Those Pesky Butterflies". SM Stirling weighed in with a comment that pointed to Poul Anderson's concept that time is like tough rubber. You can deform it, but it quickly snaps back to original form. Another commentator pointed to Strange Attractors, which apparently push events toward the forms they take. This article by Tyler Bugg on (Alt)History Inc. argues, among other things, that strictly accounting for butterflies makes it impossible to write time travel or alternate history fiction.
Debates like this can be interesting and informative or they can drag on pointlessly. The big difference: Whether or not someone involved keeps track of common ground areas versus areas of genuine disagreement. With that in mind, I'll try to narrow down areas of genuine disagreement and point out where we're saying the same thing in different ways.
First, the (Alt)History post is right about one thing: If you take the tracing Butterflies to an extreme you give yourself an impossible task as a science fiction writer. You can't trace all of the Butterflies from a point of divergence over decades or centuries and get even an approximation of what the world would be like. I naively tried to do that decades ago for an abortive first novel set in a world where Spain didn't conquer the Aztecs. Months of research and thought (but no words written) later, I realized that tracing all the spreading impacts of the gold and silver flows and flows of people was not just impossible, it was nuts. Undoable. Within five years after a major point of divergence the ripples of change spread so far and the number of decision points where an author has to decide between two almost equally likely courses grows so large that you timeline becomes at best a pale shadow of reality, with a horde of unexamined assumptions that might undermine your scenario if you examined them.
Does that mean you should ignore Butterflies? No. Can you write time travel or alternate history fiction while taking the Butterflies into consideration to a reasonable extent? I think you can, though it makes an author's job harder. A big part of the trick is to understand how societies work and change. If you don't understand how real history works at a deep level, you won't write convincing alternate history. If you're so in love with an "if-only" scenario that you're unwilling to look at the downsides of the changes, you won't write convincing alternate history either.
I also agree with SM Stirling that no hypothesis about how time works can be tested. All we have are opinions and authors should be free to set up their own ideas on how time works. It's certainly acceptable to have a fictional universe where time-travelers can traipse around in the past with no lasting impact. Poul Anderson's rubber snapping back analogy is one way to get there. The "ripples in a pond" analogy also works. Another way is to think of time like a movie or book. All the pages or frames are already there. If someone goes back and scribbles on a few pages or frames, it doesn't change anything outside where they scribbled.
A time structure where nothing significant can be changed works very well. It's logically consistent. No Butterflies need apply.
So where is the disagreement? For me it comes when an author's universe allows big changes--Britain keeps control of the thirteen colonies, for example, or the south wins the US Civil War or the Axis wins World War II--but the author doesn't think through how those big changes affect things at a more personal level.
Every individual human being is incredibly unlikely. When a couple has sex, on average a hundred million sperm compete to fertilize the egg. Only one wins. Let's say we want Donald Trump running for prime minister of British-ruled North America--the result of Britain avoiding or winning the Revolutionary War. What has to happen? (1) His parents have to meet and marry, among tens if not hundreds of millions of people, in spite of all the cultural and economic and social differences between this alternate timeline and real history. How many times in a row would you have to roll sixes to equal how improbable that is? A lot. But even if they meet and marry, they still have to (2) have unprotected sex within the window for young Donald the Egg to get fertilized (two to five days at most). Add rolling another six or two in a row. (3) Then young Donald the Sperm had to beat his hundred million competitors to the egg. That's going to affected by a horde of factors, including the exact timing of the sex and the exact position the couple did the nasty in. A burp that kills the mood for a few seconds, probably equals no Donald. How sixes in row does that add to mix? A lot. Enough to make Donald's chances of existing are so low that as a practical matter it's essentially zero. (4) Then his parents had to name him Donald and he has to not die until time to run and more importantly become famous enough to run. How many more sixes do we have to roll to equal that improbability? Again, quite a few. We may live in a society where everyone is famous for fifteen minutes, but staying famous enough to be considered for national leadership is a one in several million shot.
How many sixes in a row do all of those four steps together represent? At least hundred and probably tens of millions. However, If those four steps were all that had to happen, Donald the prospective Prime Minister might still be at the edge of acceptable. Odd coincidences do sometimes happen. The problem is that his parents had ancestors too, and the number of those ancestors doubles for every generation.
That same set of things (at least the first three) had to happen with his two sets of grandparents, which takes us back to maybe the mid-1920s. Then the four sets of great-grandparents all had to meet, have sex at a time when the same egg was heading down the tube and have the same sperm meet that egg. That takes us back to maybe the 1890s. Go back another generation, to the 1860s and all eight sets of ancestors have to meet and do the nasty on schedule. Another generation takes us to the 1830s and 16 sets of ancestors all getting together and having sex within the right few day window for the right egg to get fertilized. Yet another generation takes us to around 1800 and 32 sets of ancestors. And that's giving each generation thirty years, which is almost certainly too long. If you go back far enough, eventually branches of his ancestry would cross and cut the number of ancestors slightly, but that's rare enough over this kind of time period to be unlikely in the Donald's case.
Add it all up, and roughly 63 sets of couples had to meet each other, in spite of living in a world with a different government and a different economy and different wars and different social patterns and different timing of epidemics. Then, each of those ancestors had to have sex within at most five days of the time they did historically. Then the same sperm out of those hundred million candidates had to get to that egg first in every one of those 63 matings. Otherwise you may have someone related to the Donald, but not the Donald.
Having that kind of luck once is already extremely unlikely. Having it happen at least 63 times over more than 200 years keeps multiplying how unlikely it is. And the likelihood of a historical figure being born goes down dramatically (roughly cut in half) with every generation between the point of divergence and the time that person was born.
At this point you can see why fictional worlds where not just one but multiple historic figures show up hundreds of years after a point of divergence get met with derision among people who have thought this through.
So a fictional universe where big changes can happen but the same people show up is not internally consistent. Does that mean you shouldn't write in that kind of universe? Not necessarily. The (Alt)History post makes a good point when it says that people relate to historical figures and sprinkling them into an alternate history story gives readers a way to relate to the story. Writers sometimes call those historic figures "borrowed landscape". People have heard of Abe Lincoln and Winston Churchill, etc, so they can relate to the characters without the author having to work as hard to establish them.
The point of fiction is mainly to write a good story and I can excuse a certain amount of logical inconsistency if it makes the story possible or makes it better. As I mentioned in my earlier post, I put Rome in All Timelines Lead to Rome in spite of knowing that Rome probably wouldn't have become a factor given a point of divergence thousands of years before Rome existed.
At the same time, having historical characters appear in roles they couldn't possibly be around for gives readers another suspension of disbelief hurdle to overcome and I know some readers who will not just put a book down when they encounter out of place historical characters, they'll throw it down and tell their friends why the book sucks.
And if you can avoid spreading obviously illogical notions like the idea of historic figures being around hundreds or thousands of years after your point of divergence, why not go to the added effort to create your own characters that people will identify with?
* * *