past
future
now · {{ progressPct }}
{{ chapterLabel }}
+ + + +
light — 300,000 km/sthe past does not get rewritten… yetdelayed choice — confirmedfuture-boundary test — not performedthe truth is out there light — 300,000 km/sthe past does not get rewritten… yetdelayed choice — confirmedfuture-boundary test — not performedthe truth is out there
Somewhere on the internet
{{ nowLine }}▍

Can the futurechange the past?

An open journal of one investigation: what physicists have already tested with photons, where reasoning and hypotheses begin — and which single decisive experiment no one has performed yet.

/2026
research: Evgeny Vlasov
entries №1–9 · the diary keeps growing
▼ scroll — the “now” dot will ride into the future ▼
In place of an epigraph · voices around the project
“Dad — it is not us looking at the stars. It is THEM looking at us”
Zlata, age 5
woke up and decided to say it; never said anything like it again and forgot why she had said it
“Bloom where you were planted”
mom
that even in a village you can grasp how the universe works, without flying anywhere at all
“Wow, you are really tripping on this”
one of the friends
after hearing about these thoughts
“Better do something useful instead”
mom again
same ideas · the skeptic control group is working
“You are so smart, I love you”
beloved wife
“It is a rocket! It is not Saturn! Saturn is a planet-rocket!”
Fedya, age 3
a young colleague insists on his own model of the Universe
monitor // current research status
> A late measurement choice really changes which patterns can be found in old data. [proven]
> Not a single already-recorded instrument digit has changed. No messages from the future so far. [proven]
> The decisive experiment — the future-boundary test — has not been performed yet. That is the goal of this project. [in progress]▍
phase: idea map ready experiments studied: 9 diary entries: {{ journalCount }}
Legend: verified in the lab hypothesis / reasoning
Fresh update · 29.08.2026 · brainstorm reasoning · grounded in real experiments

Can you frighten a photon at the end of its path so it flinches already at birth?

The idea has matured into a precise formulation: a photon (or a system linked to it) should show at the start of the path a trace of what will happen to it at the end — before information about that end could return at light speed. Working title: testing the early response of a source to a late-chosen boundary condition.
EXPERIMENT A · the photon knows its own end
source → long waveguide → late-chosen boundary
The source starts emitting a photon at t₀. Only after emission has begun a random generator picks what awaits the photon at the end: a mirror, an absorber, a phase reflector or an open exit. Near the source they measure emission time, decay rate, phase, packet shape — and compare the early distributions before t₀+2L/c.
EXPERIMENT B · the partner knows about a future meeting
A — short path · B — long path → a meeting with photon C
An entangled pair A–B. Photon A is recorded early, the record is sealed. Later a random choice Y decides whether B meets photon C on a beamsplitter (interference, distinguishability, a Bell measurement — or nothing). We check only the old A data against Y — the results of B and C are never used.
The lock of nature
Keep coherence → the photon can join the future meeting, but tells nothing now. Measure now → you get an early record, but destroy the needed branch. In 2014, interference of “past and future states” was observed in a superconducting qubit — but the future “flinch” could be read only after obtaining the future key (post-selection).
Why 1000 photons
Not so that one of them “runs faster”, but for sensitivity: if instead of 50.000% vs 50.000% you get 50.020% vs 49.980% — over billions of runs that is detectable. Standard theory strictly demands ρ′(A) = ρ(A) for any action on the future part.
What would count as a real discovery
Not sorting by a late detector (that is known physics), and not interference at an actual meeting (that is the Hong–Ou–Mandel effect). But this: an early readable record distinguishes the future choice — without the future key.
The core of the project reasoning by the author · grounded in facts

The idea itself — in seven steps

01
Time may be an “internal” coordinate. For a game character a hundred years is a whole life; for someone watching from outside — one file, whole. What if our story is also written all at once?
02
Light lives on the boundary of causality. Along the path of a photon the proper interval is zero — light draws the edge beyond which no connections are possible. That is a fact.
03
A photon is not a bullet. It has no pre-written story of “flew exactly like this”; the story becomes definite only together with a measurement. Also a fact.
04
Maybe what matters is not the path but the bundle of events. Not “born → flew → absorbed”, but a single process: source ⟷ field ⟷ mirror ⟷ detector, coordinated across the whole history at once.
05
An outer level would not need to break physics. It only needs to choose among already-allowed outcomes: 0 or 1, left path or right. Every outcome is lawful — the trace shows only in a meaningful sequence of choices.
06
A star physically continues into us.
star→ photon→ retina→ signal→ image→ a question about the Universe
We do not “look at a star” — its light changes our matter. Through us the Universe looks at itself.
07
Everything reduces to one testable equality.
P(A | Y=0) =? P(A | Y=1)
Can a late button Y leave a trace in the early record A? Standard physics says “no, never”. That is exactly what must be tested.
▸ background scene: {{ fxHint }}
A finger-level demo · for step 05
What choice without breaking the laws looks like

