I’ve recently dreamed up a consistent model of time travel specifically designed to allow for good science fiction.
Up until now, I’ve typically used a branching-timeline model for the sake of simplicity, but it leaves much to desire in the way of plot: each time traveler finds themselves in their own, separate timeline, and there is eventually very little point in killing Hitler because he’s still alive in the original. To remedy that, I now introduce… [drumroll….]
The Ripple Model:
The central idea is an old cliche in lousy science fiction: you change the past, and the change propagates through the timeline until it hits the present. In most cases, it leads to absurdities like Back to the Future (“Oh my gawd, I’m fading! I have only hours of vaguely-defined metatime to make sure my parents meet, thereby leading to my conception, which is apparently not already guaranteed because only I have free will in this movie!”)
However, this approach can be made rigorous. Consider the following:
You leave 2010 and appear instantly (meaning that no subjective time passes) in 1930 in Germany. You shoot Hitler and immediately leave for France, 1940. Twenty seconds later, the world changes around you: the war-torn nation is replaced by a happy pastoral scene, but you are unaffected because your own history (starting from 2010 and including the time travel) is intact. You jump to the present, and find that WWII still happened. Seventy seconds later, the encyclopedia entry in front of you disappears, but you hardly notice, because a tiny fraction of a second after that, the ripple swallows your own remaining history and you disappear. Nearby, an alternative version of you that has never heard of Hitler or WWII carries on.
The big issue with this is that it requires a “metatime” construct, a deeper layer of time which ripples can move relative to. Metatime is tough, but it can work: you just have to view time, and the people in it, as objects moving forward steadily at one second per metasecond. Picture a steady parade of versions of yourself, separated by one second. Each ages by a second and fits perfectly the role played by the version ahead of it one second ago. When a ripple occurs, moving faster than one sec/sec (I used one year/second), it moves along this parade, changing each person in turn. When it hits the point where a time traveler disappears it either erases their journey (because the new version has no motivation to travel) or jumps to the point of reappearance (if the new version still decides to travel).
Actually, my story earlier has a major flaw. The ripple erases your journey in time, and therefore stops Hitler’s assassination: the result is a seventy-metasecond loop, in which the two possibilities alternate. This is self-consistent, but annoying.
Let’s say, instead, that you leave a note — in 1930 — to your future self, telling him/you to travel back in time and kill Hitler. Assuming you find it and obey, the new version will still travel, and the ripple will jump back to 1930 along your own timeline and overwrite the portion of your life where you kill Hitler: instead of killing him for obvious reasons, you are assassinating a man you’ve never heard of because you said to. This leads to a final timeline with no WWII, and containing an ontological paradox worthy of Heinlein (if you haven’t: read “By His Bootstraps” and “–All You Zombies–“. You won’t regret it.)
This approach solves the awkwardness of creating ontological paradoxes in the first place, and allows other exciting possibilities: fluctuating realities like our 70-sec loop, as well as a legitimate (but hilariously weird) version of Back to the Future (consider: “I have to make sure that my parents meet! The ripple is about to catch up to me, but I can at least insure my existence 50% of the time by creating a second ripple that will restore my timeline every time the first one destroys it.”)