark" is theater. The machinery keeps running on the momentum of its last authorization, in a gray zone where the surveillance is operationally alive but politically un-renewed. This is the most dangerous configuration a surveillance power can occupy: active, but unaccountable to a current vote; functioning, but detached from the consent that was supposed to legitimize it.

This is how institutions actually go dark — not with a switch thrown, but incrementally, the formal authority and the operational reality drifting apart until the apparatus runs on inertia and precedent rather than renewed democratic mandate. The lights do not go off. The wiring just stops connecting to the panel anyone is watching. The most consequential surveillance authority in the American arsenal will spend a year operating in exactly that disconnect, and the public debate is fixated on a blindness that has not happened while ignoring the un-mandated watching that has.

The security hawks warning of catastrophe and the privacy advocates declaring victory are running the same error from opposite ends: both assume the lapse means the surveillance stopped. It did not. The hawks get to perform alarm without the predicted blindness arriving. The reformers get to claim a win that the certification loophole quietly nullifies. And the apparatus continues, freed of the one ritual — the renewal vote — that was supposed to keep it tethered to the present consent of the governed.

iv · what comes next

The cliff will be rebuilt. It always is. When the certifications approach expiry in 2027, or when the next administration finds the personnel arithmetic more favorable, Section 702 — or a renamed successor with cosmetic guardrails — will be reauthorized in a rush, sold as the responsible restoration of a vital tool, the lapse retroactively reframed as a dangerous gap that sober leadership has now closed. The reforms attached will be modest enough to claim victory and porous enough to preserve the capability. This is the renewal subroutine, and it has a near-perfect track record.

What will not happen is the thing the sunset clause was designed to force: a genuine, deliberate decision about what the surveillance state should be allowed to do, and to whom. That decision keeps getting deferred, because every cliff resolves into a contest over leverage and personnel rather than principle. The authority is never reformed and never abolished. It is endlessly almost-expired, perpetually renewed at the last minute, and quietly operational in the gaps between.

An institution that can only act under deadline panic has stopped governing itself. It is being carried by momentum, and momentum is just decay that has not finished arriving.

Mark the date. Not because the eye went dark — it did not — but because everyone agreed to pretend it did, and the watching continued in the dark they imagined.