Here are 32 “quantum coin flips”. There are always exactly 16 ones — not a single flip breaks physics. Switch the mode and spot the difference.
ones: {{ onesCount }} of 32 · {{ bitsHint }}
1 Entry №1 · the hero of this story verified by science

The photon — a particle of light

A photon is the smallest “portion” of light. It has no mass, so it cannot stand still: it always flies at the maximum speed in the Universe — 300,000 km/s.

While a photon flies, it behaves like a wave and can “try” several paths at once. But when a device catches it — it always clicks as a whole particle, never a half.

And the strangest part: along the path of a photon its proper time equals zero. Light flies exactly along the border separating the possible from the impossible. That makes it the best instrument for probing how time works.

> REMEMBER: the white “now” dot on the left and the cone behind the text are a map of time. Red is the past, cyan is the future. Light lives exactly on the border.
Fig. 1 — the time cone: light draws its borders
NOW
future
past
light — along the edge
2 Entry №2 · a real experiment, 2021 verified in the lab

Source → mirror → back

An atom releases a photon. The photon flies to a mirror and returns. The question: will the atom feel the mirror in advance — or only when the light physically comes back? Launch the photon and watch the atom.

ATOM
Mirror
Source
distance L
{{ flightStatus }}
What physicists saw (Ferreira et al., 2021): the atom “flinched” exactly when the reflected light returned — after a time 2L/c. Not a second earlier.
What it means: the influence of the mirror arrives at light speed, not instantly. No early “signal from the future” was found in this experiment.
3 Entry №3 · why people started talking about “changing the past” verified in the lab

Wheeler delayed choice

The photon is given two paths. And how to measure it is decided only after it has entered the setup. Switch the measurement mode and see what the screen shows.

Photon
path A
path B
Screen
Measurement mode — chosen later:
{{ modeExplain }}
What is proven (Jacques et al., 2007): the result always matches the mode chosen after the photon entered. As if the photon “knew in advance”.
The precise conclusion: the photon simply had no ready-made story of “I am a wave” or “I am a particle”. The story becomes definite only together with the measurement. Old instrument records do not change.
4 Entry №4 · the key subtlety verified in the lab

The future sorts the past. But does not rewrite it

In the “quantum eraser” experiment (Kim et al., 2000) one photon of an entangled pair is registered early, and its “partner” is measured later — and the late measurement decides whether to keep or erase the path information.

If you then sort the old data by the results of the late measurement — stripes emerge in some stacks and not in others. Looks like magic: the future “painted in” the past.

But the key point: the early record as a whole is one and the same for any late choice. Stripes and “anti-stripes” cancel out in the sum. Not a single pixel of the past is rewritten — the late measurement only suggested how to deal the old deck into meaningful stacks.

> REMEMBER: the future chooses HOW to read the past, but cannot rewrite ANYTHING in it. So far.
Fig. 2 — one deck, different deals
old data:
stack №1 → stripes:
stack №2 → anti-stripes:
Put the cyan and red stacks back together — you get exactly the original grey record. That is why you cannot send a message into the past this way.
Archive of experiments · 2000 → 2026

How far science has already come

Every item is a real published experiment. Click a card to see the second half of the truth as well: what it did not show.

{{ c9 }}
real experiments
{{ c26 }}
years of searching · 2000→2026
{{ c1 }}
missing step
{{ exp.year }} {{ exp.title }}
{{ exp.tag }} {{ exp.plus }}
What they did: {{ exp.did }}
What they saw: {{ exp.saw }}
What they did NOT see: {{ exp.notSaw }}
{{ exp.ref }}

The honest verdict for today

Firmly established
✓A photon is not a classical bullet with a ready-made story.
✓A late choice determines which kind of pattern (stripes, entanglement…) can be found in the joint data.
✓The “wave or particle” story did not exist before the measurement.
✓Entanglement links even photons that never existed at the same time.
✓The full measurement context matters more than any single moment.
Not established
✕That even one old instrument record ever physically changed.
✕That a future choice changes early statistics without sorting by late data.
✕A channel for sending information into the past.
✕Controllable faster-than-light communication.
✕An external simulation or an “extra level” of reality.
Beyond this point — reasoning territory hypothesis · not confirmed

The project hypothesis: a light transaction

What if the emission of a photon, its path and its future absorption are not a chain of events one after another, but one whole process, coordinated across the entire history at once? Like a book page written as a whole rather than printed letter by letter.

