Nobel Prize Winner Warns Something Is Wrong With the Universe — And James Webb Just Confirmed It（原声文稿）

Transcript of 诺奖得主警告：宇宙膨胀对不上数 （中文完整文稿）。

In 2023, a team of astronomers led by Adam Rees received a batch of new measurements from the James Webb Space Telescope.

They were measurements of stars, pulsing stars in distant galaxies.

Stars this team had already measured many times with the Hubble Space Telescope. That was the point.

For years, critics had argued that Hubble's view of those stars was too blurry, that each one might actually be several stars smeared together into a single dot.

If that were true, a problem that had been haunting cosmology for a decade would simply disappear. The problem would turn out to have been a smudge all along. Webb was sharp enough to settle it.

One answer would make the problem go away. The other would make it nearly impossible to explain away.

To understand why so much was riding on those few faint points of light, you need to know about two numbers.

Right now, there are two ways to measure how fast the universe is expanding.

Both are careful, both are independent, built by different teams using different instruments to look at different eras of cosmic history.

One says the universe is expanding at about 67, the other says about 73.

That sounds like a rounding problem, but those two numbers describe two different universes, with different ages, different histories, and possibly different endings.

After decades of checking, rechecking, and building new telescopes specifically to find the mistake, nobody has found it. There are three ways this can end.

A hidden error, a strange place in the cosmos, or a missing piece of physics.

By the end of this video, you'll know which way the evidence is leaning, meaning, and the exact date, just weeks from now, when one of the most important pieces of evidence is set to arrive.

Adam Rees shared the 2011 Nobel Prize in Physics for co -discovering that the expansion of the universe is speeding up.

Since then he has led a project with one job – measure today's expansion rate as precisely as humanly possible.

The title of this video is our wording, not a quote from him, but it's a fair summary of his position.

He has argued that the gap between these two numbers is very unlikely to be a measurement mistake and that it may mean our picture of the universe is missing something.

Not everyone agrees, and one of the people who disagrees has a better claim than almost anyone to being right – you'll meet her shortly.

First, what do these numbers actually mean? The number is called the Hubble constant.

It tells you how fast galaxies are moving away from us depending on how far away they are.

A value of 73 means that for every 3 .26 million light years of distance, a galaxy recedes 73 km per second faster.

A galaxy 326 million light years away is moving off at about 7 ,300 km per second.

If the true number is 67, that same galaxy is moving at 6 ,700.

The difference between the two answers is more than 2 million km per hour.

Run the expansion backwards, and this number also tells you how long ago everything was packed together.

It's a clock for the entire universe, and right now, that clock shows two different times.

You might expect astronomers to have panicked.

For years, most of them didn't, because this exact number has fooled people before.

In 1929, Edwin Hubble published the first measurement of cosmic expansion. His value was about 500.

run the clock backwards with that and the universe comes out roughly 2 billion years old, younger than rocks that geologists were already dating on Earth.

It took decades to find out what went wrong.

Hubble had mixed up two different types of pulsing stars and he had mistaken glowing gas clouds for stars. Each correction pushed the number down. By the 1970s astronomy had split into two camps.

One insisting the number was about 50 HD, and the other insisting it was about 100, and the feud lasted for years.

What finally ended it was the Hubble Space Telescope.

In 2001, a large team called the Key Project published a value of 72, give or take about 8. The leader of that project was Wendy Friedman.

Remember that name. So, by the early 2000s the lesson seemed clear.

This number is hard to measure, the mistakes get found eventually and the answer is somewhere in the low 70s.

Then a satellite looked at the oldest light in the universe and the number it predicted wasn't 72.

Picture the universe when it was 380 ,000 years old. There are no stars and no galaxies.

Everything is a glowing fog of hydrogen and helium at around 3000 degrees, about the temperature of the surface of a cool red star.

Light can't travel through it.

Every photon is scattered, almost the instant it's released, like a flashlight beam swallowed by smoke.

The fog isn't still. Gravity pulls matter inward.

