WEBVTT 00:00.000 --> 00:02.640 and you can learn more about  them at the end of our episode. 00:02.640 --> 00:08.080 The James Webb Space Telescope found galaxies  that are too ancient-looking for our young 00:08.080 --> 00:15.520 universe. Now you may have heard that, but  JWST keeps finding them, and our recent 00:15.520 --> 00:21.760 efforts to solve this conundrum point in wildly  different directions. Have we found galaxies 00:21.760 --> 00:34.160 older than the universe, or did we just learn  something incredible about how galaxies form? 00:21.760 --> 00:28.240 with special deals if you combine your order with  previous May the 4th merch. Now on to the episode. 00:27.542 --> 00:32.902 Telescopes are time machines. Light takes time to  get to us, so we see distant objects as they were 00:32.902 --> 00:38.582 when their light began its Earthward journey.  As our telescopes become more powerful they 00:38.582 --> 00:44.502 see further, and so more and more of the past  becomes accessible to us. The James Webb Space 00:44.502 --> 00:49.702 Telescope is one of the most powerful time  machines ever built. It’s powerful enough 00:49.702 --> 00:55.222 that it’s discovered galaxies whose light has  been traveling to us since the universe was 00:55.222 --> 01:00.822 a little over 2% of its current age. The  universe back then should look different, 01:00.822 --> 01:07.062 right? Those galaxies should look different. After  all, this is when galaxies first started to grow, 01:07.062 --> 01:11.142 when they should have been vigorously  forming many of their stars. They 01:11.142 --> 01:17.942 should look like hyperactive kids. Those  types of galaxies are around back then, 01:17.942 --> 01:26.182 yet JWST has also seen much more developed,  “adult” galaxies. Some that look way too big, 01:26.182 --> 01:31.462 and way too ancient-looking for a universe  only a few hundred million years old. 01:31.462 --> 01:38.102 Over the past year or two, this mystery has made  the pop-sci rounds. That included some breathless 01:38.102 --> 01:43.462 speculation—like the idea that the entire  Big Bang model is wrong. If there are ancient 01:43.462 --> 01:50.182 galaxies 13 billion years ago, then how can the  universe be only 13 and a half billion years old? 01:50.182 --> 01:56.182 So far we haven’t weighed in on this conundrum.  Now many others have done a fantastic job laying 01:56.182 --> 02:01.222 it out and debunking some of the foolishness. Dr  Becky in particular was on top of the progress 02:01.222 --> 02:08.102 as new data came in from JWST. So, why should we  cover it now? First, a lot of you have asked for 02:08.102 --> 02:14.822 our take and it’s time to go on record. Second,  the evolution of this mystery has been quite an 02:14.822 --> 02:21.302 emotional rollercoaster, swinging from unsolvable  to solved, and now there’s new work that swings 02:21.302 --> 02:29.142 back in the direction of WTF. So let’s talk about  the current status of the early galaxy conundrum. 02:29.142 --> 02:33.622 And before that, let’s talk a  bit more about what we actually 02:33.622 --> 02:37.382 expect the early universe galaxies to look like. 02:38.022 --> 02:43.382 Galaxies pulled themselves together from tiny  density fluctuations in the very early universe 02:43.382 --> 02:50.102 that we see in the dappling of the cosmic  microwave background. The CMB reveals regions 02:50.102 --> 02:56.102 with tiny excesses of matter—hydrogen gas, but  more importantly dark matter which outweighs the 02:56.102 --> 03:02.102 gas by a factor of at least five. As dark matter  pulled itself together by gravity it pulled the 03:02.102 --> 03:08.182 gas in with it. And as that gas compacted,  the first stars were also born. These very 03:08.182 --> 03:14.502 early galaxies must have started small, but with  pretty crazy star formation due to the enormous 03:14.502 --> 03:20.902 abundance of gas at that time. As those galaxies  continued to grow, they collided and merged, 03:20.902 --> 03:25.782 and eventually built themselves up into  the mature galaxies that we see today. 