The writable window: division-coupled DNA recorders carry a built-in early bias in human cortical organoids

DNA recorders write signalling and lineage history into the genome for sequencing to recover later. They werebuilt in rapidly dividing cell lines and characterised over days. Asking when a chemical perturbs human brain development instead means running a recorder for several months in a tissue that changes while it is measured: neural progenitors divide, give rise to neurons, and leave the cell cycle permanently, so the dividing population shrinks throughout. To learn what such an experiment can resolve, I modelled a branching progenitor population coupled to sequential writing into a tandem array. Because writing happens at division, a recorder falls silent long before the experiment ends. Half of all symbols are written before day 27 of 150, and a cell is 5.3-fold more likely to record a one-day exposure given on day 15 than on day 75, a ratio that runs from 1.5 to 6.6 across parameterisations of the tissue, and 2.4 under the spread that matches measured clonal structure. An experiment comparing exposures at several developmental stages therefore carries a built-in gradient shaped like the window it is meant to detect, and a genuine five-fold late window comes back flat. Of seven mechanisms that might remove the gradient, tested as 14 arms across 50 simulated populations, eight arms move it detectably and none removes it. Gliogenesis, the one mechanism that restores proliferation rather than redistributing it, blunts it by a tenth to a seventh depending on when the switch falls. A writer decoupled from division entirely gives a gradient of 1, so the whole of the effect is the coupling. Held at a fixed writing rate, the recording model also reproduces, without refitting, the array occupancy measured in one published mouse embryo, which tests the writing chemistry, but not the writable window. Records are clonally correlated, so precision comes from cultures and founder lineages, not sequencing depth.