[theme music] Did all of dark energy just vanish?

A team of scientists have analyzed new data and claim that we need to completely rethink its existence.

[theme music] Back in 1998, two independent teams of astronomers made an extremely controversial announcement, that the universe is not only expanding, but that expansion is accelerating.

They discovered this by watching for the explosions of type 1a supernovae.

These exploding white dwarf stars have predictable brightnesses that allow astronomers to figure out how far away they are.

MATTHEW O'DOWD: With a set of reliable distances extending back several billion years, the teams were able to map the expansion history of the universe.

They expected to see that this expansion rate was slowing down due to the gravitational effect of all of the matter in the universe.

Instead, they found the expansion rate has been accelerating for half of the age of the universe.

It appeared that something was acting to counter gravity, an outward pressure that has come to be known as dark energy.

The finding led to a shared Nobel Prize for the leaders of the supernova hunter teams-- Adam Riess, Brian Schmidt, and Saul Perlmutter.

For a deep dive into the mysteries of dark energy, we actually already made this entire playlist on the topic.

It's pretty hardcore, so definitely check it out, but maybe after this video.

So why are we talking about dark energy again?

Because another team has just announced a new analysis of updated supernova data.

They claim that the data are consistent with there being no dark energy, no accelerating expansion.

They suggest the universe may be just expanding at a constant rate, never speeding up but also not slowing down.

This claim is just as controversial as the original discovery of dark energy, because dark energy is now the industry standard.

Of course the media jumped all over this and provided almost no useful info.

We thought it would be a good idea to figure out whether we should throw away nearly 20 years of work on dark energy and delete our playlist based on this result.

In October 2016, the team of Nielsen, Guffanti, and Sarkar published a paper titled "Marginal evidence for cosmic acceleration from type 1a supernovae."

It appeared in the prestigious Nature journal, and that helped to get a lot of attention.

As with the initial discovery of dark energy, these scientists used type 1a supernovae to track the expansion history of the universe.

However, in the 18 years since the first studies, we've observed a lot more of these exploding white dwarfs-- 740 of them compared by the 10 used by Riess and Schmidt and the 49 by Perlmutter.

The new guys claim that the much larger sample is consistent with there being no acceleration.

And more data equals more confidence, right?

So did dark energy just go away?

Actually, not at all, and here's why.

The new study actually agrees with the old dark energy results, mostly.

It finds that an accelerating universe containing dark energy still fits the data best.

The difference is that it finds that the data is also consistent with a wider range of possible expansion histories.

And that range now includes a history in which there never was any acceleration.

An accelerating expansion is still preferred.

It's just that a non-accelerating history is not excluded with quite the same confidence.

I want to talk about how scientists analyze astrophysical data like this, so let's talk numbers.

The new study claims a three sigma confidence that there is a positive cosmological constant.

So quick aside-- the cosmological constant, written as lambda, is the thing you add to Einstein's equations of general relativity to give the anti-gravitational effect of dark energy.

If the cosmological constant exists and is larger than zero, then dark energy is a real thing.

So three sigma confidence in a positive cosmological constant basically means this-- if you repeated this experiment many, many times, about 0.27% of the time, the uncertainties-- so The messiness in the data-- would cause a universe with no dark energy to just happen to look like one with dark energy.

So on average, about one in 300 experiments gives you a false three sigma result.

But given that many thousands of different experiments are being run by professional scientists at any one time, false three sigma results do happen.

So for a really big scientific claim, like the existence of dark energy, three sigma just doesn't cut it.

Scientists like to get at least five sigma significance.

False positive five sigma results only happen once per 3.5 million experiments.

So the new result doesn't give a high enough significance from the supernova data alone.

No one could claim a proof based on that.

It's at best in a very strong hint.

But here's something the press skipped-- the original papers by our Nobel laureates also claimed a significance of three sigma or lower for a positive cosmological constant based on that early supernova data alone.

So why did anyone pay any attention?

Because they didn't consider the supernova data alone, and nor should we.