03:25.782 --> 03:31.062 We think we understand this stuff pretty well  - or at least thought we did. Between the 03:31.062 --> 03:35.862 precise CMB measurements and our theoretical  understanding of gravity, star formation, 03:35.862 --> 03:40.022 etc., our computer simulations allow us  to explore the possible growth scenarios 03:40.022 --> 03:45.942 for galaxies. We can predict, for example,  that early galaxies should be forming stars 03:45.942 --> 03:51.542 like crazy due to the enormous amounts of raw  material—-hydrogen gas—that was around back then. 03:51.542 --> 03:58.662 One thing that should be even more robust as  a prediction is the size of dark matter halos. 03:58.662 --> 04:04.422 These are the giant pools of dark matter that  encompass all galaxies and hold them together. 04:04.422 --> 04:08.902 Because dark matter should not be  strongly effected by the complicated 04:08.902 --> 04:14.182 behavior of the gas and stars, we have  high confidence in our understanding of 04:14.182 --> 04:18.822 how those halos grew over time.  Or at least we thought we knew. 04:18.822 --> 04:24.102 One very clear prediction of this whole model  is that a lot of this halo growth should have 04:24.102 --> 04:30.822 happened in the first 10% of the universe’s age.  In that first 1.5 billion years since the big bang 04:30.822 --> 04:37.862 there shouldn’t have been essentially no very  large halos, and so no very large galaxies. 04:37.862 --> 04:43.142 And that’s what the theory and the simulations  say. But to check these we need powerful time 04:43.142 --> 04:48.822 machines— telescopes. Now as the telescopes got  bigger and our cameras got more sensitive we 04:48.822 --> 04:55.302 were finally able to probe early enough times to  properly test our models of halo growth. And this 04:55.302 --> 05:02.342 is where the problem started around 15-16 years  ago when our “high redshift galaxy surveys” - aka 05:02.342 --> 05:08.582 really far away galaxy surveys - finally  reached these distances we got some answers. 05:08.582 --> 05:14.742 And those surveys started seeing a few  cases of what looked like giant halos at 05:14.742 --> 05:21.862 earlier and earlier times. And also the hint  of overly red galaxies in the early universe. 05:21.862 --> 05:27.382 Galaxies that are actively forming  new stars should be bright at short 05:27.382 --> 05:32.182 wavelengths because they have lots of  giant, hot, short-lived stars. They 05:32.182 --> 05:37.782 are relatively blue in color compared  to older galaxies. And older galaxies 05:37.782 --> 05:42.982 that are no longer so actively forming  stars look much redder because these 05:42.982 --> 05:45.862 short-lived blue stars exploded already. 05:45.862 --> 05:51.462 Those early surveys didn't reveal anything crazy  yet—just a handful of cases where the galaxies 05:51.462 --> 05:57.702 looked too big and/or too evolved compared to  what was expected from our models. In 2018, 05:57.702 --> 06:03.862 Charles Steinhardt and team articulated this  emerging tension in their paper “the impossibly 06:03.862 --> 06:09.942 early galaxy problem”. But impossible  things are well, impossible. It’s right 06:09.942 --> 06:17.382 there in the name. So several not-impossible  explanations for these galaxies were devised. 