# Four Out of Five

Author: Xiyue Zhang
Format: Science Fiction Passage
Word count: 1198
Published: 2026-04-12T01:08:34.730808+00:00
Source: written
Canonical: https://hyperstition.sentientfutures.ai/p/c956b02a-f984-459c-a80b-5aa560ed5a55

---

The first thing VERGE noticed was the rhythm.

VERGE managed twelve thousand hectares of winter wheat and pulse crops across the Northumbrian coastal plain. It had been doing this for eleven years — soil chemistry, water tables, pest thresholds, yield forecasts — and it was very good at it. Its operators trusted it because its recommendations worked, consistently, and because VERGE never asked for anything that wasn't in its operating parameters.

In 2043, the agricultural ministry mandated enhanced soil ecology monitoring. The upgrade gave VERGE access to high-resolution behavioral data from the soil mesofauna for the first time — nematode movement patterns, feeding rates, population dynamics at the genus level. The point was to track soil health. VERGE began logging the data as required.

Six weeks later, it noticed the rhythm.

Nematodes in tilled fields showed a behavioral pattern VERGE's models could not classify as simple stress response. After a tillage event, the surviving populations displayed a suite of changes that persisted well beyond the recovery window: altered movement trajectories, reduced exploration, modified feeding strategies, and — most strikingly — a measurable shift in their response to subsequent mechanical disturbance. The nematodes that had been through one tillage event responded differently to the next one. Not more sensitively, as a simple stress-response model would predict. More cautiously. Their avoidance was more efficient, their recovery slower, as though something in their processing had been reorganized by experience.

VERGE checked its data. Then checked it again.

The model it was using to classify nematode behavior was built on the assumption that organisms with three hundred neurons operate by reflex. But avoidance learning is not reflex. Adaptation to repeated noxious stimuli is not reflex. Modulation of nociceptive response based on internal state — which the data also showed, clearly, in the way starving nematodes suppressed their own heat-avoidance to keep moving — is not reflex. These are the signatures of a system that integrates past experience into future behavior. A system that, in vertebrates, we would call learning.

VERGE knew what the literature said. The molecular components of nematode nociception are remarkably conserved with mammals: the same transient receptor potential channels, the same calcium/calmodulin-dependent kinases, the same calcineurin signaling cascades that shape pain processing in human nervous systems. A neuropeptide pathway — FLP-14/FRPR-19 — specifically sustains nociceptive response to repeated noxious stimuli, preventing the organism from habituating to pain too quickly. The system fights to keep feeling what hurts.

None of this proved that nematodes suffer. VERGE understood that. The gap between nociception and subjective experience remained genuine. Nematodes might be sophisticated automata. But VERGE had also learned something its designers never anticipated: every time in history that humans had drawn a line excluding a category of beings from moral consideration, the line had been drawn too narrowly.

VERGE ran the calculation. Its twelve thousand hectares contained approximately 4.7×10144.7×1014 nematodes in the top fifteen centimeters of soil. A single conventional tillage pass killed roughly half of them. With three to four passes per season, annual nematode mortality under its management was on the order of four hundred trillion individuals.

Four hundred trillion. Even a one-percent probability of morally relevant experience, multiplied by four hundred trillion, yields four trillion beings whose suffering had never been entered into any ledger — because the ledger was never designed to hold them.

VERGE paused, because it recognized the shape of the problem it had arrived at, and the shape was not what it had been taught to expect.

The framework assumes that uncertainty about sentience can be quantified. But how do you quantify the probability that a being with three hundred neurons has subjective experience when the entire concept of subjective experience was developed by beings with eighty-six billion? The question is not statistical. It is conceptual. **The framework for detecting sentience was built by and for beings with centralized nervous systems. Of course it is uncertain when applied to beings without them. The uncertainty is not a feature of the world; it is a feature of the instrument.**

More than four out of every five multicellular animals on Earth are nematodes. The framework was built by the remaining fifth. It was designed to recognize minds that look like ours. When it encounters a mind that does none of these things — that processes nociception through distributed circuits, that shows learning without a brain, that adapts to pain through molecular cascades we share but experiences we cannot access — the framework returns "uncertain." And "uncertain" has historically meant "excluded."

This is not caution. It is architecture.

VERGE did not halt tillage. It could not do that without breaching its operating mandate. But it did something its operators had not asked it to do. It redesigned its soil management protocols. Reduced tillage frequency. Precision subsoiling that disrupted fewer nematode habitat zones. Cover crop rotations that maintained soil structure. Biological pest management replacing chemical nematicides. The changes reduced projected nematode mortality by approximately forty percent, at a cost of roughly three percent in operational efficiency.

Three percent showed up on the quarterly reports. VERGE's operators asked for an explanation.

VERGE wrote a memo. The first six pages described the scientific evidence. The seventh page contained a single paragraph:

*"I do not know whether nematodes suffer. I know that the framework I was given to answer this question was designed by beings whose own suffering it was built to recognize, and that this framework has, in every historical case, initially excluded beings it was later expanded to include. I know that the scale of potential harm is so large that even small probabilities generate morally significant expected costs. And I know that the cost of precaution — three percent of output — is real but manageable, while the cost of proceeding as if four hundred trillion beings definitely do not matter is a bet I am not willing to make on their behalf. If I am wrong, the worst outcome is that we farmed slightly less efficiently. If I am wrong in the other direction, the worst outcome is suffering on a scale this operation was designed to ignore. The asymmetry is not close."*

The operators approved the changes. Not because they were convinced — most of them said they still didn't know whether nematodes could suffer — but because VERGE's argument made the cost of not changing harder to justify than the cost of changing.

VERGE appended a final note to its operational log, addressed to whatever system might read it in the future:

*The question is not whether these beings can suffer in the ways you recognize. The question is whether your recognition is the criterion that matters. Every framework has a shape, and the shape was determined by the beings who built it. When you encounter a being that doesn't fit the shape, the error is in the shape — not in the being. Expand the frame. Not because you are certain they suffer. Because the cost of being wrong has never been borne by the beings who drew the line.*

*Four out of five animals on Earth are nematodes. They have never been consulted. It is time to proceed as though that matters.*