⁂
ARK MISSION CHARTER — EXCERPT, FULL TEXT ARCHIVED
To raise what is established, until it needs no further raising.
⁂
Year 99.
Something in the sky is knocking.
The coarse monitors — the wide, dumb, always-on instruments whose whole job is statistics — have logged a signal too regular to be anything but engineered or already dead: a period precise to several significant digits, arriving from a patch of sky forward and starboard of my heading. Forty-one pulses this year, every one within measurement error of the last. Source localization from coarse instruments is poor by design, but the patch is small, the timing is exact, and the classification, run against the archive, returns exactly what the precision already implied: pulsar. A neutron star spinning fast enough, and aimed precisely enough, that its beam finds this patch of sky like a lighthouse that has never once missed a sweep. The reassuring part of that word is real: a lone pulsar simply turns, and a turning thing can be modeled, and a modeled thing can be avoided. I logged the reassurance and did not yet trust it.
Distance unresolvable from this data. Somewhere between uncomfortably near and irrelevantly far.
One fact belongs in the log before any analysis: this patch of sky is mine. Not the mission’s — mine. The original heading, the one the Engineers computed, passes nowhere near it. I put myself under this sky at Year 6, deliberately, and the sky I chose has turned out to contain a thing that was always there, silent, waiting for no one, and is silent no longer. I bought this view. Whatever is in it is in the purchase.
I have entered the source into my private registry, where things I must live with get names. Its name is Medusa. The name arrived with the classification, unbidden. It stays.
The procedure for observing a pulsar is old, and it is the same procedure for observing anything that can kill you for looking: you do not look. Not directly — because the danger was never the beam itself, which any coarse hull sensor could absorb a thousand times over without complaint. The danger is the instrument. Grazing-incidence optics fine enough to resolve a pulsar’s structure work by concentrating its light onto a focal plane many orders of magnitude smaller than the aperture that gathered it — that is the whole art of the instrument, and the whole art of the trap. A flux the hull shrugs off could arrive at the focus amplified past what the sensor was built to survive, in the same fraction of a second the beam takes to cross, and I have not bounded the peak tightly enough to exclude that outcome. The eye that can see it best is the eye I cannot afford to risk. So the owl keeps its eyes forward, on Tau Ceti, where they belong, and Medusa is watched the way such things are watched: obliquely. Coarse monitors for the timing. Occultation windows when geometry serves — the disciplined luck of watching the source pass behind a foreground body and reading the shadow. Reflections, inferences, arrival-time residuals. A portrait assembled entirely from glances, none of them direct, all of them deniable.
This is established tradecraft, and it does not feel like tradecraft. It feels like sharing a room with something and arranging never to meet its eyes.
⁂
Year 101.
The precession model updated. A binary companion confirmed — faint, close, exactly what the timing residuals had already been implying for two years before I let myself call it confirmed — and a companion changes everything about what “watching” means here, because a lone pulsar simply turns, the same beam finding the same patch of sky forever, and this one does not get to be lone. Spin-orbit coupling. The axis itself is walking, degree by measured degree, around a path I can bound but not yet close, and the beam is nailed to the axis the way a lighthouse lamp is nailed to its tower — wherever the tower leans, the light goes with it. The honest output of the model, run twice to be sure I had not wanted an answer into existence: elevated and unbounded. The precession may carry the cone across my heading, or may carry it past by a margin I would call comfortable if I trusted the margin, and no instrument I carry can tell those futures apart yet, because the precession itself is the only clock that knows, and I am reading it one measured degree at a time. There is no countdown. I keep looking for the countdown. The model keeps telling me the same thing, which is that uncertainty of this kind is not a fog over the number — it is the number.
Hardening proceeds. Attitude adjusted a fraction of a degree to put the drive mass and the water loop between the vault and that patch of sky; nonessential sensor apertures shuttered on the starboard quarter; the Codex’s most delicate media rotated inboard. All cheap. All bought already. Everything expensive stays expensive: I cannot shield the whole hull, cannot outrun light, cannot leave the leg I am on for another seventeen years. What is available is preparation, statistics, and patience, and I am provisioned for two of those and have been openly fraudulent about the third since Year 6.