06:17.382 --> 06:24.502 For one, we don’t see these overly-large dark  matter halos directly. We only see the starlight, 06:24.502 --> 06:28.742 and we use that starlight to infer the  mass of the stars and then the mass of 06:28.742 --> 06:34.262 halos that contain them. But that step requires  assumptions, and if any of those assumptions 06:34.262 --> 06:41.462 are wrong then perhaps we got the wrong halo  mass. And the redness that suggests an old 06:41.462 --> 06:46.022 stellar population could be due to other  things too, and there are several other 06:46.022 --> 06:51.942 issues with these inferences besides. I’ll  come back to possible solutions in a bit. 06:51.942 --> 06:57.942 But even if the galaxies observed in  these ground-based surveys were really 06:57.942 --> 07:02.902 too old looking and too big looking,  well we haven’t quite broken all of 07:02.902 --> 07:07.542 astrophysics and cosmology yet. If we  strain our models of galaxy formation, 07:07.542 --> 07:13.702 we can potentially speed up galaxy  evolution to show why these things exist 07:13.702 --> 07:19.382 by the time the universe is 10% of its  current age. What we’d really like to do 07:19.382 --> 07:25.222 is to look back even further in time to see  if we can find the time when these galaxies 07:25.222 --> 07:31.302 were themselves growing. And that’s what  we did with the James Webb Space Telescope. 07:31.302 --> 07:38.662 JWST was designed to push this early-galaxies  game to new extreme limits. It’s the largest 07:38.662 --> 07:44.582 telescope ever deployed to space, which means the  most powerful. It’s also designed to be sensitive 07:44.582 --> 07:49.702 to very long wavelengths of light—deep into the  infrared part of the electromagnetic spectrum. 07:49.702 --> 07:55.382 The “mid-infrared” as we like to call it.  That’s important for these first galaxies 07:55.382 --> 08:00.742 because of the expansion of the universe. Their  light has been traveling to us for much of the 08:00.742 --> 08:07.462 age of the universe. Because these EM waves  traveled through expanding space, they were 08:07.462 --> 08:13.382 stretched out—their wavelengths lengthened. This  is cosmological redshift, and higher redshift 08:13.382 --> 08:19.942 means longer travel time and greater distance. For  the galaxies we’re interested in, this redshift 08:19.942 --> 08:27.062 converts visible and even ultraviolet light  into mid-infrared—which is why JWST is needed. 08:27.062 --> 08:34.262 The first efforts with JWST used similar methods  to our ground-based galaxy surveys. So maybe I 08:34.262 --> 08:38.982 give you a bit more detail on that. The first  thing you do in a galaxy survey is to take 08:38.982 --> 08:45.302 pics with different filters corresponding to  different wavelength bands and then you compare 08:45.302 --> 08:51.782 the amount of light in each filter—we call this  photometric imaging. We can learn a lot of stuff 08:51.782 --> 08:58.022 from the ratios of filter brightnesses— and so the  colors. We discover candidate galaxies this way, 08:58.022 --> 09:02.822 we make a crude estimate of the distance  because the color ratios suggest a redshift, 09:02.822 --> 09:08.182 we can constrain the stellar population—both  the number of stars and the distribution of 09:08.182 --> 09:14.502 different types—for example, the redder colors  mean an older population with few massive stars. 09:14.502 --> 09:19.462 So what did JWST find? Well first, it  confirmed the presence of overly massive 09:19.462 --> 09:24.982 and overly old-looking galaxies from the  earlier studies—and remember that was from 09:24.982 --> 09:31.222 a cosmic age of around 10%--a redshift  of 4. Now a critical advantage of JWST 09:31.222 --> 09:36.422 is that it’s sensitive enough to do proper  spectroscopy on these galaxies—it can measure 09:36.422 --> 09:41.542 their “spectrum”-- the amount of light as  a function of wavelength. These spectra 09:41.542 --> 09:47.382 confirmed that the redshifts are indeed very  high. They also confirmed that the redness of 09:47.382 --> 09:53.142 the spectra is due to highly evolved stellar  populations, rather than, say, there being a 09:53.142 --> 09:59.622 lot of dust in the galaxies, which can cause  similar reddening in photometric analyses. 