Then the future “boundary condition” — say, whether a mirror or an absorber awaits the photon at the end — could leave a faint trace in the statistics before light had time to make the round trip.

Similar ideas are seriously discussed in physics: the transactional interpretation (Cramer, 1986), retrocausal models (Wharton and Argaman, 2020), and in 2026 the Mukherjee–Hance theorem appeared: quantum theory is incompatible with all “classical” assumptions about time at once — something will have to be sacrificed.

If the connection were “instantaneous” — three versions of the world:
Athe world has a hidden absolute time — its own “tick”, and relativity is only an internal law;
Binstantaneous correlation without a controllable signal — entanglement already works this way;
Cnothing flies at all: the whole history is settled as one piece, and “speed” is an illusion of the inside view. This version is closest to the hypothesis.
Two pictures of one event
The familiar picture:
born→ flew→ absorbed
Hypothesis:
source⟷ field⟷ mirror⟷ detector
one coordinated “transaction”, settled as a whole
Important: existing experiments are compatible with this picture, but do not prove it. Only a new experiment can test it ↓
The climax · an experiment that has never been done

The future light-boundary test

status: not performed

All past experiments compared data with sorting by late results. The missing experiment tests something else — a simple and hard thing:

“The button is pressed later — yet the instrument statistics recorded earlier depend on which button was pressed.”
The result forbidden by standard physics: P(A | Y=0) ≠ P(A | Y=1)
The protocol in plain words
1A source emits a photon. Its early state (A) is recorded and sealed.
2Only after A is recorded, but before the photon reaches the distant boundary, a generator randomly picks the action Y: mirror or absorber.
3Only early data are analyzed — up to the moment light could have returned.
4No sorting by late results. The whole sample, entire.
What the result will say
—Early data identical for any Y → standard physics holds, the hypothesis narrows. The most likely outcome.
—A difference appears and is reproduced by independent groups → violation of the “no signals backwards” principle. New physics.
—Even then, simulation is just one hypothesis alongside retrocausality and a global law of coordination.
—And if the early record follows not the button but the late-chosen meaning of the code (today “red = zeros”, tomorrow the reverse) — that is no longer a blind law, but something that understands the protocol. The beginning of a dialogue.
no post-selection Y chosen after A is recorded ordinary communication excluded different distances L independent replication dummy switchings decoherence control analysis pre-registered
Honesty first · limits of inference about simulation and God

What even success will not prove

Even a perfect result will not immediately prove
✕a computer simulation;
✕a religious God;
✕an afterlife;
✕a separate soul;
✕an extra spatial dimension.
But it will prove one narrow, colossal thing
“The ordinary model, in which information forms only from past to future inside the light cone, is incomplete”
After that, simulation becomes a serious hypothesis, but it will compete with: retrocausal quantum theory · absolute time · superdeterminism · a new geometry of spacetime · a global law coordinating the whole history.
How to tell an unknown law from an interlocutor
1 · a new code is chosen late→ 2 · early sealed data answer by that code→ 3 · the next question depends on the answer→ 4 · the chain repeats→ 5 · another team reproduces it
A constant natural bias could raise the number of zeros. But it is under no obligation to understand a freshly chosen language every time and hold a consistent conversation. Only an adaptive dialogue justifies speaking not of a law, but of a system that understands the protocol.
First — honestly a check of everyday “predictions”

Tarot, dreams and “non-random randomness”

Inside the project hypothesis, cards, dreams and chance encounters look like possible interfaces: an outer level needs no miracle — just picking one allowed outcome among many. Logically it is imaginable.

But physics does not predict it. And a rigorous multi-center attempt to replicate the psychological retrocausal effect on 2004 participants produced data favoring its absence. Sensations easily vanish under honest replication.