The pressure of light pushes it back out and sound waves ripple through the entire young universe.

Then the universe cools enough for the fog to clear.

Light breaks free and it has been travelling ever since.

Stretched by billions of years of expansion, it now reaches us as faint microwaves from every direction in the sky.

This is the cosmic microwave background, the oldest light there is.

Frozen into that light is a pattern left by those ancient sound waves.

The waves could only travel a certain distance before the fog cleared, so the pattern contains a ruler of known size, written across the whole sky.

Today, that ruler has been stretched to roughly 500 million light years.

Between 2009 and 2013, the European Space Agency's Planck satellite mapped that pattern across the entire sky.

Telescopes in Chile and at the South Pole measured it independently.

But the microwave background doesn't hand you today's expansion rate directly. It's a baby photo.

To get today's number, you have to feed that photo into a model, a recipe that predicts how the baby grows into the adult.

That recipe is called Lambda CDM.

In it, ordinary matter, every star, planet and person, makes up only about 5 % of the universe.

About 27 % is dark matter, which has never been directly detected.

About 68 % is dark energy, which nobody can explain.

Feed Planck's map through that recipe, Ask how fast the universe should be expanding today, and the answer is about 67 .4, with an uncertainty of about half a point.

The other number comes from a very different place, and it starts with a woman almost nobody has heard of.

In the early 1900s, Henrietta Swan Levitt worked at Harvard College Observatory as one of the women employed to study photographic plates of the sky by hand.

Examining stars in the Magellanic Clouds, two small galaxies near our own, she noticed something about the pulsing stars called Cepheids, the brighter ones pulsed more slowly.

That meant that if you time a Cepheids pulse, you know how bright it truly is.

Compare that with how bright it appears and you know how far away it is. Levitt published the relationship in 1912.

She had given astronomy its first reliable measuring tape for the universe, and she died in 1921 before most of its consequences were understood.

Modern astronomers use her discovery as the middle of what's called the distance ladder. The bottom rung is pure geometry.

As Earth orbits the Sun, nearby stars shift slightly against the background and that shift gives their distance directly. That calibrates Cepheids close to home.

Cepheids then measure the distances to galaxies containing a certain type of exploding star, the Type 1A supernova, which reaches a reliable peak brightness and can be seen across billions of light years.

Those supernovae carry the ladder out into the deep universe.

Reese's project has spent about two decades tightening every rung.

Its answer is about 73, with an uncertainty of about one point.

So we have 67 .4 on one side and 73 on the and the error bars don't come close to touching.

In physics the standard for a signal too strong to be chance is called 5 sigma, roughly a 1 in 3 .5 million probability that it's a statistical fluke.

It's the bar the Higgs boson had to clear before anyone could call it a discovery.

The gap between these two measurements reached about 5 sigma.

Put the higher expansion rate into the standard model without changing anything else and the universe comes out roughly a billion years younger.

That's longer than complex animal life has existed on Earth.

One of these numbers had to be wrong, or the recipe connecting them was. Most astronomers bet on the ladder.

Long chains of measurements have more places to break, and there was one weak link everybody kept pointing at, the blurry stars.

The galaxies hosting those supernovae are tens of millions of light years away.

Their cepheids are faint dots in crowded fields of other stars, and Hubble measured them in infrared light, where its vision is less sharp.

If neighbouring stars were blending into each cepheid, every one of them would look slightly brighter than it really was.

Brighter means closer, and closer means a higher expansion rate.

It was a reasonable suspicion, the kind of mundane era that has fooled astronomers before.

If it was true, 73 would come down and the crisis would fade. That brings us back to Webb's first measurements.

Webb's 6 .5 -metre mirror and infrared cameras see several times more sharply than Hubble at those wavelengths.

Pointed at the same stars, it could pull apart anything that Hubble had blended.

Across 2023 and 2024, Reese's team measured more than a thousand sephiids with Webb.

The crowding was there in the images, and Webb separated it cleanly. The distances barely moved. Webb confirmed Hubble's measurements.

