SGU Episode 1006

This episode needs: proofreading, links, 'Today I Learned' list, categories, segment redirects. Please help out by contributing!

How to Contribute

SGU Episode 1006
October 19th 2024
"An early morning launch lights up the sky, inspiring dreams of space exploration."
SGU 1005                      SGU 1007
Skeptical Rogues
S: Steven Novella
B: Bob Novella
C: Cara Santa Maria
J: Jay Novella
E: Evan Bernstein
Quote of the Week
"One of the beautiful things about science is that it allows us to bumble along, getting it wrong time after time, and feel perfectly fine as long as we learn something each time."
- Martin A. Schwartz
Links
Download Podcast
Show Notes
SGU Forum

Intro

Voice-over: You're listening to the Skeptics' Guide to the Universe, your escape to reality.

S: Hello and welcome to the Skeptics' Guide to the Universe. Today is Wednesday, October 16^th, 2024, and this is your host, Steven Novella. Joining me this week are Bob Novella...

B: Hey, everybody!

S: Cara Santa Maria...

C: Howdy.

S: Jay Novella...

J: Hey guys.

S: ...and Evan Bernstein.

E: Good evening everyone.

S: Hey, so has anyone seen this once in 80,000 year comment recently?

E: No. Photos of it?

J: No, I have not seen it.

E: Does a photo count?

S: No, I mean in person.

C: No.

S: No, it doesn't count. I tried to see it twice so far. I have not been able to spot it. I just think there's too much light pollution where I am, to be honest with you.

E: And what time of day or evening?

S: Just after sunset, if you look to the west, just after sunset, but it's sort of climbing in the sky every day and also getting farther away. The closest approach was Saturday before this recording. By the time the show goes out, it's be at the very tail end. This is the northern hemisphere. But if you look at Arcturus, which you guys all know how to find Arcturus in the sky, right?

E: Oh, sure.

S: You follow the arc of the Big Dipper. That's kind of the mnemonic. And then it's the brightest star that that kind of arc goes to. And then you look to the upper left of that. That's where the comet is. Take binoculars. But I could not find it even with my binoculars. I think there's too much light pollution. I was disappointed. But this comet comes around once every 80,000 years. So of course all the news outlets are saying, the last time it was here the Neanderthals saw it. Of course there were humans at the time as well. 80,000, yeah. The Neanderthals died at 35,000.

E: Was it 35?

B: How about the Inuit Sea, the Aurore?

S: Yeah, that was the other thing.

E: Oh gosh, there were so many pictures of that.

S: I did not see that either. I went out that at night to see that as well.

E: That was last week.

S: Yeah, that was last week.

C: It's still going on. I mean, I think we're, we're at solar maximum, right? Like, we're in it. We're like fully in it.

B: Well, I was at the hunt and everyone's looking up and like, what the hell? And there was a, you could see a red glow. It was faint, but there was definitely...

C: Did you take a photo?

B: I took a photo and of course the picture was much better than what I was seeing with my eyes. Not ridiculously better, because I've seen some pictures people took that were just like, holy crap. Mine wasn't a holy crap picture, but it was like, oh, cool picture. Better than what I saw. So I was happy that I saw it, but I want to see the shimmering ribbons of color. This is just like a diffuse red glow, which is fine. I was happy, because you never see that in Connecticut. But man, I want to see the real deal though. Oh well. I still saw it.

S: Had to go to Scandinavia or Iceland or something.

B: But I would guess that even southern Canada is seeing some good aurora.

S: I think you've just got to get to a dark sky location. I'm just too close to New Haven. There's so much glow around the horizon if you're looking for anything especially low. I remember when we were kids, the summer nights, we would go out and you would see the Milky Way and it was gorgeous. I just haven't seen that in decades.

B: Oh, really? You haven't.

S: I mean, except when I'm deliberately going to a dark sky location, but I mean like around my house.

B: We used to see it growing up.

S: I mean, maybe you could kind of sort of see it, but it's not like it was when we were kids.

B: It's worse, yeah, no, and it's worse where you are.

S: Oh, yeah.

B: It's so close to New Haven, such a big city.

C: I told you guys I'm going to Iceland in January, right?

'S: Yes.

C: So hopefully I'll have good news to report back on the Aurora front.

B: Oh boy, yeah.

C: Fingers crossed.

S: Unfortunately, we were looking into Iceland because the next eclipse is going to go through there. And it's like 70% overcast. That's the downside of being here.

E: Yeah, unfortunately for that. Spain, heading to Spain.

C: Well, I mean, but were you looking at the entire country? Or were you just looking at where the eclipse would be? Because that's right on the coast.

