More to the point, if you 'rerolled the dice' the volcano would erupt every time, as the eruption is pretty much deterministic.
Napoleon has nothing to do with some peasant in China - but rerolling the dice will give different sperm hitting different eggs. (Fertilization being pretty random and chaotic).
Ah, you're the one who says that, then.
Or is this sperm/egg example just a commonplace here?
Obviously this is true in any one time-line. Once you introduce even a tiny change, chaotic effects cascade.
However--by the very nature of speculating in Alternate History, we are accepting, for the purpose of discussion, alternate timelines. Once we do that, there is no reason to exclude the uncountable number of alternate timelines arising in every instant, one for every possible alternate quantum outcome. Obviously then there isn't just one timeline emerging from any POD you care to name--there are within some tiny fraction of a second of the POD, uncountable trillions multiplying to quadrillions and so on. They form a widening sheaf, a set within the larger set of all possible timelines.
And so, within that set, chaotic events than can go any number of ways
can nevertheless replicate events that also did happen in any other timeline you care to name--say, ours for instance.
Thus, in a world where for example the island of Greenland did not exist--that's a POD going back many tens or hundreds of millions of years (depending on whether you want to say the rock that forms Greenland today never formed at all, or whether it just got shuffled elsewhere during continental drift)--we could nevertheless outline human history happening very much as it did OTL, up until human history as we know it OTL started to specifically interact with Greenland itself.
Of course global weather patterns depend on both ocean currents and interactions with landforms. Greenland holds a fair amount of ice; if we don't shift that ice onto other lands--which broadens the divergence of course--we have to say that water just never existed, or we have different shorelines all over the world. (Actually if the rocks that form Greenland OTL never formed, there is probably more water on this particular set of alternate Earths with our sea level and other landforms the same--because the landmass of Greenland displaces a lot of water, I am guessing more volume than the ice on top of it). Assuming sea levels as OTL and that we achieve them by saying there has been a different amount of water all along, we come to what I think should be the criterion for whether something is "ASB" or not--how physically plausible is it to get similar results some distance away?
If there are possible configurations of ocean currents and prevailing winds that would leave the Bering Strait and northern Eurasia and western North America much as OTL without this particular big island there, then fine. There would be a fair number of alternate Earths which would not have Greenland, and a good fraction of them would have very similar climates some distance away from Greenland, and within that sheaf of worlds, there would be a fraction--tiny compared to the whole subset, but still comprising a huge number of possible alternate Earths--where our species evolved as it did, colonized the planet (except for Greenland! and its immediate neighborhood would be inevitably different) and even the details of who begets whom, which genes are in which sperms which impact which eggs, can be as close to OTL (some distance from Greenland) as we like. The Inuit are different, but we can just limit that difference to absence of the particular ones who OTL did settle Greenland.
Or perhaps someone can show that, without some large island of much the same area in much the same places throughout geological history, the large-scale climate would
necessarily be different, either now or at some time in the past, and if that is true then there would be no physical way to get a world similar to ours even far from Greenland. In that case, I'd call it ASB proper.
But if the overall evolution of global climate did not require Greenland strictly, then never mind that back in the Mesozoic, whole linages of animals and plants did not find the precursor landmass of Greenland where it was OTL. The point is, equivalent ones could have arisen elsewhere, probably nearby, and their descendants evolved to converge on the range of species that exist today OTL. Not, I stress, because there is some destiny that demands our current ensemble, but because we happen to choose to look at the worlds that did produce similar ones. Meanwhile there are lots of other timelines with completely different species by our timeframe--just as there are lots of other Earths with exactly the same geography as ours, but completely different species--because our timeline is not particularly privileged, except in the sense that we choose to privilege it from our parochial point of view.
Because after all the cascading divergences from any arbitrary difference you care to name are true of OTL too--our timeline is one of a practically infinite number of possible outcomes of events pretty much indistinguishable from OTL at any point in the past; if you go back to the Tambora eruption and do nothing but take notes, hidden by some Star Trek duckblind and scrupulously avoiding all contact with everyone and scrubbing out every interaction with the environment that might have the slightest consequence, but stay with any one particular timeline, different sperm hit different eggs, it rains on different days--everything is different. However, the world you are in is no more--and no less-probable than our own.
When, to achieve a particular state of affairs, you need to imagine a whole series of events that are all not reasonably likely as individual events, then you are venturing into strictly ASB territory. What that means in terms of the huge and proliferating sheaves of possible timelines is that you are selectively looking for the relatively rare timelines that contain a sequence of highly improbable events; it's a small target to hit. It is also easy to imagine scenarios which are simply physically impossible.
And I started to go with a fun example, but this is long enough already. I just want to stress, that it seems to me there is a Church of the One True Butterfly around here that only considers half of the implications of Lorenzian chaos, and letting the other shoe drop gives us what I call "anti-butterflies." We can consider timelines that derive from very early PODs and yet still have features recognizably close to our own, if they could plausibly arise in the changed situation.
---
As for the main topic of this thread--I'd use the same criteria I'd use in judging whether a particular ATL was "ASB" or not to judge how likely a given POD among humans would be to affect say the timing of the Tambora eruption or the like. However--if I want a world where Napoleon was never born
and Tambora erupts in 1821, I wouldn't be
restricted to cause-and-effect results of Napoleon's non-birth, and in this case, with only decades for absurdly tiny variations to affect deep molten rock I'd opt for another divergence--that meanwhile, while Napoleon was not getting born, the mantle rock flowed a little differently. How far back that POD would have to be would depend on knowing a lot more than I do about geology. But however far back it has to go, we can still have France and Corsica as OTL in say 1750, even though the rock is flowing differently underground.
So for people who are considering highly implausible physical mechanisms that are unlikely to work to get meteors and comets to hit the Earth at different times, it all depends on whether it is important to you that there be a cause-and-effect relationship between some change among humans and some large physical event, or whether you were just trying to use butterfly chaos to your advantage to enable something to happen. If the latter, I invite you to consider anti-butterflies instead--Earth evolving to much as OTL up to a certain point, in a solar system that is already different. You can have this if you want it.