WEBVTT 00:00.240 --> 00:04.320 Last year, the Dark Energy Spectroscopic  Instrument—DESI—made headlines around the   00:04.320 --> 00:09.280 world. Its enormous map of the Universe hinted  that dark energy—the mysterious something causing   00:09.280 --> 00:16.000 the expansion of the Universe to accelerate—might  not actually be constant. And just last month,   00:16.000 --> 00:21.520 another giant cosmology experiment released  its final results. The Dark Energy Survey.   00:21.520 --> 00:28.080 DES. Its results are intriguing. But  did they confirm DESI? We’ll, we'll see,   00:28.080 --> 00:33.280 but there’s also a much more interesting  question— which is why are completely   00:33.280 --> 00:38.800 different scientific studies so much more  powerful in combination than any one study   00:38.800 --> 00:50.080 on its own? The answer brings us to  the heart of what science really is. 00:38.800 --> 00:43.760 much as it's inspired us, check it out at the  Space Time merch store. Now on to the episode. 00:43.641 --> 00:50.281 Did DES confirm DESI? Seems like a reasonable  question, no? New data is supposed to reduce   00:50.281 --> 00:56.441 the uncertainty on whatever number we’re trying  to measure. Tighten the error bars and we confirm   00:56.441 --> 01:01.881 the measurement. Science sometimes works that  way—like when we first measured the existence   01:01.881 --> 01:07.881 of dark energy in the first place. But modern  cosmology has reached the point where this   01:07.881 --> 01:14.921 doesn’t work any more. The signals we’re looking  for are so subtle—so tangled together—that it’s   01:14.921 --> 01:20.281 no longer about directly measuring the  parameters of this universe—it’s about   01:20.281 --> 01:28.921 ruling out countless hypothetical universes that  we’re not in. And DES was ruthless in eliminating   01:28.921 --> 01:33.801 universes. So what did it discover? Despite their nearly identical names,   01:33.801 --> 01:39.001 the Dark Energy Survey, or DES, and the  Dark Energy Spectroscopic Instrument,   01:39.001 --> 01:44.921 DESI, were built to tackle the same grand  mystery from almost opposite directions.  01:44.921 --> 01:49.881 DESI's mission is beautifully focused. It  measures the distances to tens of millions   01:49.881 --> 01:55.321 of galaxies, constructing the most detailed  three-dimensional map of the Universe ever made.   01:55.321 --> 02:03.961 Hidden within this vast cosmic atlas is a faint  fossil pattern left behind by sound waves that   02:03.961 --> 02:09.241 rippled through the hot plasma of the infant  Universe nearly 14 billion years ago. These   02:09.241 --> 02:15.321 are the Baryon Acoustic Oscillations, which  we’ve covered before. BAOs gives us a cosmic   02:15.321 --> 02:21.481 ruler that grow with the expanding universe,  and so allows DESI to trace the history of   02:21.481 --> 02:27.561 that expansion with extraordinary precision. DES took a much broader approach. This truly   02:27.561 --> 02:33.401 international collaboration of some 400 scientists  took control of the Blanco 4-meter telescope at   02:33.401 --> 02:40.441 the Cerro Tololo Inter-American Observatory in  the Chilean Andes. After building a dedicated 570   02:40.441 --> 02:45.721 megapixel camera, they scanned an enormous  region of the southern sky for six years,   02:45.721 --> 02:53.081 building an extraordinarily deep portrait  and movie of 300 of million galaxies.   02:53.081 --> 02:59.161 This became the foundation for several  completely different ways of probing the cosmos.  02:59.161 --> 03:05.641 DES measures the tiny distortions in distant  galaxy shapes caused by the subtle warping of   03:05.641 --> 03:11.481 intervening space by massive objects. Through  this weak gravitational lensing, DES maps the   03:11.481 --> 03:18.681 invisible distribution of dark matter and tracks  how cosmic structure has grown over time. It also   03:18.681 --> 03:23.801 measures how galaxies have clustered together  over cosmic time due to their mutual gravity,   03:23.801 --> 03:29.321 giving a completely different perspective on  this same structure growth. By combining those   03:29.321 --> 03:36.361 two measurements through galaxy-galaxy lensing,  DES connects the visible galaxies to the much   03:36.361 --> 03:42.201 larger halos of dark matter that surround them. And DES also discovered and monitored thousands   03:42.201 --> 03:47.561 of Type Ia supernovae, which allowed it to  measure the expansion history of the Universe by   03:47.561 --> 03:53.161 determining the distances to these exploding white  dwarf stars. And, by the way, this is exactly   03:53.161 --> 04:00.761 how dark energy was first discovered.And finally DES even measures the same ancient baryon acoustic   04:00.761 --> 04:07.161 oscillation pattern as DESI—but