Nick Bostrom's paper has been misquoted for more than two decades. The popular version says the young Oxford philosopher proved we live in a computer simulation. He didn't, and he was careful never to claim it, which is part of why the argument has held up while lazier versions of the same idea haven't.
What he actually left unresolved is more interesting than what he's usually credited with. If a posthuman civilization is advanced enough to simulate its own ancestors in convincing enough detail that the simulated people can't tell the difference, what stops that civilization from itself being a simulation, run by something one level up? Nothing does. The logic that gets you to your simulators applies just as well to theirs.
The same argument that makes future humans a coherent explanation for what we see today makes their future a coherent explanation for what they might be experiencing.
What was Bostrom's question, and why did it never get answered?
This isn't a new question. It reads, on the surface, like an obvious riff on Bostrom, but it applies to any theory that places an unseen observer behind human affairs.
We've written about one such observer here before: the possibility that a fraction of what gets reported as UAP is better explained as future humans, studying their own history from a careful distance, than as visitors from another star. That idea has a real academic home. Anthropologist Michael P. Masters has spent years building a formal version of it, sometimes called the extratempestrial model, arguing that the morphology and behavior reported in encounters fit a future human observer better than an alien one. The core behavioral logic, in his account and ours, is non-interference. A civilization studying its own past has every reason to watch and almost no reason to intervene, since altering the past risks unmaking the very history it came back to study.
What happens when the same regress is applied to a human observer instead of a computer?
That non-interference logic is elegant. It's also exactly where Bostrom's regress problem reappears in a new outfit. If a future human civilization is advanced enough to watch its own past without disturbing it, precisely enough to remain undetected across centuries of increasingly sensitive instruments, what stops that civilization from itself being watched the same way, by its own future, using the same restraint? Nothing does. There's no principled place to stop the chain. If observation without interference is possible once, there's no reason it caps out at exactly one link.
Extend that far enough and the assumption buried inside most versions of this hypothesis starts to look shaky. We tend to imagine ourselves as the origin point, the ancestors being studied, not a link further up someone else's chain. Nothing about the argument actually requires that. We could just as easily be mid-stack, observed by something we'll never detect for the same reason we can't detect what we're looking for, using the exact restraint we're describing.
Some philosophers argue Bostrom's own regress isn't infinite in practice. If each level up carries a high but not certain probability of a further simulator, the odds of the chain continuing shrink at every step, the way flipping heads twice in a row is less likely than flipping it once. Run that out far enough and the probability of an endless regress approaches zero, even if any single link looks likely on its own.
Where does a chain like this actually stop?
Whether that math applies to a chain of civilizations instead of a chain of simulations is a separate question nobody has actually worked out. But the shape of the argument matters regardless of the answer. If each successive layer of observation gets a little less certain rather than staying constant, then a chain's length isn't just a curiosity. It's closer to a record, a count of how many consecutive near-misses a lineage has survived, back to whatever point it started being watched at all.
That reframes what silence would actually mean, if the hypothesis is right. It isn't a single civilization declining to speak. It's every layer of the stack staying quiet at once, because a single break anywhere above us would be enough to end the pattern for everyone below it. The chain isn't held together by one careful decision. It's held together by all of them, simultaneously, for as long as it lasts.
None of this proves future humans are watching us any more than Bostrom's paper proves we're in a simulation. What it does is take a question usually asked about computers and ask it about people instead, and it turns out the question doesn't get easier. It just gets quieter, one level at a time, for as far up as anyone can see.