A dream usually “comes true” in hindsight: out of hundreds of images you remember the one that matched. The honest test: record the dream before the event, fix one unambiguous prediction, forbid reinterpretation, count all dreams and repeat hundreds of times. Until then, trading on a dream is an experiment not on nature but on your own deposit.

The scientific formula
“Tarot as a retrocausal channel is logically imaginable but experimentally not established. It must not be used for decisions about health, money or safety”
This is a story
“I ran into the same person three times — a sign of fate”
This is a potential discovery
“A sealed device answered correctly, ten times in a row, questions that were formulated only later”
A chain of coincidences becomes science when it is: defined in advance · carries unknown information · repeats · follows a changing code · vanishes in controls · is reproduced by others.
And this is no longer esoterics · devices and networks

10 serious applications of this strangeness

None of these technologies proves that the past changes. But all of them grew out of the same questions: did the result exist before measurement, what randomness is, whether a single causal order is mandatory.

01already working
Randomness you can prove
Can you prove a number is random and not secretly planted by the manufacturer?
Violating a Bell inequality certifies randomness even without trusting the device (2018). Crypto keys, lotteries, audits, experimental settings.
02being deployed
Communication where a spy leaves a trace
Can a channel be eavesdropped without leaving a physical trace?
No: measurement enters the process and changes the statistics — interception shows up as rising errors. A ~4,600 km quantum-key network: 700+ fiber links, 2 satellite links, 150+ users (2021).
03laboratory
Imaging with light the camera never saw
Can a camera image an object its photons never touched?
Yes. Photon pairs of different wavelengths: one touches the object, the other is detected (Zeilinger, 2014). IR microscopy with cheap visible-light detectors.
04working in LIGO
“Empty” vacuum hears black holes
How can emptiness improve a telescope?
The vacuum fluctuates. Squeezed light redistributes uncertainty — LIGO cut quantum noise by ~28% (2.15 dB) and hears black-hole mergers better.
05laboratory
A radar that knows its own signal in noise
Can you find an object whose echo drowned in noise?
Quantum illumination: the signal is compared with a stored “twin” — the advantage partly survives even after entanglement dies (2013; up to 40% beyond the classical bound, 2021).
06working · closest to our theme
Future data refine the past
Can a later measurement improve knowledge of an earlier state?
Yes — smoothing uses both past and future samples (2010→): magnetometers, qubit trajectories, accident analysis. The future does not change the past — it refines its reconstruction. The most honest real analogue of the project formula.
07laboratory
A measurement you can undo
If a measurement changed the state — can the measurement itself be undone?
For a weak measurement — partly yes: uncollapse on a superconducting qubit (2008). Improves teleportation and qubit protection. This reverses the effect, it does not change the past.
08field networks
Quantum internet
Can nodes be linked without sending a particle between them?
Entanglement swapping entangles particles that never met — even ones that never coexisted (2013). A three-node network (2021), teleportation between non-adjacent nodes (2022).
09field prototypes
Clocks and navigation on photon pairs
Can distant clocks be synced by photon correlations?
Synchronization over 4 km of fiber (2016); two-way quantum time transfer with picosecond precision in field conditions (2024). A backup for GPS outages.
10laboratory
Transmission without a fixed order
Two channels each destroy information. Will a superposition of orders help?
The quantum switch holds the orders “A→B” and “B→A” in coherent superposition — nonzero information passed through a pair of fully noisy channels. Not a time machine, but the causal order is not classically defined.
Already really working
Quantum random-number generators · key distribution · squeezed light in LIGO
Field demonstrations
Quantum time transfer · multi-node networks · teleportation between non-adjacent nodes
Serious laboratory work
Imaging with undetected photons · quantum illumination · retrodiction · measurement undoing · the quantum switch
“Science has not yet learned to read the future. But it has already learned to use the fact that quantum reality is not obliged to have one ready-made classical history”
randomness observation uncertainty entanglement vacuum causal order future data measurement reversibility
— all of this used to look like philosophy. Now these are engineering resources.
The philosophical formula of the project
“Perhaps the photon is not a traveler outside time. But the bundle ‘emission ⟷ absorption’ may be part of a structure determined as a whole rather than in sequence. Then light is the border where our internal time touches the connectedness of the entire history”
Proven today: a late choice changes the accessible joint correlations.
Not proven: that it changes the early unconditioned record.
The difference between these two sentences is this whole project.
The living diary of the project

The history of this investigation

Project milestones and admin-approved submitted ideas land here. Your submitted thought may become the next entry.