E: Isn't that the point? Yeah.

S: But I still think the whole thing is overcast frequently, depending on what season you are. So definitely when I looked at the season when the eclipse is going to happen, which is why it's August or something?

E: Yes, it'll be August of 2026.

C: I mean, we're going to Iceland in January because it's you want to go for the eclipse when it's dark and cold.

E: You ain't going for the beach, that's for sure.

S: Well, we have a great interview coming up with Christian Hubicki later in the show. We're going to talk all about robotics. So let's get right to it.

From TikTok: Fake Space Station (04:42)

https://www.tiktok.com/@james.1.1.1.1.1.1.1/video/7415844279718776097?_r=1&_t=8qa2MthYQIQ

S: I'm going to start us off with a quick from TikTok. This is a quickie. So in this video, the guy who made the video was called NASA Lies. That's the name of the contact letter. NASA Lies. It's a video of the Chinese space station and taikonauts aboard that space station. And they're trying to say that it's fake because they think they found an anomaly. It's like, well, you know that NASA has nothing to do with that, right? That's a completely different country, different space organization. That seemed to be lost on him. And all the commenters were like, they lie about everything. Which they are we talking about, Matt?

E: They.

S: Yeah, they. So this was the anomaly they found. There's basically a cup of water on the table. The cup is velcroed to the table, but the water is sitting at the bottom of the cup. And apparently everyone on the comments was convinced, or most of the people there, That the water should be floating away because it's in microgravity. But of course, that's not true. That's not how water works in zero-g. There's so many videos online that you could find that demonstrate this. So you guys know how water behaves like on the ISS?

J: Absolutely.

B: Well, it's ruled by surface tension, right?

S: Yeah, exactly. Water sticks to itself, and it sticks to surfaces, some surfaces more than others, especially glass. It likes to stick to glass and stick to itself, because you have weak interactions between the hydrogen of one water molecule and the oxygen of another, and one's positive, one's negative, so they're slightly attracted to each other. Now, of course, in 1G, gravity overwhelms that weak force, except for a little bit of surface tension, so like, yeah, mosquitoes could walk on top of it, but water's still going to seek its lowest level. In microgravity, it's just going to if you did sort of release it so that it was free-floating in the air, it would just form a sphere, right, just the surface tension. You've seen the basic little spherical globules. I remember there was one video of one of the astronauts squeezing out a rag of water, and the water just glommed around his hands and didn't go anywhere. So again, it didn't float off into the air. If it has a surface to cling to, it would cling to it. So the thing is there's other videos of these same people and that same jar of water on the station where they show them putting the water into it and interacting with it, showing how it will adhere to a pencil if you put that in there. You could put a ping pong ball in there and it won't float because there's no buoyancy without gravity. So clearly proving that they're in microgravity, plus all the other reasons, right, the people's hair floating and all that sort of stuff. But they look at just 10 seconds of the video, they zoom in on one thing, they completely misinterpret it because they don't understand the physics, and they just spin conspiracy theories about it. Total nonsense. Meanwhile, there's some pretty cool science there, if they actually were curious.

News Item #1 - AI Common Sense (07:56)

• Can AI have common sense? Finding out will be key to achieving machine intelligence 

S: All right, Jay, tell us about artificial intelligence and common sense.

J: Yeah, so since the release of the large language models, right, we call these LLMs, like the ones behind ChatGPT. So they've shown there's a good amount of power there. There's a lot of people using them. There's a lot of excitement about it. But there are limitations, of course, to current machine intelligence, and I think as we all get accustomed to these LLMs that we're using, that we're starting to see, yeah like I use Suno.com, it can make music, and sometimes you can really knock something out of the park, but for the most part, it's a slug trying to go through a hundred iterations until you find something that's reasonably good. Currently, some researchers are speculating that LLMs might be the step towards artificial general intelligence. This is AGI. What we're basically saying here is that we can create machines that can handle a range of tasks similar to human intelligence. Humans can operate in the real world. We have common sense. We have understanding of the world's physics and lots of other things, social interactions and everything. Right now, LLMs, they don't think. They just regurgitate information and they construct information based on statistics. So this is something AI researchers, they've been chasing this artificial general intelligence idea for 70 plus years. Now historically, here at the SGU, we've reported that narrow AIs like ChatGPT, these are not the beginnings of general AI, right? Because the knowledge up until very recently has basically been it's not going to be where general AI comes from. But this could be changing with some new ways of thinking about how to progress current LLM's ability to understand common sense knowledge. So I'll give you a few ideas here. First of all, an important milestone that needs to be achieved is this machine learning and the ability to deal with the idea of common sense. As humans, it's typical that we gain common sense as we age, right? Kids, they learn by trying, that's why kids do things that sometimes might seem crazy to an adult, like touching something really hot or whatever, but this is the way that we learn. We also learn about people, we learn about everyday life, and again, like I said, social interactions and all that stuff. So for instance, we know that glass is fragile, right? Even my eight-year-old knows that glass inherently is fragile, and if you drop it, it's going to break. Or it's rude to serve meat to a vegan friend of yours, right? You wouldn't do that. Why? Well, it's obvious to us why, because they don't eat meat in some of them could be very offended by the idea of meat. We get this and we accept it. When someone lacks common sense though they make mistakes that most people wouldn't similar to an LLM because the LLM doesn't have any of this common sense they have strengths but they have massive weaknesses. So LLMs excel at what? Essentially tasks that involve memorization, tasks that involve collecting data and disseminating that information. So for example, ChatGPT-4, it can pass licensing exams for doctors and lawyers, right? I'm sure you guys have read the news items that come out where they say, I could do it better now than 80% of lawyers that took the bar and all that stuff.

