I really dislike this old idea, mainly because it was supported by Hoyle et al during his "steady state universe with transpermia" pseudo-creationist period. (But it can have older roots.)
Besides being a cherry-pick, as you yourself notes at the end "evolution can produce trillions of species without ever selecting for high-intelligence". Confusing growth processes with any of the many ways to measure complexity is the other sleight of hand often used. Parasitic simplification is at least as common (with other half of species parasites), meaning there is no specific trend but a Gouldian diffusion to fill up a niche space. Mass extinctions tests that nicely, with diversity recovering to random levels, it can be a more or less complex system after a ME. [I don't have that ref handy.]
Other minutiae is that biologists haven't checked with planetary scientists. Mojzics et al has shown that any realistic intensity of late bombardment can't sterilize a planet, cells procreate and disperse faster than impactors sterilize crust. (There is even a Goldilocks zone ~ 1 km down where crust busters are survivable somewhere else on a planet.)
The earliest potential trace fossils are now the Isua BIFs @ 3.8 Ga bp, since small scale isotope analysis has shown that the later metamorphosed rocks were originally biologically deposited. So with Mojzics et al in mind, we don't really know when life appeared, because of later plate tectonics - few older rocks. A simple model [of my own] using the latest protein fold phylogenies gives a (linear, btw) extrapolation of the first gene ~ 4.10 Ga bp, which is reasonably after the first oceans appeared ~ 4.25 Ga bp [from the Jack Hill zirconium/diamond evidence.]
Other concerns raised:
- Transpermia. Can't compete with local abiogenesis rates with the latest pathways such as Lane & Martin, or Russell et al. Even a martian tramway raining down flash frozen spores has too low transfer rate to make it the likeliest pathway if you take experimentally observed factors of survivability et cetera into account.
- Fermi question. Too unconstrained due to potentially unobservable pathways (silent civilizations), making for false negatives. Need observed positives for making the full Drake equation a useful estimate. (The short DE for life at large works well for potentially observable inhabited, oxygenated worlds on the other hand, as we will only look for positives.)