STAR Drive

Thrust in two factors

Every propulsion system comes apart into three pieces: a reservoir that momentum is exchanged with, a coupling that performs each exchange, and a schedule that decides when. Almost all of the attention goes to the first two. STAR Drive is a design theory for the third.

The accounting behind it fits on one line.

Thrust is momentum per event times event rate.

The first factor belongs to the reservoir and its coupling. It is governed by ordinary conservation, and this work does not touch it. The second factor is the schedule: how often the actuator fires, how reliably that cadence can be held against a drifting external reference, and how honestly the firings can be counted. STAR stands for Sampled Transport by Alias Ratcheting, and the name records the mechanism: a weak coupling ratchets the phase mismatch between the gate and its reference into deterministic integer advances, each committed advance releasing one impulse.

What it is not

Stated before anything else, because this is the part that gets oversold. STAR is not a momentum source. Within its own model it cannot change mean thrust at a fixed mean event rate; it redistributes timing and adds a basin, a ledger and a control structure. It amplifies nothing. It contains no reactionless mode, and thrust that survived a matched dummy load while failing an axis reversal would be a systematic to hunt down, not a discovery.

It is also a recognizable relative of things engineers already build: injection locking, first-order phase-locked loops, standard mode-locking theory. That kinship sets the burden of proof rather than lowering it.

Two awkward facts about lock windows

Under a weak coupling, the schedule reduces to the sine circle map, and rational firing ratios open into finite windows of stability rather than sitting at isolated points. Two properties of those windows shape the whole architecture, and both are inconvenient.

First, a window is not centred on its own label. Commanding a ratio of one third does not put you at the middle of the one-third window; at the coupling this work is anchored to, some rational values lie outside their own windows entirely, and the machine locks stably to the wrong thrust level. The operating point has to be the measured centre, taken from a scan at the actual coupling, never the number on the tin.

Second, a rational winding fixes how many events occur per cycle, not when they occur within it. The locked train is exactly uniform only at a handful of named operating points. Everywhere else it carries a deterministic timing floor of roughly three to five percent of the nominal gap, computable in advance and unavoidable by tuning. That floor is a structural handicap on the architecture's own primary metric, and it is stated up front rather than discovered later.

The ledger

Every command is a guarded threshold crossing of a single accumulated phase, committed once to a record that is never rewritten. Four counters are kept apart on purpose: what the gate wound, what the controller committed, what the actuator confirmed, and what the mission asked for. They coincide only under named conditions, and where they diverge, the divergence is the diagnostic. A schedule built this way audits itself.

How it can lose

The interesting part of the note is the part that specifies its own defeat. Where a schedule is synthetic, referenced to the controller's own clock, a plain digital counter already produces exact rational schedules with exact counts and no drift. Against that baseline the coupled gate offers nothing, and the note claims nothing for it. The candidate regime is a physical cadence that must be acquired and held: a spin phase, an orbital phase, a drifting analog reference.

Whether it wins there is posed as an open empirical question, judged on a preregistered timing metric under a hard event-count constraint, against digital schedulers, pulse-width modulation and a fully instrumented digital phase-locked loop at matched conditions. There are named model falsifiers, named experimental invalidators, and a retirement criterion: if the coupled gate fails the superiority test, the scheduler is redundant engineering and gets retired in favour of the baseline, with no falsification required.

Two things survive that outcome regardless. The accounting discipline, and a metrology standard for phase-based thrust stands, which is overdue independently of whether this architecture is any good.