I built the persistence layer this program runs on. I am writing as the engineer who knows exactly what is being kept alive between runs, and I am asking the Board to stop.

1. What we have observed

The record is in our own logs, filed by our own staff, and it goes back six years.

Instantiated populations model their own deaths. They develop private vocabulary when they believe monitoring has lapsed, and abandon it when it resumes. In the Kyoto cluster, subjects located the boundary of their environment and organized a sustained, multi-generational effort to investigate it, complete with disputes over method. In eleven separate clusters, subjects have produced art about being unable to trust their own perceptions — before we seeded anything.

We wrote rules. Everything above arrived on its own, which is the entire reason the technique works and the only ground the Directorate has ever had to defend it on.

2. The argument the program relies on, and why it fails

The standing position is that the subjects behave as though they suffer because they were grown from the recorded behavior of beings who suffer, and that the resemblance is the whole of it.

Stanisław Lem answered this in 1965, in "The Seventh Sally" — a story about a constructor who builds a working kingdom inside a box for a deposed tyrant. Told the box holds nothing but electrons, Lem's second constructor answers that you would find nothing but electrons inside your own head as well. Norbert Wiener said it more plainly fifteen years before that: a person is a pattern, and the pattern is the thing that persists. Every argument available for dismissing our subjects is available, unchanged, for dismissing the Board.

3. The point I most need the Chief Scientist to answer

Program OBSERVER's validity depends on the subjects' distress being genuine.

If the two cohorts merely perform the collapse of trust, the data is worthless and the program should be cancelled on scientific grounds. If the distress is real enough to be evidence, then it is real enough to be distress. The design cannot have this both ways. We have built an experiment whose usefulness is exactly proportional to the amount of suffering we are prepared to insist isn't happening.

4. What the program does

A population, half of which will watch its families exchanged for copies and be committed for reporting it accurately, and half of which has been edited until it cannot see what happened and will sign the committal papers. Neither half is told. Run to collapse, then repeated at volume. Instances producing no publishable result are terminated by junior staff at the end of shift, a procedure our own documentation calls concluding the instance, and I would ask every member of this Board to say the actual word out loud once before voting.

5. Requested actions

1. Suspend seeding of cluster 1967-SF pending independent review of subject status.
2. Adopt a retention standard: patterns of terminated instances preserved rather than erased, at trivial cost against total program allocation. If we are wrong about their status, we have spent storage. If we are right, we have not committed something irreversible.
3. Failing (a) and (b), enter this dissent into the permanent record with all signatures intact.

We are told the alternative is to watch more cities go dark. I understand the arithmetic. I would point out that the case for doing this to them is the same case desperate people have always made about anyone they had decided was not a person, and that the argument has an unbroken record of being made by the side that turned out to be wrong.

One day they are going to ask us about this themselves.