At a macroscopic scale, though, that won't look like much at all. Air currents, temperature, and so on don't really depend on what specific microstate the atmosphere is in. It's like shifting every object in a room at random by a tenth of a millimeter, for all intents and purposes it doesn't change anything. Usually. From time to time, some microscopic details can snowball into macroscopic effects. Well, pretty much any small change anywhere in the world is going to shuffle all of these, everywhere in the world, a few years down the road.
This means that, perhaps counter-intuitively, the more time passes, the more likely a small change is to have an effect. First it will shift everything a tenth of a millimeter, but eventually one of these shifts will snowball into a one-millimeter shift... which will propagate everywhere, setting the stage for a ten-millimeter shift, and so on. You'd be unlikely to notice, though, because the proximal cause for a big change will almost always be a medium-sized change, not a tiny one. Furthermore, it could take a very long time for the effects to really ramp up: large scale natural processes are very robust. Still, if a process has any kind of sensitivity to noise, it's just a matter of time before it shifts.
Then, overall, the makeup of the world is still different. Will this really never cause anyone to get a speck of dust in their eye that they otherwise wouldn't, flinching, preventing them from having the though "oh I have to buy milk" that they were about to have?
I don't know... maybe it's just that some changes are indeed so tiny and remote that it takes them a long long time to have an effect, but once an obvious effect is present, I guess we're back to everything changing again?
You're underestimating the effect now. You only have to play about with a simple chaotic function and see how quickly the most infinitesimal change leads to utterly different results to appreciate that a butterfly flapping its wings sufficiently in advance literally alters the course of all subsequent tornados.
If your home had a 99% chance of making it through tornado season unscathed one year but for the flap of a butterfly's wings, then it may still have a 99% chance, however it would be a completely new role of the dice.
You can dip your finger in a river and tell me how a bunch of molecules are displaced and then those molecules displace other molecules and so on in a way that wouldn't otherwise have happened. But for all intents and purposes, the river carries on in its original course as if nothing happened. It doesn't turn into a seething mass, it doesn't reverse course, it doesn't spiral off into infinity.
Sensitivity to initial conditions is relatively rare, and even most chaotic functions only get interesting around certain values.
Consider the world in terms of large numbers of probability distributions.
When you combine distributions it's the variance and not the deviation that you add together.
I.e.: at larger scales the distribution of probabilistic variables tends toward the mean.
All that's left at smaller scales is quantum white noise.
The people will largely be genetically the same, the population and its distribution will be largely the same - they'll fit into the same ecological niche, they'll be inventing the same things and polluting the same rivers.
There is potential for the butterfly effect to do something - the right change to an unstable equilibrium certainly has the potential to trigger a cascade of events.
In general though, such events are highly improbable - and even dramatically cascading events will have their impacts dampened across time and space.