C: Geez, will it take by my licensing exam for me?

J: No, Cara, you're safe because you have to actually talk to people and there's a lot of...

C: Not in my licensing exam.

J: But it can do it. Why can it do it? It's a huge amount of information. Well, information to large language models, they got that. That's what they do, and they're really good at disseminating that information. They can scan through it, they can find answers, and they can churn out the answers. But if you give a large language model a puzzle, a simple puzzle, and this is where we can see things go wrong. Like if you ask, if you say, hey, ChatGPT, my friend Frank is in pain. How would Frank feel right after the thing that hurt them that caused the pain? How would they feel? It could and it's not uncommon for it to choose a nonsensical response like Frank would would have more awareness now than he than he did before. The correct answer is Frank would feel pain. But it just doesn't know that. And these are things, of course, to the human brain, they seem so obvious to us, but again, these are not thinking machines. These are machines that are going through processes to come to their answers in a very different way than humans do. So multiple choice questions are widely used to measure machines' common sense, but these tests typically miss the mark because they don't reflect the real world. They don't reflect what people like us interact with and what the intuitive grasp that we have on physical laws and all the social norms. So even though LLMs perform well on paper, and they do, they fail to display the flexibility of human reasoning that we all have. Research on machine common sense began back a long time ago, back in 1956, with an AI workshop that set the vision like create machines that learn from experience just like humans. So easy to say and possibly hard to make happen. So these early approaches that they had, they used symbolic language, programming machines with if-this-then-that type logic gates and rules to teach them things like unsupported objects fall under Earth's gravity, right? These are just basic tenets of understanding. But what they discovered was that common sense is more than just learning rules. It isn't a set of rules. It requires abstract things like self-reflection, certain types of abstract thinking and reasoning, very difficult things to even understand that's happening in the human mind, let alone being able to transfer that knowledge and understanding into a machine. Machines need to understand not only things like factual information, which they absolutely do, but also how to make decisions in uncertain or changing situations. That sentence right there, it's a big sentence. How to make decisions in uncertain or changing situations. Like projects that had happened back in the 80s, there was one called CYC, and they tried to build databases.

B: I remember that.

J: I did too, Bob, it's interesting. I didn't realize what they were doing, I just heard about it, but what they did was they were building these massive databases of knowledge to give machines this common sense reasoning. And it would just be like statements of fact. This happens. This should typically happen, but not all the time.

B: Here's an example, Jay, that I specifically remember that they typed in. It was that a son would need to be younger than his mother.

J: Right, of course.

B: Ridiculous common sense, right? But it's like not to a machine.

