Last year, the Dark Energy Spectroscopic Instrument-DESI-made headlines around the world.

Its enormous map of the Universe hinted that dark energy-the mysterious something causing the expansion of the Universe to accelerate-might not actually be constant.

And just last month, another giant cosmology experiment released its final results.

The Dark Energy Survey.

DES.

Its results are intriguing.

But did they confirm DESI?

We'll, we'll see, but there's also a much more interesting question- which is why are completely different scientific studies so much more powerful in combination than any one study on its own?

The answer brings us to the heart of what science really is.

Did DES confirm DESI?

Seems like a reasonable question, no?

New data is supposed to reduce the uncertainty on whatever number we're trying to measure.

Tighten the error bars and we confirm the measurement.

Science sometimes works that way-like when we first measured the existence of dark energy in the first place.

But modern cosmology has reached the point where this doesn't work any more.

The signals we're looking for are so subtle-so tangled together-that it's no longer about directly measuring the parameters of this universe-it's about ruling out countless hypothetical universes that we're not in.

And DES was ruthless in eliminating universes.

So what did it discover?

Despite their nearly identical names, the Dark Energy Survey, or DES, and the Dark Energy Spectroscopic Instrument, DESI, were built to tackle the same grand mystery from almost opposite directions.

DESI's mission is beautifully focused.

It measures the distances to tens of millions of galaxies, constructing the most detailed three-dimensional map of the Universe ever made.

Hidden within this vast cosmic atlas is a faint fossil pattern left behind by sound waves that rippled through the hot plasma of the infant Universe nearly 14 billion years ago.

These are the Baryon Acoustic Oscillations, which we've covered before.

BAOs gives us a cosmic ruler that grow with the expanding universe, and so allows DESI to trace the history of that expansion with extraordinary precision.

DES took a much broader approach.

This truly international collaboration of some 400 scientists took control of the Blanco 4-meter telescope at the Cerro Tololo Inter-American Observatory in the Chilean Andes.

After building a dedicated 570 megapixel camera, they scanned an enormous region of the southern sky for six years, building an extraordinarily deep portrait and movie of 300 of million galaxies.

This became the foundation for several completely different ways of probing the cosmos.

DES measures the tiny distortions in distant galaxy shapes caused by the subtle warping of intervening space by massive objects.

Through this weak gravitational lensing, DES maps the invisible distribution of dark matter and tracks how cosmic structure has grown over time.

It also measures how galaxies have clustered together over cosmic time due to their mutual gravity, giving a completely different perspective on this same structure growth.

By combining those two measurements through galaxy-galaxy lensing, DES connects the visible galaxies to the much larger halos of dark matter that surround them.

And DES also discovered and monitored thousands of Type Ia supernovae, which allowed it to measure the expansion history of the Universe by determining the distances to these exploding white dwarf stars.

And, by the way, this is exactly how dark energy was first discovered.And finally DES even measures the same ancient baryon acoustic oscillation pattern as DESI-but with a completely different observing strategy.

So DES is really an entire laboratory of cosmological probes built from one extraordinary survey of the sky.

DESI, meanwhile, takes one of those probes and pushes it farther than anyone ever has.

Why try to measure dark energy by so many different methods?

It's not because we're worried that any one of those methods might be wrong-at least, not mostly.

It's because each one tells us something slightly different about the universe.

And it's only in the combination of all of our "cosmological probes" that we find the answer that we were looking for.

The truth is, there's no one experiment that actually measures dark energy.

Instead, we observe galaxies and their distances to each other and tiny distortions in their shapes and the faint flashes of exploding stars.

We observe the miniscule temperature fluctuations in the oldest light in the Universe-the cosmic microwave background.

None of these directly measure dark energy.

Instead, each measurement is influenced by a set of unknown cosmological parameters that together determine the way the universe expands and how structure forms, which in turn drive the observables that we can collect.

It's these cosmological parameters that we really want-things like the proportions of dark energy to dark matter to regular matter to light to neutrinos.

Or The expansion rate at the beginning and the slowing and speed of that rate due to gravity and dark energy.

The parameters of the equation of state that determine if and how dark energy changes over time.

The subtler details of how light and atoms and dark matter and dark energy interact with themselves and with each other.

Each tiny variation in the configuration of these parameters represents an entire possible universe.

Cosmology is about figuring out which of these infinite possible universes we are in.

But we can't do that just by measuring the cosmological parameters one by one-there is no observation that will give us these parameters.

Instead, we use our observations to determine which hypothetical universes we are NOT in, over and over until we're left with the universe we are in, like a cosmic game of 20 questions.

The cosmological parameters that are of most interest to DES and DESI are the parameters of the dark energy equation of state.

This is perhaps the simplest way to describe a changing dark energy-we have the baseline dark energy strength, omega-0, and the rate of change of dark energy, omega-a.

Before we ever discovered dark energy, as far as we knew we might have been any one of a vast distribution of possible universes in the space defined by these parameters.

With the first observations of white dwarf supernovae we culled huge numbers of these universes to demonstrate that we're in a universe with at least some dark energy described by omega-0.

The supernova studies had ask the universe a question: they asked how has cosmic expansion dimmed distant supernovae?

The universe answers in a way that many possible universes couldn't have answered.

So we were able to eliminate those universes.

Then the cosmic microwave background studies asked a different question: what did the early universe look like?

The answer rejects different universes.

The baryon acoustic oscillation studies like DESI ask: how did the fossil parameters from the early universe expand?

Again, different hypothetical universes are eliminated.

DES asks some of the same with new data, and also asks how dark matter has warped galaxy images and drawn galaxies together against that expansion.

Again, counterfactual realities were slain by DES in this case.

