The reason to give a hundred-run parameter sweep any weight is what the code behind it was tested against. The Bern Smoothed Particle Hydrodynamics code has been in development at the University of Bern for around two decades, represents colliding bodies as millions of particles, and lets the modellers dial gravity, density and material strength [9]. The same approach was used earlier to simulate NASA's DART spacecraft hitting Dimorphos [10], which remains the only full-scale impact on a small body that anyone actually performed and measured. That lineage is the argument for trusting the Deimos numbers, and it is worth naming, because nothing about Deimos itself is available for a second look.
The mechanism the team is proposing is narrow. The impact has to be energetic enough to excavate the depression near the south pole and still leave the moon intact [13], and the debris has to end up everywhere rather than piled around the crater. The obliquity is what buys both: according to the University of Bern release, a roughly 320-metre asteroid arriving at an angle threw material around the entire body without breaking it apart [2]. Neither the depression nor the widespread debris layer had a settled explanation before this [16], and the team varied impactor size, speed, angle and its assumptions about Deimos's interior to find the combination that reproduced them [8].
Which is where the Hera data does its real work. The flyby observations point to a moon that is more porous and more fragile than assumed [3], and a weak, porous target is precisely the interior assumption that makes a survivable global-resurfacing impact plausible. Sabina Raducan led the study, which appears in Nature Astronomy and is the first publication to use anything from Hera's pass at Deimos [4], with collaborators at the Observatoire de la Cote d'Azur, the University of Arizona and the University of Tokyo [5].
The authors say the findings could help guide future missions, and they name JAXA's Martian Moons eXploration mission [6]. Porosity is the operating part of that. A body described as fragile and porous does not respond to contact the way competent rock does, and anchoring and sample acquisition both depend on how much load the surface carries before it collapses. If the regolith really is ejecta from one event that never escaped, then loose material is the default condition across the moon rather than a patch to steer around [13].
The compute bill is worth stating plainly. About a hundred simulations at roughly a week each [7] is on the order of a hundred weeks of runtime, close to two years if the runs were laid end to end [15]. Raducan says the code runs on Bern's cluster and is one of few capable of this class of simulation [11]. That is a short list of places able to check the result, on a moon that was not even the spacecraft's destination [12].