J: But yeah, it's not, because where would a machine learn that if it wasn't told that, specifically? Now, human rationale, Bob, can infer things like that. And inference is a huge part of our knowledge base and our wisdom because we have the ability to think horizontally about things because we can infer things on the fly as if we already know them, which I think is fascinating. But they can't do that. They could not, back in the 80s, make this massive database function well enough to give it anywhere close to a match of the human flexibility of thinking, especially when new information came in and changed the context of a situation. This is all really complicated stuff that they're trying to figure out ways to get a computer to be able to process. So, currently LLMs rely heavily on probability, right? They choose the next word in the sequence based on what's most likely to be statistically to be the next word. But this doesn't help them with the idea of common sense reasoning. So, if you take my earlier example of whether to serve meat to a vegan, if you add details like but they're having a cheat day. The LLM might struggle because the context changes, and it is a very unusual thing. That's an unusual idea that it might not know about. It might not even understand what the two words together, cheat day, even mean. But a human could figure out, if they've never heard the phrase, I'm having a cheat day, a human would pretty easily be able to figure out what that means in the context and just move on like it's no big deal. I get it. I understand what you're saying. So the main question here is how can we improve what we have right now? So scientists can improve AI, and they can do that by programming them to be able to explain why they gave a certain answer, and I think this is really cool. Now right now, this information is hard to get from the LLMs that are functioning because they simply weren't programmed to spit out the linear thinking, the series of steps that it takes. I think the scientists know what they're supposed to be doing, but we want to see what it did specifically for this specific request. So if they asked why coffee left outside cools down, an LLM should be able to reason through the concepts like heat transfer and not just spit out the right answer, right? So an LLM could easily look up the right answer in its vast database of information, but we want it to also see the path of the reasoning that it takes to come to the answer, and that'll help us understand the processes that they're actually going through. Because there is a lot of black box stuff happening inside LLMs as they process, the complexity gets so profound that the human programmers don't have a full, full, full understanding of every single thing that's going on everywhere because a lot of these decisions were concocted by the computer itself on how to get from point A to point B. So what they want to do now is they want to go in and be like, no, show us every decision that you're making and why you're making that decision. And I think it's pretty obvious how beneficial that would be. Another idea is to give an LLM an open-ended task that requires planning or strategy. So in one test, an LLM was asked to navigate a board, like a game board, to collect energy tokens and then deposit them in the right spots, right?

E: Ooh, I like it.

J: You're right, it's pretty cool. And Ev it could be a little complicated for a human to learn a new board game and what the movement of the pieces are and what each spot on the board means and all that stuff.

E: Yes.

J: So the LLM understood the rules, and it did understand the rules. It's able to play with the rules, but it made weird and basic errors. And these are errors that a human with common sense wouldn't make. And what's funny is the LLM will find these really weird fringe cases of things that could make a weird decision, like actually placing the pieces on a part of the board that isn't a place where they should go on the board. Right? Because it doesn't understand. There is no understanding. So the LLM, they're far from handling these real world planning problems. They just can't do it. So what's the path forward? Researchers need to involve fields like cognitive science, psychology, and philosophy to understand how humans learn and apply common sense. We know a lot about that, but it would be great if we can get more into the weeds on that. And then this interdisciplinary approach will guide the development of AI systems that can replicate human reasoning. So we need to understand human psychology, cognitive science, and esoteric ideas like philosophy so we can better understand how to apply the common sense aspects of them to machine learning. That's a big deal.

S: I question whether LLMs are even on the path to doing that. We might need a completely different process.

J: LLMs don't seem to have the guts, the inner programming, to get them to the general AI nuanced thinking. They don't seem to have it. But I think the idea here, Steve, is that they're going to continue to broaden what LLMs can do. We're going to add this, then we're going to add this, we're going to add these modules that expand the way that they process information and make them more robust and more robust. So it's like from the ground up, they could, sure, expand an LLM to the point where one day it's no longer a large language model, it's something else. We have to define it differently because of how much all these other things that it can do and all these augmentations that have taken place. So that's fine, sure, you can get to a general AI by starting with this piece of software and dramatically improving all of these different pieces of it. But my gut still feels like there's something wrong about saying that an LLM is going to become a general intelligence, artificial intelligence. Sure, but I would think, and this is where I think we all agree, is that to get to a general intelligence type of artificial intelligence, the premise of the entire thing would have to just simply be functioning in a completely different way than the current day's LLM's function, right?

B: It's generally accepted, I think, that we need something beyond an LLM. It's not going to get there by itself to AGI.

J: It'll be like saying, let's take this scooter and turn it into a 747. And all the steps you'd have to take to get there, that's not the proper way to do it. Just don't start with the scooter. Start with a foundation that makes more sense with what you're trying to get to.

S: All right, thanks Jay.

News Item #2 - Malnutrition in Children (21:35)

• None 

S: Cara, tell us about the malnutrition problem in Africa.