If we include other stuff about the universe, our confidence in the existence of dark energy rockets well above five sigma for all of these studies.

We can't yet observe dark energy directly.

We can only infer its existence based on how it affects the expansion of the universe.

But that means we have to consider all of the things affecting that expansion when we decide whether dark energy is part of that equation.

In fact, there are only two factors that can change the way a universe expands-- there are things that tend to accelerate expansion, which we call dark energy; and there are things that slow expansion, which is just the gravitational effect of regular energy.

And that's a mostly dark matter, but also stars, planets, gas, radiation, et cetera.

This graph is how we like to show the balance of these energy types-- dark energy on the y-axis and normal energy-- so matter-- on the x.

To be a bit more precise, these numbers-- omega lambda and omega m-- represent the fraction of the total energy in the universe that these two types would comprise, assuming that the universe is flat, which it probably is, and I'll get back to that.

Those blue balls represent the ranges of combinations of matter and dark energy that are consistent with the new supernova measurements.

Now statistical hypothesis testing is a whole big topic, and I'll put some links into description.

But very crudely, the inner circle is the most likely region-- that's the 95% confidence, one sigma region.

But really, omega lambda and omega m could be anywhere in here, although the further from the center, the less likely.

The big controversy here is that the three sigma contour touches the zero omega lambda line.

So if we consider the supernova data by itself, represented by these contours, there appears to be a small chance that we lie on this part of the graph-- little or no dark energy.

Except that bottom left corner of the graph also represents a universe that has almost no matter in it either-- zero omega lambda but also zero omega in.

But our universe definitely has matter in it-- we even have a pretty good idea how much.

Counting galaxies and weighing dark matter tells us that omega m is probably around 0.3, but it's at least around 0.2.

That means we can rule out this entire section of the graph.

That alone rules out the region of the supernova results that suggest there's no dark energy.

Now this was known in the late '90s, which is why the first supernova results were taken seriously.

Another really powerful piece of evidence is that the balance of omega lambda and omega m define the geometry of the universe.

If these add together to equal one, then the universe is flat.

And by that I mean that parallel lines stay parallel, and the angles of triangles add up to 180 degrees, and all the regular rules of geometry work.

In a relatively empty expanding universe-- as represented by this part of the graph-- space would not be flat.

It would have a weird hyperbolic curvature.

I'm going to have to refer you to another for details on universe geometries, but the main point is that we can also figure out where we should be on this graph by measuring the geometry of this universe.

And, as I also talk about in that episode, we can do this using the patterns in the cosmic microwave background to measure the angles of universe-sized triangles.

Their geometry appears to be very flat, so our universe should lie on this line here.

In fact, I oversimplified slightly, but the CMB results place our universe somewhere in these orange regions, the one, two, and three sigma contours, based on the CMB measurements.

That little region where the supernova and CMB results overlap represents the most likely combination of dark energy and matter.

When you mathematically combine the certainty contours of two completely independent measurements, they give you a much tighter range of possibilities.

And the no dark energy region down there is so far from the combined likely region that we can rule it out with much more confidence than even five sigma.

And there's other evidence leading us away from that little corner also, like baryon acoustic oscillations, but I'm going to have to leave that one to our previous episodes also.

But to wrap up-- everything we know about the way gravity works, combined with the expansion history that we measure, tells us that there has to be something out there countering the gravitational effect of matter and flattening the geometry of space.

It's not even like accelerating expansion is that weird anymore.

We're pretty sure it happened even more quickly in a separate episode soon after the Big Bang in the event we call cosmic inflation.

So this is a thing that seems to happen in our universe.

Dark energy, whatever it is, is still a thing.

But this new study is still very important.

It demonstrates one of the greatest qualities of the scientific process and culture-- no matter how well-accepted a result is, everything we think we know is always subject to being questioned and retested.

We now need to understand why the new result edged the confidence down.

There were differences between the experiments, so what effect did these differences have?

You can be sure that many scientists will be taking a long, careful look at the evidence for dark energy and its effect on the expansion of spacetime.

[theme music]