Standard gate-fidelity calibration scans the beam across the ion, watches the transition rate, and stores the position where the number peaks. That works right up to the point where the peak moves depending on which transition you happen to be driving, which is what the ETH Zurich maps show [3].
The two measured offsets make the case. Against a 729-nanometer drive [2], a 240-nanometer offset is about a third of a wavelength and a 463-nanometer offset is close to two thirds [13]. The two transitions sit roughly 223 nanometers apart from each other [14], so this is not a common offset you absorb once into a pointing solution and forget. One stored beam position is a compromise between two different right answers.
The match to prediction is what makes this usable rather than merely interesting. Theory said roughly 232 and 464 nanometers; the measurements came in at 240 plus or minus 16 and 463 plus or minus 20 [4], which puts them 8 nanometers above and 1 nanometer below prediction, inside the quoted uncertainties [15]. A team can compute its own offset from its own focusing geometry instead of hunting for it empirically.
Teams often treat the beam center as the addressing coordinate, since it is the thing you can measure with a camera and a knife edge. The ion instead reports a different coordinate for each internal state [3], because tight focusing promotes field components that are negligible in a loose beam, and produces polarization gradients along with a longitudinal electric field [5]. Nothing is curving through space; the structure of the light shifts to wherever the coupling is strongest [12].
The proposed use and the everyday task are different things. The proposed use is treating the effect as a resource, and the source points to a 2023 theory paper proposing Magnus forces to couple trapped-ion qubits, plus later work demonstrating related entangling operations from transverse polarization gradients [8]. The everyday task is subtracting an offset nobody wrote into the calibration procedure, in a system where the same fields that shift the coupling peak also couple a qubit's internal state to its motion [7].
Read the write-up's own hedge, too. It credits the group with directly observing and spatially mapping an optical analog of the Magnus effect [1], then states that the experiment used related transverse polarization gradients rather than demonstrating the optical Magnus effect itself [9]. That distinction matters if you were planning to cite this as a demonstrated gate mechanism rather than as a measured systematic.
Fixing this only takes an added calibration scan, not a hardware redesign. Measure the coupling optimum separately for every transition in your gate set, then compare the spread between those optima to your day-to-day pointing drift. If the spread sits inside the drift, it is noise you already tolerate and price. If the spread is larger, the difference is in your error budget whether or not it has a name, and the honest fix is one stored beam position per transition instead of one per beam. Philip Leindecker of ETH Zurich, the lead researcher, describes the ion as a tiny sensor for feeling out the structure of the laser light at the few-hundred-nanometer scale [10], which is to say the instrument is already sitting in the trap.