The team reported that the crowding explanation could be rejected at about eight sigma, far beyond the threshold for ruling something out.

Usually in science, a strange result meets a sharper instrument and dissolves.

Here, the sharper instrument made the problem harder to explain away.

The stars weren't a smudge.

They had been measured correctly all along, and they still pointed to 73. That should have been the end of the argument.

And it wasn't, because of the person who had ended the last one.

Wendy Freedman, now at the University of Chicago, has never been convinced the tension is real.

She doesn't think Reese's team is careless.

Her concern is that Cepheids are complicated stars, affected by dust, by their chemistry, history, and by the crowded regions they live in, and that a ladder built mostly on one kind of star can hide errors nobody has thought to look for. So she built a different ladder.

Her team uses red giants at a specific stage of their lives.

Just before a red giant ignites helium in its core, it reaches a maximum brightness that is almost identical from star to star, which makes it an excellent distance marker.

These stars live in quieter, less dusty parts of galaxies, away from the crowded spiral arms where cepheids are found.

Using web and Hubble together, Friedman's team gets an answer of about 70 .4. That lands between the two camps.

Her team argues that within its uncertainties it's consistent with the early universe value and that there may be no crisis at all.

Others disagree sharply.

Her The critics point out that 70 sits in the middle of the tension rather than resolving it.

They also argue that when you apply the same stars to the complete set of supernova galaxies rather than a smaller selection, the answer climbs back towards 73.

Friedman's side counters that their selection was about quality and that the cepheid measurements still carry hidden risks.

Same telescope, many of the same galaxies and two of the most experienced teams in the world reading different answers out of them. This disagreement is real and it isn't settled.

So the question becomes, who else can testify?

When two witnesses disagree, you look for others who have no connection to either.

The first comes from gravitational lensing.

When a distant quasar sits directly behind a massive galaxy, the galaxy's gravity bends its light into several images.

Each image follows a different path, so when the quasar flickers, the images flicker days or weeks apart.

Those delays depend directly on the expansion rate, without any distance ladder at all.

Lensing measurements have mostly come out high, closer to 73, though with larger uncertainties.

The second comes from ripples in space time itself.

In 2017, In 2017, gravitational wave detectors caught two neutron stars colliding and telescopes saw the explosion.

The shape of the wave signal gives the distance directly, with no ladder involved.

That single event gave an answer around 70, but with error bars so wide it can't yet pick a side. The third comes from galaxies themselves.

Those same ancient sound waves left a faint pattern in how galaxies are spread across space.

The dark energy spectroscopic instrument, DESI, has mapped millions of galaxies to read that pattern.

Combined with early universe information, it comes out close to 68.

Line all the witnesses up and something strange appears. The disagreement isn't really between teams.

Methods that measure the nearby universe directly keep landing high.

Methods that start from the early universe and project forward through the standard model keep landing low.

The This split follows a line in time, not a line between people, and that line is exactly where the three possible endings diverge.

The first ending is that the error is still hiding in the late universe, something subtle in how stars or supernovae behave that affects many local methods the same way. Friedman's work keeps that possibility alive.

If it's true, the standard model survives, and this becomes one more chapter in the long history of a number that's very hard to measure.

The second ending is that we live in an unusual place.

If our region of the universe sits inside a vast zone with less matter than average, nearby galaxies could appear to be moving away slightly faster than the universe as a whole.

It's an appealing idea, but most tests so far suggest such a region would need to be unusually large and empty to explain the full gap.

The third ending is the one that makes cosmologists nervous.

Both measurements are right and the recipe is incomplete.

Remember, the early universe number is never measured directly. It comes from running the baby photo through a model.

If the model is missing an ingredient, the prediction will be off, even if every measurement is perfect. Physicists have proposed several candidates.

One is a 200 ,000 years, often called early dark energy.

It would shrink that ruler printed in the microwave background, and a smaller ruler means faster expansion today.

Others add unknown particles to the early universe, or suggest gravity behaves differently on the largest scales. And there's a newer hint from DESI.

