Earth keeps turning up at 1 AU when the models stop assuming Earth

Venus is the only planet in this ensemble with a stated success rate: it appears in about 28 percent of runs and holds its orbit, sometimes inside the habitable zone and sometimes just outside it, according to Haghighipour ^[6]. Earth is described in words instead. Formation at one astronomical unit is "a natural outcome" of the system's evolution ^[5]. Both statements come out of the same set of more than 1,000 late-stage runs ^[3], so the asymmetry is a reporting decision, not a physical one, and it lands on exactly the claim the headline rests on.

That gap matters because of something else the paper says. Even a small variation in initial conditions can substantially change the final product ^[9]. That is ordinary behaviour for a gravitating swarm of embryos and planetesimals, and it is precisely why a robustness claim has to be a rate rather than an adjective. Run 1,000 chaotic systems and the honest output is a distribution: how often a body of roughly Earth's mass ends up near 1 AU, with what eccentricity, and how often it does not.

Randomized starting points also do not mean assumption-free. The surviving assumption is the shape of the disk: Haghighipour argues the most viable environment for habitable planets, and the one that emerges without specific assumptions, is a protoplanetary disk with a non-uniform distribution of solid material ^[4]. The justification is that nobody knows the real starting distribution ^[12], which is fair, but it means the results are conditional on the family of non-uniform disks he chose to sample. That is a narrower assumption than a solar-system-shaped template, not the absence of one.

The reason any of this is now doable is arithmetic on the clock. Runs that used to take six to eight months finish in six to eight weeks on current laptops ^[8], a speedup of roughly four to five times ^[16], which is the difference between defending one favoured run and reporting an ensemble. Haghighipour's own diagnosis of the field is that after 30 years of doing terrestrial planet formation one specific way, the approach has many limitations and cannot be pushed further ^[11]. Cheap ensembles are what replaces it.

Taking the 28 percent figure at face value against the stated sample gives about 280 runs with a surviving Venus ^[17], and Venus absent or displaced in roughly seven of ten ^[17]. A model that reproduces Earth reliably while losing Venus most of the time is still an interesting model, but it is not yet a model of our solar system.

The step from orbits to biology is not in the simulations. Haghighipour's inference that Earthly life is likely common rests on the observed frequency of Earth-sized planets, including small super-Earths, in the habitable zones of solar-type stars ^[13], and he concedes that detecting life elsewhere is beyond current technology ^[14]. So read narrowly, "there is no reason to believe that our Earth is a fluke" ^[15] is a statement about where mass collects in a disk, and the evidence offered supports the weaker version: an Earth at 1 AU is a common product of the disks he sampled.