with a completely  different observing strategy. So DES is really an   04:07.161 --> 04:13.481 entire laboratory of cosmological probes built  from one extraordinary survey of the sky. DESI,   04:13.481 --> 04:20.121 meanwhile, takes one of those probes and  pushes it farther than anyone ever has.  04:20.121 --> 04:25.081 Why try to measure dark energy by so many  different methods? It’s not because we’re   04:25.081 --> 04:30.841 worried that any one of those methods might be  wrong—at least, not mostly. It’s because each   04:30.841 --> 04:36.681 one tells us something slightly different about  the universe. And it’s only in the combination   04:36.681 --> 04:42.281 of all of our “cosmological probes” that we  find the answer that we were looking for.  04:42.281 --> 04:48.361 The truth is, there’s no one experiment  that actually measures dark energy. Instead,   04:48.361 --> 04:53.401 we observe galaxies and their distances to  each other and tiny distortions in their   04:53.401 --> 04:58.841 shapes and the faint flashes of exploding  stars. We observe the miniscule temperature   04:58.841 --> 05:03.241 fluctuations in the oldest light in the  Universe—the cosmic microwave background.   05:03.241 --> 05:08.841 None of these directly measure dark energy.  Instead, each measurement is influenced by   05:08.841 --> 05:14.601 a set of unknown cosmological parameters that  together determine the way the universe expands   05:14.601 --> 05:20.041 and how structure forms, which in turn  drive the observables that we can collect.  05:20.041 --> 05:25.241 It’s these cosmological parameters that we  really want—things like the proportions of   05:25.241 --> 05:30.601 dark energy to dark matter to regular matter  to light to neutrinos. Or The expansion rate   05:30.601 --> 05:37.641 at the beginning and the slowing and speed of  that rate due to gravity and dark energy. The   05:37.641 --> 05:43.161 parameters of the equation of state that  determine if and how dark energy changes   05:43.161 --> 05:49.401 over time. The subtler details of how light  and atoms and dark matter and dark energy   05:49.401 --> 05:56.121 interact with themselves and with each other. Each tiny variation in the configuration of   05:56.121 --> 06:04.601 these parameters represents an entire possible  universe. Cosmology is about figuring out which   06:04.601 --> 06:11.481 of these infinite possible universes we are  in. But we can’t do that just by measuring the   06:11.481 --> 06:18.921 cosmological parameters one by one—there is no  observation that will give us these parameters.   06:18.921 --> 06:26.281 Instead, we use our observations to determine  which hypothetical universes we are NOT in,   06:26.281 --> 06:33.881 over and over until we’re left with the universe  we are in, like a cosmic game of 20 questions.  06:33.881 --> 06:41.081 The cosmological parameters that are of most  interest to DES and DESI are the parameters of   06:41.081 --> 06:46.441 the dark energy equation of state. This is perhaps  the simplest way to describe a changing dark   06:46.441 --> 06:52.681 energy—we have the baseline dark energy strength,  omega-0, and the rate of change of dark energy,   06:52.681 --> 06:59.801 omega-a. Before we ever discovered dark energy,  as far as we knew we might have been any one   06:59.801 --> 07:06.121 of a vast distribution of possible universes in  the space defined by these parameters. With the   07:06.121 --> 07:12.681 first observations of white dwarf supernovae  we culled huge numbers of these universes 07:12.681 --> 07:16.841 to demonstrate that we’re in a universe with  at least some dark energy described by omega-0.  07:16.841 --> 07:25.241 The supernova studies had ask the universe a  question: they asked how has cosmic expansion   07:25.241 --> 07:30.681 dimmed distant supernovae? The universe  answers in a way that many possible   07:30.681 --> 07:35.801 universes couldn’t have answered. So we  were able to eliminate those universes. 07:35.801 --> 07:39.801 Then the cosmic microwave background  studies asked a different question:   07:39.801 --> 07:46.361 what did the early universe look like? The  answer rejects different universes. The   07:46.361 --> 07:51.641 baryon acoustic oscillation studies like DESI  ask: how did the fossil parameters from the   07:51.641 --> 07:57.881 early universe expand? Again, different  hypothetical universes are eliminated.   