In the end, we're left with a much smaller pocket of possible universes where the cosmological parameters are now much more tightly constrained than they were before we began DES.

Here's an analogy.

I'm thinking of two whole numbers between one and ten and you can ask a series of questions to figure them out, but you can't ask what the actual numbers are.

Let's say you ask what the sum of the numbers that you're thinking of.

I tell you it's 10.

You immediately know that the numbers are 1 and 9 or 2 and 8 or 3 and 7 or 4 and 6 or 5 and 5.

That's five possibilities assuming that order of the numbers doesn't matter.

Or you could instead ask what the product is of the two numbers.

I tell you it's 24-so that could be 4 and 6, 3 and 8-two options.

But only one of the answers to the second question is consistent with the answer to the first question.

My numbers had to be 4 and 6.

Only by asking multiple questions can we infer the original numbers.

In this analogy, the numbers are our cosmological parameters, and the function-the sum or product-is one aspect of the complex physics that turns those numbers into observables.

For example, one "function" might be "how do the cosmological parameters govern dark matter halo growth and so warp the shapes of distant galaxies through weak lensing" or "how do they determine baryon acoustic oscillation formation and then expansion".

Unfortunately, we don't get to choose these "functions" or questions.

They're what the universe offers us.

When we come up with a cosmological probe, we're identifying observables that best pry the cosmological parameters in the context of all the other available probes.

What we're trying to do is to minimize what we call degeneracies.

In the context of our cosmological probes, degeneracies are the multiple plausible cosmological parameters that could have led to the same observations.

Like how there are several combinations of whole numbers that sum to 10.

Sometimes a probe is good at constraining one parameter over others, sometimes a probe contains a combination of parameters without giving any single constraint.

We want a combination of probes that whittle down the degeneracies and leave us with tight constraints on the parameters we're interested in.

All we need to do is combine the datasets from these different probes.

These various cosmological probes are not independent of each other and so we can't combine the constraints they give in a simple way.

These different probes may observe many of the same galaxies, or rely on the same potentially biased distance calibrations, or they make similar assumptions about how galaxies form and evolve.

That means some of their uncertainties are also shared.

In statistics, those shared uncertainties are called covariance.

Ignoring covariance is a bit like polling a hundred people who all got their news from the same rumor-you think you have a hundred independent opinions, but really you've only measured the same bias a hundred times.

So before cosmologists compare two experiments, they have to figure out exactly how much information is genuinely new, and how much is information they've effectively counted already.

It turns out that understanding those relationships between datasets is just as important as understanding the datasets themselves.

In modern cosmology, the challenge isn't simply collecting more evidence-it's making sure every new piece of evidence is genuinely asking a different question.

But in the end, reality is self-consistent.

There aren't separate Universes for DESI and DES, or for supernovae and the CMB.

There's just one universe, and it will always give consistent answers to the questions we ask of it.

We just need to choose the right questions that allow us to efficiently rule out the universes that could not have provided the answers that we receive.

So, what happened when DES finally asked its questions of the universe?

Analyzed on its own, the new DES results are beautifully consistent with the standard picture of cosmology.

A Universe dominated by a cosmological constant still provides a perfectly good description of everything DES observed.

It very slightly favors a changing dark energy.

DESI is similar-consisstent with the standard cosmology-there are surviving universes whose physics is exactly what we thought.

But there are slightly more surviving potential universes whose dark energy is actually diminishing.

Remember, DES wasn't built to verify DESI or vice versa.

Both were built to interrogate the Universe in different, complementary ways.

That means that together they have the power to eliminate more possible universes than either in isolation.

And when the DES results are compared with DESI, and the cosmic microwave background, and supernova observations, many of the same Universes do survive, but not all.

Further universe-culling pushes us slightly further towards favoring a decreasing dark energy.

Not enough to rule out physics-as-we-thought-it, but the intrigue does increase.

DES independently points to the same exciting region of parameters space as DESI, which is extremely important to rule out bias, while also shaving the edges because DES has different degeneracies to DESI.

The several different questions asked by DES and DESI were answered consistently by a smaller and smaller fraction of possible universes.

Perhaps the most exciting "science result" from DES is that the intriguing hints from DESI weren't immediately ruled out or disfavored.

That could've happened, and would have meant that some bias in the DESI method was probably at work.

No, that island of intriguing universes with decreasing dark energy is still in play.

So what's next?

We ask more questions of the universe.

We give those counterfactual universes more chances to fail our grilling and get eliminated.

That's why the next decade of cosmology is going to be incredible.

The Rubin Observatory will discover billions of new galaxies and measure the changing Universe with unprecedented depth and monitor it over a decade.

Euclid is already in space, mapping a vast patch of the sky with incredible spatial resolution.

The Roman Space Telescope will perform this exquisitely sharp imaging over a smaller scale, but with even better calibration and will also monitor over time.

DESI will continue expanding the largest three-dimensional map of the Universe ever assembled.

And we'll continue culling universes.

Modern cosmology is really a beautiful example of how humans have always learned about nature.

We know in our hearts that there's a world out there and that the world is self-consistent.

And when we ask questions of it-with our senses and with our experiments-it'll tell us the truth.

We believe in the model of the world that never fails a test-never answers wrong.

Every independent line of evidence needs to agree on the same story.

And even when we think we have the story, we keep asking questions.

At the frontier of science, the hardest part isn't discovering something new, it's convincing ourselves that we haven't fooled ourselves.

Our current cosmological models, along with all of our best scientific theories in their broad strokes-from evolution to climate change to germ theory to the Big Bang-are the explanations of a reality that have withstood every question that we've so far asked of it.

But these also still represent families of possible universes that still need whittling down.

So we keep asking questions, bringing us closer to pinpointing in parameter space this singular space time.