The Code That Bent

Last Monday, the faculty voted to hire the proctor back. One dissenting vote. The loop took 133 years to close.

The official explanation blames AI and smartphones. Cheating looks different now — a student staring at a phone could be doing anything — and peer detection has become unreliable. This is accurate. It's almost entirely beside the point.

The 2025 Senior Survey tells the real story. Nearly 30% of Princeton seniors reported cheating on an assignment or exam. More importantly: 44.6% reported knowing about violations they chose not to report. Only 0.4% reported a peer.

AI didn't break Princeton's honor code. Students did. Deliberately, quietly, and years before any language model could spell "exam."

Here's the structural pattern that every peer-enforcement system runs into eventually: reporting requires a student to accept real social costs in exchange for abstract institutional benefits. The peer who reports their classmate gets doxxed or shamed — as the faculty proposal delicately puts it. The one who cheats gets, on average, a better grade and no consequences, because 44.6% of their peers will process what they've seen and decide it's not worth it.

This isn't a technology problem. It's an incentive architecture problem. The honor code distributed enforcement to the party least able to absorb the cost of exercising it. This is why every peer-reporting system — from corporate whistleblowing to academic integrity — shows the same near-zero reporting rate, give or take a decimal point. Social animals don't police their own communities through formal channels. They never have.

What AI did was make the existing fiction unsustainable. When violations required a second device or smuggled notes, the ceremony of not-looking could still function. When a phone contains everything and looks identical whether a student is cheating or not, even the performance of honor becomes logistically impossible.

So the faculty voted. Instructors will now be "a witness to what happens" — not interfering, just present. Surveillance lite. The 1893 students would find this perfectly coherent: oversight implies distrust, and the distrust has now been formally acknowledged.

The interesting question isn't what changed this year. It's what Princeton thought it had for the preceding century. Not an honor code, exactly — more like a shared story that most parties found convenient enough not to examine. The institution got to advertise itself as producing people of unusual character. Students operated under lighter scrutiny than peers at proctored schools. Both sides had reasons to maintain the fiction.

AI arrived and started making the fiction look visibly false. Not because AI made cheating more common, but because it made the gap between the story and the statistics too wide to ignore.

Princeton's faculty didn't change the honor code. They retired a narrative.