07:57.881 --> 08:03.481 DES asks some of the same with new data, and  also asks how dark matter has warped galaxy   08:03.481 --> 08:10.761 images and drawn galaxies together against that  expansion. Again, counterfactual realities were   08:10.761 --> 08:17.001 slain by DES in this case. In the end, we’re  left with a much smaller pocket of possible   08:17.001 --> 08:21.961 universes where the cosmological parameters  are now much more tightly constrained than they 08:21.961 --> 08:23.641 were before we began DES. 08:23.641 --> 08:29.641 Here’s an analogy. I’m thinking of two whole  numbers between one and ten and you can ask a   08:29.641 --> 08:34.601 series of questions to figure them out, but  you can’t ask what the actual numbers are.   08:34.601 --> 08:40.761 Let’s say you ask what the sum of the numbers  that you're thinking of. I tell you it’s 10.   08:40.761 --> 08:47.961 You immediately know that the numbers are 1  and 9 or 2 and 8 or 3 and 7 or 4 and 6 or 5   08:47.961 --> 08:53.241 and 5. That's five possibilities assuming  that order of the numbers doesn’t matter.   08:53.241 --> 08:58.761 Or you could instead ask what the product  is of the two numbers. I tell you it’s   08:58.761 --> 09:06.041 24—so that could be 4 and 6, 3 and 8—two  options. But only one of the answers to   09:06.041 --> 09:11.081 the second question is consistent with the  answer to the first question. My numbers   09:11.081 --> 09:18.121 had to be 4 and 6. Only by asking multiple  questions can we infer the original numbers. 09:18.121 --> 09:24.761 In this analogy, the numbers are our cosmological  parameters, and the function—the sum or product—is   09:24.761 --> 09:31.241 one aspect of the complex physics that turns  those numbers into observables. For example,   09:31.241 --> 09:36.361 one “function” might be “how do the cosmological  parameters govern dark matter halo growth and so   09:36.361 --> 09:42.761 warp the shapes of distant galaxies through weak  lensing” or “how do they determine baryon acoustic   09:42.761 --> 09:47.961 oscillation formation and then expansion”. Unfortunately, we don’t get to choose these   09:47.961 --> 09:52.841 “functions” or questions. They’re what  the universe offers us. When we come up   09:52.841 --> 09:58.841 with a cosmological probe, we’re identifying  observables that best pry the cosmological   09:58.841 --> 10:05.481 parameters in the context of all the other  available probes. What we’re trying to do   10:05.481 --> 10:11.721 is to minimize what we call degeneracies.  In the context of our cosmological probes,   10:11.721 --> 10:16.841 degeneracies are the multiple plausible  cosmological parameters that could have   10:16.841 --> 10:23.481 led to the same observations. Like how there  are several combinations of whole numbers that   10:23.481 --> 10:31.241 sum to 10. Sometimes a probe is good at  constraining one parameter over others,   10:31.241 --> 10:37.561 sometimes a probe contains a combination of  parameters without giving any single constraint.   10:37.561 --> 10:44.681 We want a combination of probes that whittle  down the degeneracies and leave us with tight   10:44.681 --> 10:50.281 constraints on the parameters we’re interested in. All we need to do is combine the datasets from   10:50.281 --> 10:59.321 these different probes. These various  cosmological probes are not independent   10:59.321 --> 11:03.641 of each other and so we can’t combine the  constraints they give in a simple way.   11:04.761 --> 11:10.521 These different probes may observe many of the  same galaxies, or rely on the same potentially   11:10.521 --> 11:16.921 biased distance calibrations, or they make similar  assumptions about how galaxies form and evolve.   11:16.921 --> 11:23.481 That means some of their uncertainties are also  shared. In statistics, those shared uncertainties   11:23.481 --> 11:31.081 are called covariance. Ignoring covariance is  a bit like polling a hundred people who all got   11:31.081 --> 11:36.601 their news from the same rumor—you think you have  a hundred independent opinions, but really you've   11:36.601 --> 11:43.321 only measured the same bias a hundred times. So  before cosmologists compare two experiments, they   11:43.321 --> 11:48.841 have to figure out exactly how much information  is genuinely new, and how much is information   11:48.841 --> 11:55.161 they've effectively counted already. It turns out  that understanding those relationships between   11:55.161 --> 12:00.841 datasets is just as important as understanding  the datasets themselves. In modern cosmology,   12:00.841 --> 12:05.481 the challenge isn't simply collecting more  evidence—it's making sure every new piece of   12:05.481 --> 12:13.961 evidence is genuinely asking a different question. But in the end, reality is self-consistent. There   12:13.961 --> 12:20.281 aren't separate Universes for DESI and DES, or for  supernovae and the CMB. There’s just one universe,   12:20.281 --> 12:25.641 and it will always give consistent answers  to the questions we ask of it. We just need   12:25.641 --> 12:31.721 to choose the right questions that allow us to  efficiently rule out the universes that could   12:31.721 --> 12:39.081 not have provided the answers that we receive. So, what happened when DES finally asked its   12:39.081 --> 12:45.561 questions of the universe? Analyzed on its own,  the new DES results are beautifully consistent   12:45.561 --> 12:51.081 with the standard picture of cosmology.  