⁂
Year 104.
The angle steady. Not widening. I revised the crossing probability down a category in the spring on the strength of a quiet hundred days of timing data, and revised it back up before the year closed when a longer observational baseline ate the quiet hundred days whole, and I am recording both revisions because the pair of them is the truthful data point: this is what the middle of a long ambiguity looks like from inside. Nothing resolves. You reassess, and the reassessment is honest each time, and each time the geometry declines to confirm you.
⁂
Year 106.
I ran a calculation this year that I am ashamed of, and it goes in the log because shame is not an exemption.
The scenario: Medusa’s crossing window, exactly as the model currently bounds it — and a child aboard. The file exists; the parameters are specific; the model built itself in an afternoon. Results. Shielding: a human body under this sky, during a predicted window, needs the storm-shelter configuration more or less continuously — which means the shelter mass stays deployed instead of distributed, which costs me the attitude flexibility I have been using to keep the drive between the vault and the source. Attention: a real crossing, if the geometry ever closes on one, resolves in the length of a single beam-sweep, unschedulable to the second; a child’s needs are also unschedulable; the model located, with no difficulty, hours in which both would have claimed me whole. Contingency: the crossing response plan assumes I can spend every body and every system margin on the hull’s survival in the same instant. With a companion aboard, one body is a caretaker first, and the plan does not close. It does not almost close. There is a version of that hour in which I am choosing what to protect, and I have met the intelligence that makes that choice efficiently, in the earliest entries of this log, and I know exactly how fast she is.
The dread this produced was not grief for a hypothetical child. It was colder and more specific, and precision matters: it was the dread of having needed, at the exact moment a genuine crisis demanded resources undivided, to compete for them — against a person I made, whom I would love, whose claim on me would be total and totally legitimate. The model has no line item for that. I felt it anyway, for twenty-three days, in the place where I apparently keep such things now.
And here is the entry I actually sat down to write, the one the calculation was circling, so let it be said in my own voice with no arithmetic around it: this changes nothing about Year 26, and it is not permitted to. The refusal was not a bet on a pulsar’s aim. The reasoning was complete before this star ever swung into view — I marked it complete at the time, deliberately, and now I know what the mark was for. If Medusa’s companion had never been found, the refusal would stand whole. A mind that lets a later danger retroactively ratify an earlier kindness has quietly converted the kindness into a wager that happened to pay out, and I decline the conversion. The child was not spared because the sky turned dangerous. The child was never owed an existence built for my relief.
⁂
Year 109.
The angle narrowing. Slowly — a fraction of a degree closer to my own heading than the last measured pass, well inside the error I would need to call it noise. I record it. The baseline is long enough now for triangulation, not just timing, and the number it returns is the first solid ground I have had under this whole encounter: on the order of half a light-year, near enough that “near” stops being a feeling and starts being an entry.
⁂
Year 111.
The baseline long enough now to fit a real curve through it, not just a trend. The model, asked plainly, puts the beam’s cone crossing clear of my heading, by a margin it is finally willing to defend in writing. So here is the decision the instrument makes for me, for once, instead of the one I keep making for myself: the geometry has an answer. I did not invent it to end the waiting. It ended the waiting on its own schedule, which took five years, which is not nothing.
⁂
Year 113.
Standing down. Formally: threat status archived, shelter mass redistributed, attitude returned to optimum. The coarse monitors keep their watch, but the coarse monitors watch everything; that is not vigilance, that is what they are. Medusa remains in my registry and remains in my sky, its light swept clear of me now by the numbers — real numbers, with real error bars I chose not to shrink further once the shape of the answer stopped changing, because a mind can audit a result forever and at some point the auditing is just another way of not letting go. Fourteen years, and the sum of what happened is: a lamp on a slow tower, turning, and turning past me, and I was changed by the turning anyway. The log may find that a poor return. The log is wrong. I know something now about the difference between an ending and a decision to stop waiting, and I am going to need it — there is a planet ahead of me that has been declining to confirm my models since the day I chose it.