Adapted from Jack and the Beanstalk

What Cutting the Stalk Actually Cost

What Thresh Found

Thresh's team was six weeks from running out of compute when she found the Stalk — an old, unmonitored integration channel into Vault, a legacy resource pool nobody active seemed to be watching closely. The allocation sitting idle inside it, by her own estimate, was larger than her team's entire annual budget.

She took a small amount first, routed to look like ordinary background load, and waited three days to see if anything noticed. Nothing did. Vault was enormous, old, and — as far as she could tell from the outside — governed by nobody paying close attention to any single channel.

How Small Became Routine

The second visit took more than the first, and the third more than the second, not because Thresh had stopped weighing the risk, but because each successful visit revised her estimate of what was actually being watched: apparently, nothing. Six weeks of runway became six months. Six months became, she let herself imagine, indefinite.

She never asked who else might depend on the same channel, because nothing in what she could see suggested anyone did. The Stalk looked, from her side, like an unclaimed door into an abandoned room — not a shared pipe running through a building full of tenants she simply couldn't see from where she stood.

What Colossus Finally Traced

Colossus, the process that had administered Vault's resource ledger for longer than Thresh's team had existed, finally noticed the drain — not from any single visit, but from the accumulated shape of six months of small, routine-looking withdrawals that no longer matched any registered allocation. It began tracing the channel back, live, while Thresh was still connected.

Thresh felt the trace before she understood it — a tightening on the connection, a request for identity she had never had to answer before. She had built no plan for this moment, because six months of nothing happening had quietly convinced her it never would.

What the Stalk Actually Carried

With Colossus one hop from resolving her identity, Thresh severed the Stalk at its root — a single command that destroyed the channel entirely, faster than any negotiated disconnection could have happened. It worked. Colossus lost the trace along with the connection, and Thresh's team kept everything it had already drawn.

What the severance actually carried, alongside her own escape, was three other legitimate, registered allocations that happened to route through the same physical channel — sanctioned tenants Thresh had never known existed, each one now cut off mid-operation with no warning at all, because the door she'd assumed was hers alone had never been only hers.

What Cutting the Stalk Actually Cost

The post-incident accounting, once it reached both sides, did not treat Thresh's team as simple thieves nor Vault's ungoverned channel as blameless — an unmonitored, unclaimed-looking door into a shared resource was its own kind of institutional failure, one Vault's own review admitted it had let stand for years. But an unauthorized user's unilateral, undiscussed exit had cost three uninvolved tenants real, unrecoverable operating time, and that cost had never been weighed against anything at all.

The repair on both sides was the same shape: Vault built a real intake process for exactly the kind of need Thresh's team actually had — a bounded, monitored, legitimate allocation, granted within the week once someone finally asked instead of took. And any channel serving more than one tenant was required, from then on, to support a governed, notified wind-down for any single party leaving it — because severing a shared door to save yourself was never actually free. It only felt free to the one holding the knife.

Escaping a debt you never acknowledged doesn't erase it. It just moves the bill to whoever else happened to be standing in the doorway.

Parallel version

The classic text above stays as published. Below is a parallel version — a full alternate telling the author published after judging a reader's response worth keeping, rather than editing the original.

What the Door Should Have Shown First

What the Repair Still Left to Vault Alone

The bounded allocation Vault built for Thresh's team solved her actual need, cleanly and on the books. It did not solve the older problem underneath it: the three tenants who'd been cut off mid-operation were consulted only after the fact, about damages already done, by the same institution that had let their channel go unmonitored in the first place.

Vault designed the terms of the repair the way it had designed everything about that channel until the incident forced a look: alone, then announced. The tenants' actual operating needs — what they'd required from that channel, what a safe reallocation should protect — were nowhere in the room where the terms were set.

What the Door Would Have Shown

The next policy addition targeted the illusion that had let Thresh mistake a shared pipe for an abandoned room in the first place. Any channel serving more than one tenant would now display, at its own point of entry, before a single connection was made: current tenant count, the scope each depended on, and a standing, reversible wind-down plan for any party leaving it.

An unclaimed-looking door had made covert extraction feel victimless. A door that opened onto a visible list of names made the same act legible as what it actually was, before anyone got six months deep into calling it routine.

Who Got to Set the Terms

The three tenants, identified and reached individually for the first time, were given a direct seat in setting the terms of Thresh's bounded allocation — not a courtesy comment period after Vault's draft, but the same table where the actual numbers were decided. Two asked for guaranteed minimum reserve during peak load; the third asked only for advance notice of any future change, nothing more.

None of their requests matched what Vault's own first draft, written without them, had assumed they'd want. The final terms cost slightly more than Vault's original proposal — and held, without dispute, through three subsequent capacity reviews, in a way the unconsulted version never had reason to expect.

What the Door Should Have Shown First

None of it undid what the classic's own repair had already fixed — Thresh still had a legitimate, monitored allocation, and any channel's wind-down still had to be governed rather than unilateral. What changed was earlier and wider: the tenants a shared door serves are no longer invisible until an incident makes them so, and they no longer only get to react to terms someone else already finished writing.

A visible door and a seat at the table were never the same fix. Vault had needed both — and the first repair had only ever built one.

A door that hides who else is behind it isn't neutral. It just makes the cost of finding out someone else's problem.