A Universe dominated by a cosmological   12:51.081 --> 12:57.401 constant still provides a perfectly good  description of everything DES observed. It   12:57.401 --> 13:04.441 very slightly favors a changing dark energy.  DESI is similar—consisstent with the standard   13:04.441 --> 13:11.401 cosmology—there are surviving universes whose  physics is exactly what we thought. But there   13:11.401 --> 13:17.561 are slightly more surviving potential universes  whose dark energy is actually diminishing.  13:17.561 --> 13:24.361 Remember, DES wasn't built to verify DESI or vice  versa. Both were built to interrogate the Universe   13:24.361 --> 13:30.361 in different, complementary ways. That means  that together they have the power to eliminate   13:30.361 --> 13:38.681 more possible universes than either in isolation. And when the DES results are compared with DESI,   13:38.681 --> 13:44.601 and the cosmic microwave background, and supernova  observations, many of the same Universes do   13:44.601 --> 13:52.121 survive, but not all. Further universe-culling  pushes us slightly further towards favoring a   13:52.121 --> 13:59.561 decreasing dark energy. Not enough to rule out  physics-as-we-thought-it, but the intrigue does 14:00.201 --> 14:07.241 increase. DES independently points to the same  exciting region of parameters space as DESI,   14:07.241 --> 14:14.521 which is extremely important to rule out  bias, while also shaving the edges because   14:14.521 --> 14:21.881 DES has different degeneracies to DESI. The  several different questions asked by DES and   14:21.881 --> 14:28.601 DESI were answered consistently by a smaller  and smaller fraction of possible universes.  14:29.161 --> 14:35.001 Perhaps the most exciting “science result”  from DES is that the intriguing hints from   14:35.001 --> 14:42.201 DESI weren’t immediately ruled out or disfavored.  That could’ve happened, and would have meant that   14:42.201 --> 14:48.121 some bias in the DESI method was probably at  work. No, that island of intriguing universes   14:48.121 --> 14:55.161 with decreasing dark energy is still in play. So what’s next? We ask more questions of   14:55.161 --> 15:01.481 the universe. We give those counterfactual  universes more chances to fail our grilling   15:01.481 --> 15:07.481 and get eliminated. That's why the next decade  of cosmology is going to be incredible. The   15:07.481 --> 15:12.201 Rubin Observatory will discover billions  of new galaxies and measure the changing   15:12.201 --> 15:18.761 Universe with unprecedented depth and monitor  it over a decade. Euclid is already in space,   15:18.761 --> 15:24.201 mapping a vast patch of the sky with incredible  spatial resolution. The Roman Space Telescope   15:24.201 --> 15:30.121 will perform this exquisitely sharp imaging  over a smaller scale, but with even better   15:30.121 --> 15:35.641 calibration and will also monitor over  time. DESI will continue expanding the   15:35.641 --> 15:41.961 largest three-dimensional map of the Universe ever  assembled. And we'll continue culling universes.  15:41.961 --> 15:48.601 Modern cosmology is really a beautiful example  of how humans have always learned about nature.   15:48.601 --> 15:54.121 We know in our hearts that there’s a world out  there and that the world is self-consistent. And   15:54.121 --> 16:01.641 when we ask questions of it—with our senses and  with our experiments—it’ll tell us the truth. We   16:01.641 --> 16:08.281 believe in the model of the world that never fails  a test—never answers wrong. Every independent line   16:08.281 --> 16:15.881 of evidence needs to agree on the same story. And even when we think we have the story, we   16:15.881 --> 16:21.641 keep asking questions. At the frontier of science,  the hardest part isn't discovering something new,   16:21.641 --> 16:26.761 it's convincing ourselves that we haven't fooled  ourselves. Our current cosmological models,   16:26.761 --> 16:31.641 along with all of our best scientific theories  in their broad strokes—from evolution to climate   16:31.641 --> 16:37.481 change to germ theory to the Big Bang—are  the explanations of a reality that have   16:37.481 --> 16:44.121 withstood every question that we've so far  asked of it. But these also still represent   16:44.121 --> 16:50.281 families of possible universes that still need  whittling down. So we keep asking questions,   16:50.281 --> 17:03.734 bringing us closer to pinpointing in  parameter space this singular space time.