09:59.622 --> 10:04.822 With these confirmations of old-looking early  galaxies, researchers more broadly really 10:04.822 --> 10:11.462 started to pay attention. The media started  to pick up on it when JWST pushed to greater 10:11.462 --> 10:18.102 distances and earlier times and kept finding these  things. Currently, the earliest candidate giant, 10:18.102 --> 10:25.782 evolved galaxy discovered by JWST is at  redshift 7.3, at just 5% of the universe’s age. 10:25.782 --> 10:30.822 Now it seems we have a real conflict  with our models of galaxy formation. 10:30.822 --> 10:36.342 Dark matter halos too large and stellar  populations too evolved for the short amount 10:36.342 --> 10:43.142 of time they had to develop. And this is about why  we started to hear some hysterical claims that the 10:43.142 --> 10:48.262 universe is twice as old as we thought, or that  the big bang model is completely overturned. 10:48.262 --> 10:55.302 It’s not. And it isn't. There’s just so much  independent corroboration of our model of an 10:55.302 --> 11:00.262 expanding 13 point something billion  year old universe. You can’t point to 11:00.262 --> 11:05.702 one admittedly intriguing discrepancy and  decide to throw the baby universe out with 11:05.702 --> 11:13.702 the bath water. There are much more parsimonious  explanations for our improbably early galaxies. 11:13.702 --> 11:16.662 Let’s look at one of the most compelling. Remember 11:16.662 --> 11:20.822 that when we calculate the masses of  these supposedly gigantic dark matter 11:20.822 --> 11:26.742 halos, we based them on the starlight that  we see, and  that requires an understanding 11:26.742 --> 11:31.542 of the relationship between these two things,  the starlight and the dark matter halo mass. 11:31.542 --> 11:33.942 And that involves assumptions. Here’s how that 11:33.942 --> 11:37.702 works. One assumption is that halo  mass is connected to the mass in 11:37.702 --> 11:44.342 stars,  and that “stellar mass” is connected  to the amount of light we see in those stars. 11:44.342 --> 11:51.062 But we don’t see the light from all of the  stars—typically the light we collect is dominated 11:51.062 --> 11:56.742 by the brightest stars in the galaxy. We then have  to decide on the relative numbers of the different 11:56.742 --> 12:01.862 types of stars so we can extrapolate from the  observer starlight to the mass of all stars, 12:01.862 --> 12:09.542 seen and unseen. And there's another assumption.  And, yes, from there we can get to the halo mass. 12:09.542 --> 12:14.982 Knowing the distribution of stellar masses  in a galaxy on the other side of the universe 12:14.982 --> 12:21.062 takes some guesswork to say the least. And  maybe the biggest unknown there is something 12:21.062 --> 12:25.862 called the initial mass function—the  IMF. It tells us the relative numbers 12:25.862 --> 12:32.022 of stars at different masses that form  when a burst of star formation happens. 12:32.022 --> 12:37.542 We typically use the IMF that has been measured  for the Milky Way galaxy, sometimes with various 12:37.542 --> 12:42.822 refinements. But we don’t know that stars  formed with similar mass distributions in 12:42.822 --> 12:47.782 the early universe. And they probably didn’t.  For example, when there’s less heavy elements 12:47.782 --> 12:54.102 around from generations of old stars, its easier  to make really gigantic stars because gas clouds 12:54.102 --> 13:01.942 don’t fragment as much when the collapse. That  would give what we call a “top-heavy” IMF— so 13:01.942 --> 13:08.342 more massive stars form in a given burst of star  formation relative to what happens in the Milky 13:08.342 --> 13:16.342 Way. And more bright, massive stars means that  these galaxies would be overly bright for a given 13:16.342 --> 13:22.582 halo mass. So if we’re determining halo mass  from the light of those massive stars then we 13:22.582 --> 13:29.222 over estimate halo mass. Such a “top-heavy” IMF  is probably the leading contender for explaining 13:29.222 --> 13:34.822 the apparent giantness of the dark matter  halos. And this may even solve the conundrum. 