Dark energy itself may not be constant, and may have been weakening over billions of years.

Depending on which data are combined, the evidence ranges from about 2 .8 to 4 .2 sigma. That isn't proof, but it isn't nothing. So far, no single proposal fixes everything.

Several ease the Hubble tension while making other measurements fit worse.

In physics, that usually means you haven't found the real answer yet.

You've found something nearby. There's a precedent for exactly this kind of standoff.

In the 1960s, the chemist Ray Davis built a detector nearly a mile underground in a South Dakota gold mine, a tank holding hundreds of thousands of litres of cleaning fluid designed to catch neutrinos streaming from the Sun.

Physicists had a detailed model of the Sun that predicted how many he should catch. He caught about a third of them.

For roughly 30 years this was the solar neutrino problem.

Many people assumed the experiment was flawed. Others blamed the model of the Sun.

Both were checked again and again, and both held up. The answer was neither.

Neutrinos were changing from one type into another on their way to Earth, and Davis's detector could only see one type.

The experiment was right, the model was right, and the missing piece was new physics. Two Nobel Prizes came out of it.

Nobody knows yet whether the Hubble tension is another Solar Neutrino problem, but it has the same shape, two careful measurements, decades of checking, and a gap that refuses to close.

This matters beyond cosmology, because this number runs through almost everything we say about the universe.

It sets the age of the cosmos and every distance to a faraway galaxy, and it feeds every calculation about how early galaxies formed.

When a documentary says the universe is 13 .8 billion years old, that figure assumes the recipe is complete.

It also shapes the ending.

If dark energy is truly constant, the universe expands forever, with galaxies drifting out of sight until the sky goes dark.

If dark energy is changing, that ending is no longer guaranteed.

The fate of the universe, which a generation ago was treated as settled is back to being an open measurement. None of this means the Big Bang is wrong.

The evidence that the universe was once hot and dense and has been expanding ever since is overwhelming.

What's in question is the recipe we use to describe that history. So when do we find out?

The first major piece of new evidence arrives on December 2nd, 2026.

That's the date the European Space Agency releases the fourth major data release from its Gaia mission, based on five and a half years of measurements of stars across our galaxy, with significantly smaller uncertainties than before.

Gaia measures the bottom rung of the ladder, the geometric distances to stars close to home, including the cepheids that calibrate everything above them.

If there's a flaw in that foundation, it's one of the places it would show.

If the foundation holds, the case for 73 gets stronger and the pressure on the recipe grows.

It won't settle the question on its own, but it's the first in a line of tests now arriving.

NASA's Roman Space Telescope launched on August 30th, 2026 and is built to map dark energy and find supernovae across enormous areas of sky.

Europe's Euclid Telescope is mapping billions of galaxies.

The Rubin Observatory in Chile is discovering supernovae and lensed quasars in huge numbers.

Gravitational wave detectors keep catching new collisions, each one an independent measurement with no ladder at all.

And new microwave experiments will test the early universe side more precisely than Planck could.

If the gap is an error, these instruments should expose it.

If it survives all of them, the question stops being where's the mistake and becomes what's missing from physics? Go back to those first web measurements.

A lot of people hoped those stars would turn out to be a blur because a blur would have meant the universe still matched the recipe. They weren't a blur.

They were stars, measured correctly, telling the same story they had been telling for years.

Maybe that story still has an ordinary ending, an error found in some corner nobody thought to check. That has happened before, with this very number.

Or maybe it's the first faint sign that the universe contains an ingredient we haven't named yet. The same way a missing two -thirds of the Sun's neutrinos turned out to be.

Either way, the universe was never obligated to match our recipe.

The recipe was our best guess, and the evidence is now testing whether that guess is complete. So, here's my question for you.

If you had to bet on the true expansion rate of the universe, Would it be 67, 70 or 73?

Leave your number in the comments and when the answer comes in, we'll see who was right.

Gaia's release lands on December 2nd. Subscribe if you want to find out what it shows.