+++
author = "Dixi Yao"
title = "The Future Is the Past: A Random Thought Experiment"
date = "2026-11-13"
+++

I finally have time to write something now that I have submitted the first version of my ICLR 2027 rebuttal. It has been pretty exhausting, and I haven't received any responses yet. During the rebuttal, though, I have had a feeling that today's conferences and AI community are very random. Two years ago, in my blog post [Prepare Your Application for CS PhD.](#post:csphdapp), I already said that a good paper may not get accepted, and an accepted paper is not necessarily a good paper. Today, this feels even more true. Whether a paper gets accepted seems to have become a lottery.

This motivated me to think about two things:

1. If we already knew the results, would things be different? Could we really go to a parallel universe or go back to the past and change something?
2. Or is randomness purely a human feeling? Perhaps there is no randomness at all, and everything is definite within some wave function or probability function $\phi$.

So I came up with the following sketch: the sufficiently distant future is the past, and maybe we can find some way to get back to it.

## Preliminaries

### Time Dilation

According to the theory of relativity, time dilation occurs because the speed of light in a vacuum is fixed. Suppose I am on a spaceship traveling at velocity $v$ relative to Earth. An observer on Earth measures an elapsed time of $\Delta t$, while I experience an elapsed time of $\Delta \tau$. Then we have

$$
\Delta t = \frac{\Delta \tau}{\sqrt{1 - \frac{v^2}{c^2}}}.
$$

For example, if we could design a super powerful spaceship and travel far enough into the universe, we might need only 0.5 seconds to get there and 0.5 seconds to get back. We would spend only 1 second traveling, while over a trillion years would pass on Earth. The people (or actually, some rocks) on Earth would see me fly to a place a trillion light-years away and return, while I would experience only 1 second.

### Poincaré Recurrence Theorem

If we assume that the universe is a finite-dimensional, closed quantum system undergoing unitary evolution, then for an arbitrarily small $\epsilon > 0$, there will always be a sufficiently distant time $T$ such that

$$
\left\lVert |\Psi(T)\rangle - e^{i\phi}|\Psi(0)\rangle \right\rVert < \epsilon.
$$

This means that if the current time is $t$, then 10 minutes ago was $t - 10\,\mathrm{min}$, and we can find a sufficiently distant future time $T$ when the universe's quantum state is approximately the same as it was 10 minutes ago. However, the assumptions about the universe needed for Poincaré recurrence are pretty strict, especially considering that our universe is continuously expanding. They may not hold, but let's assume they do.

## Back to the Future

There are three proposed ways of getting back to the past:

1. Wormholes, though we have not found any evidence for them.
2. Closed timelike curves in solutions such as the Gödel universe, where time forms a loop rather than a line.
3. Faster-than-light travel.

Suppose the chronology protection conjecture is true, meaning that we can never go back to the past. Our universe would be like a computer system with a hardware lock that prevents certain access permissions from ever being granted. But what if we could get back to the past by going to the future?

We could use a spaceship to travel so fast that we experience only 1 second, while a sufficiently long time $T - 10\,\mathrm{min}$ passes on Earth. That 1 second still takes us into the future. We are not really going back; we are using time dilation to reach a sufficiently distant future with a quantum state similar to that of the past.

## Dilemma?

If the whole universe can repeat according to the Poincaré recurrence theorem, the arbitrarily small $\epsilon$ also applies to our minds and memories. So our memory would also have to return to its state at $t - 10\,\mathrm{min}$, without any memory of the trip. Otherwise, the universe would not be the same.

However, as we said, $\epsilon$ is arbitrarily small. Maybe we could allow the difference between these two memories to fall within $\epsilon$, so that we could get back to the future and remember it. However, this would also seem to require the spaceship's quantum state to be independent of the closed universe. Otherwise, at every point during $T$, the spaceship's quantum state would be predictable, meaning that the universe's evolution would have to be conditional on it. It may be impossible to find such a trip.

## What If?

What if consciousness does not actually rely on the physical state of our bodies? Or what if it is carried by some particles, fields, or something that is not even a particle, which no current physics model can describe? Perhaps it is carried and transferred by something we have not discovered yet.

In that case, imagine a quantum device that can carry our consciousness. It does not carry our mind, brain, or body, only our consciousness, which might not even have mass. The device carries that consciousness at a very high speed, and the consciousness then enters the brain at time $t + T - 10\,\mathrm{min}$. That future brain receives consciousness from the very distant past. In that way, we are back to the future. The future is the past, and the past is the future.

What if $\tau$ is not 1 second, but just an instant, like a blink? Sometimes, we suddenly feel that we have encountered or experienced something before, or that it feels very similar to something we have experienced. Maybe we have just received a consciousness, a thought, from the distant past. So what is the future, and what is the past?

In summary, we can represent the hypothetical trajectory as

$$
(t_0,B_0,C_0)
\xrightarrow[\Delta\tau\approx1\,\mathrm{s}]{\Delta t\approx T}
(t_0+T,B_0',C_0)
$$

where $B_0'$ is a reconstructed brain and $C_0$ is the hypothetical consciousness carrier.

## Limitations

1. Such a transfer of consciousness may require tremendous energy, more than the current universe can supply. The energy available at any time $t$ may not be enough to support travel to $T$.
2. We assume that the quantum state of consciousness and its transfer is independent of the closed universe, but we do not know whether this assumption holds. It cannot be independent of the universe.
3. The discussion of consciousness in the "What If?" section is not scientifically valid.

## Collaboration

If you are interested in this topic and want to collaborate or conduct research projects based on the current materials, please let me know. You can reach out to me via email.