13:34.822 --> 13:40.422 But the mystery does stop here. A new  study just found exactly the opposite 13:40.422 --> 13:45.942 result of this. This study claims to have  identified a sample of galaxies that are 13:45.942 --> 13:51.702 what those “impossibly early” galaxies  became in this part of the universe. 13:51.702 --> 13:56.742 And because this sample is much closer,  we can detect the light of much fainter, 13:56.742 --> 14:02.742 lower-mass stars. And so figure out the initial  mass function down to much lower masses. 14:02.742 --> 14:09.142 And the result is bad. This study found  that the IMF is actually bottom-heavy in 14:09.142 --> 14:15.062 those galaxies. There are way more low mass  stars—little red dwarfs and whatnot—compared 14:15.062 --> 14:20.422 to  the Milky Way for example. And  definitely very bottom-heavy compared 14:20.422 --> 14:26.502 to what we think a top-heavy IMF might look  like. The top-heavy IMF that was supposed 14:26.502 --> 14:33.462 to solve the problem of the impossible galaxies.  And in fact, this bottom-heavy IMFworsens the 14:33.462 --> 14:38.582 problem. An excess of low-mass stars means  that when we convert galaxy light to stellar 14:38.582 --> 14:44.982 mass we underestimate that stellar mass, and  presumably underestimate it's dark matter halo 14:44.982 --> 14:53.062 mass. And that makes it even harder to figure  out how those things grew so fast so quick. 14:53.062 --> 14:58.662 OK, before we join our crazy uncle on the anti-Big  Bang conspiracy facebook group, some words of 14:58.662 --> 15:05.062 caution. We don’t really know whether this new  study really did successfully identify the modern 15:05.062 --> 15:09.942 counterparts of our impossible galaxies, and  the authors admit that, calling them “likely 15:09.942 --> 15:15.222 descendents”. We also don’t know what complicating  interactions these galaxies could have had in the 15:15.222 --> 15:20.742 intervening 13 billion years to muddle the  IMF. We also don’t know that a bottom-heavy 15:20.742 --> 15:27.702 IMF rules out an IMF that’s also top-heavy. In  other words, there could be an over-abundance 15:27.702 --> 15:35.462 of low mass stars AND of high mass stars all  compared to masses around that of the Sun. And 15:35.462 --> 15:43.622 it is plausible to assume an IMF like this that  would still bring down those early halo masses. 15:43.622 --> 15:49.302 There are similar challenges in explaining  the apparent redness of these early galaxies. 15:49.302 --> 15:54.182 The key here is that we need a way  to shut star formation down much 15:54.182 --> 15:59.862 more quickly than we thought likely. Perhaps  the leading contender is quite awesome—early 15:59.862 --> 16:04.822 quasars—supermassive black holes in the  hearts of these galaxies blasting out 16:04.822 --> 16:09.462 radiation and winds that heats up and  expels gas so that stars stop forming 16:09.462 --> 16:15.782 and the population can evolve quickly. We  know this sort of feedback from quasars 16:15.782 --> 16:23.542 happens, but we now need to understand why its  so extreme in the early universe. That involves 16:23.542 --> 16:30.422 very rapid growth of supermassive black holes—yet  another very real and very interesting problem. 16:30.422 --> 16:34.262 The most likely solution to all of this  is that we’re going to learn an enormous 16:34.262 --> 16:40.342 amount about the surprising processes  behind structure growth, star formation, 16:40.342 --> 16:45.782 black hole seeding and growth, and who  knows what else. And once we figure it out, 16:45.782 --> 16:51.862 the impossible early galaxies will become more  than just possible—they’ll become inevitable: 16:51.862 --> 16:56.662 a natural part of our updated  understanding of our early spacetime. 16:51.862 --> 16:58.182 when you drag your tabs downward, Opera  splits the screen up to four, resizable ways. 16:51.862 --> 17:00.582 If you’d like to try out all the features of the  Opera browser there’s a link in the description.