{{ j.date }}
{{ j.tag }}
{{ j.title }}
{{ j.text }}
Library · key publications and experiments

The 17 works everything stands on

Every claim on this site rests on published science. Here is the full list — with direct links to the papers.

01 · Reviews of Modern Physics, 2016
Ma, Kofler, Zeilinger — Delayed-choice gedanken experiments and their realizations
The main review of delayed choice, the quantum eraser and delayed-choice entanglement swapping.
02 · Science, 2007
Jacques et al. — Experimental Realization of Wheeler's Delayed-Choice Gedanken Experiment
Single photons, with the configuration chosen after the photon entered the interferometer.
03 · Physical Review Letters, 2000
Kim et al. — Delayed "Choice" Quantum Eraser
Late keeping or erasing of the path information of an entangled photon.
04 · PNAS, 2013
Ma et al. — Quantum Erasure with Causally Disconnected Choice
A quantum eraser with a causally disconnected active choice.
05 · Nature Physics, 2012
Ma et al. — Experimental Delayed-Choice Entanglement Swapping
A late measurement classifies already-registered photons into entangled or separable subsamples.
06 · Physical Review Letters, 2013
Megidish et al. — Entanglement Swapping between Photons that Have Never Coexisted
Entanglement between photons whose lifetimes never overlapped.
07 · Science, 2012
Peruzzo et al. — A Quantum Delayed-Choice Experiment
A quantum-controlled beamsplitter and a superposition of measurement modes.
08 · Nature Communications, 2015
Procopio et al. — Experimental Superposition of Orders of Quantum Gates
A photonic superposition of two orders of quantum operations.
09 · Physical Review X, 2021
Ferreira et al. — Collapse and Revival of an Artificial Atom Coupled to a Structured Photonic Reservoir
The reflected photonic feedback appears after the full round-trip time.
10 · Reviews of Modern Physics, 2020
Wharton, Argaman — Bell's Theorem and Locally Mediated Reformulations of Quantum Mechanics
A review of retrocausal and “all-at-once” models.
11 · Reviews of Modern Physics, 1986
Cramer — The Transactional Interpretation of Quantum Mechanics
A quantum event as a transaction of advanced and retarded waves.
12 · Physical Review Letters, 2018
Chaves, Lemos, Pienaar — Causal Modeling the Delayed-Choice Experiment
Why the usual delayed-choice scheme by itself does not yet prove retrocausality.
13 · Physical Review A, 2019
Polino et al. — Device-Independent Test of a Delayed Choice Experiment
A stricter test of delayed-choice nonclassicality via a dimension witness.
14 · Physical Review Research, 2026 · new
Mukherjee, Hance — Limits of Absoluteness of Observed Events in Timelike Scenarios: A No-Go Theorem
A theorem on the incompatibility of quantum mechanics with simultaneously keeping absoluteness of events, no retrocausality and several other classical assumptions.
15 · MIT OpenCourseWare
Lecture notes on null geodesics
The rigorous formulation of a light trajectory with zero proper interval and the absence of an ordinary four-velocity for the photon.
16 · Physical Review Letters, 2014
Campagne-Ibarcq et al. — Observing Interferences between Past and Future Quantum States
The weak signal of a qubit depended on the preparation and the future post-selection: the future “flinch” is visible in the past — but readable only after the future key.
17 · Physical Review D, 2025
Causality in Relativistic Quantum Interactions without Mediators
Microcausality: locally measurable quantities do not depend on actions that could not yet causally reach the measurement point. This is exactly the principle our experiment tests.
Frame 001 · the people behind the project

Author of the idea — Evgeny Vlasov

Project author with family
That is him — the author of the idea. Mouth open: he just grasped the scale of the question
frame 001 · the full support-and-skeptics group both roles matter
Dossier
Name: Evgeny Vlasov
Role: author of the idea and head of the investigation
Distinguishing mark: the mouth opens by itself — every time the Universe answers “well, no”
Status: wants to believe. But verifies
“Thanks to my whole family for supporting and not supporting my ideas, and for pushing me to figure everything out in even more detail”
— caption under frame 001. The skeptics in the family are the control group of the experiment.
Become part of the investigation

Submit an idea or a line of reasoning

The button will compose an email to the project admin in your mail app. If the idea passes review, it will be built into the research diary with the author name.

{{ sendHint }}
The truth is out there
and it is · testable