C: Yeah, we're taking a hard left turn today, aren't we? UNICEF, the United Nations Children's Fund, put out a statement on Monday, as of this recording, about something called wasting. Sometimes you'll hear it called wasting, other times you'll hear it called severe acute malnutrition. And so this is a condition that is the result of insufficient calories, insufficient fat, insufficient proteins, and other nutrients. Children not getting enough of that to cover their individual needs. It also is compounded and complicated because very often when children are experiencing extreme poverty and therefore showing signs of severe acute malnutrition, they also have problems with their metabolism. They have problems with their immune systems. And it's easier for them to get infections. They often have – then they may have severe diarrhea and be dealing with severe dehydration. And this is a major source of death around the world. And so UNICEF has long worked towards – and a lot of other organizations – but has long worked towards providing treatment for severe acute malnutrition. And there's a long protocol. You can look it up online according to a lot of different guidelines. MSF, Doctors Without Borders guidelines, or the UNICEF guidelines for treatment. And other organizations have those guidelines as well. But one of the kind of staples of treatment of severe acute malnutrition is something called ready-to-use therapeutic food. And ready-to-use therapeutic food is a very nutritionally dense food. It's usually sort of in a little packet or bar. It's often made with, like, peanut paste, oils, sugars, fats. It's very highly calorically dense. And there is an action, a 2024 – actually, it started, I think, in 2022 – call-to-action called No Time to Waste that UNICEF put out calling for the funding necessary to be able to purchase and maintain enough stockpiles of this nutritional treatment across, I think, 45 of the countries that have the highest mortality rate due to malnutrition, due to severe wasting in children under five years old. And so they put out this call-to-action, and their goal was to raise between, I think, October 2022 and September 2023 to raise and deploy $1.2 billion in treatment funds for early prevention, detection, and treatment of wasting. You know, really just to give these timely therapeutic feedings, but also the antibiotics, the malaria medications and tests, the HIV-AIDS medications and tests that are necessary in these cases. They were able to raise $927.8 million, which is still quite a lot. They raised that globally, and it came from a bunch of different sources – $200 million from the U.S. government, almost $100 million from the government of the Netherlands, $35 million from Canada, $33 million from Ireland. The Bill and Melinda Gates Foundation raised a lot. Interestingly, the Latter-day Saints charities raised quite a lot – $5 million. So, like, made it in the top donor list. But it wasn't enough. It's not enough. And so UNICEF put out a statement just this week detailing the fact that in both Mali and Nigeria, stock-out – meaning running out of the RUTF, that energy-dense pace that I told you about – stock-out began this summer. In Chad and Niger, it's already started as of October of this year. And by December, they're projecting that stock-out will have begun in Cameroon. Starting next year, March of 2025, Pakistan and Sudan may start to run out of their supplies. And then April, May, June, we're seeing Madagascar, Kenya, South Sudan, the DRC, the Democratic Republic of the Congo, and Uganda running out of their supplies of this very simple and very necessary treatment for severe acute malnutrition for wasting.

S: Now, Cara, I understand that this is made partly out of peanuts. Would it help if Bob stops eating all peanut butter? Would that help? Or peanut butter related?

C: It would help if Bob sent all the money he spends on peanut butter over the course of his lifetime. Yes, probably do.

B: Wow, I gotta think about this.

C: And here's the issue, right? It's not just a funding issue, although funding is a huge part of it. You know, money is fungible and everything does come down to money. But part of the issue here is conflict, economic shock, climate crises, and supply chain problems. And so a big part of this program is to actually build out the ability to produce these foods locally so that they're not having to be made in other countries and then brought in. So the more there can be local production of RUTFs, those ready to use therapeutic foods, the closer they are to the source, right, the closer they are to the people that need them. And let me just kind of give you a couple of, a few statistics real quick, because I think that really drives this home. In Mali this year, over 300,000 children under the age of five are expected to suffer from severe wasting. In Chad, over 500,000 children are projected to suffer. And in the report that was put out by UNICEF, they were able to do some statistical analyses to see how many lives are saved through this severe wasting treatment. These 40, it was 47, I said 46 before, sorry, 47 countries with high, what they call under five mortality. They did an analysis based on numbers that they looked at from 2023 and they were able to estimate that in 2023, 1.2 million deaths were averted due to this program. 1.2 million deaths. And so this new call to action is stating that if we don't get the money necessary and continue this program, nearly 2 million people could die just of malnutrition, a completely preventable disease or syndrome, I should say. So this is an urgent call to action from UNICEF to the entire globe. And to be clear, like those of us here in the U.S., I know we have listeners from all over the world. Those of us here in the U.S. are giving, and the U.S. was a very large funder of this program, $200 million of it came from the U.S., but there's still more to give. I know for I don't want to make this political at all, but we're in an election season right now. And when I see the amount of money thrown away on campaign finance and how much that money could go towards saving lives, it's hard, yeah, it's hard to sit with that right now. And knowing that this is a very pressing need, if you are looking to give right now I don't want to, I don't want to bend you in any direction, but this is definitely a worthy cause.

S: It's interesting to think of nutrition as a treatment. I know they're treating a condition, but basically it's food.

C: We're talking about food, yeah.

S: But it's super concentrated. Does that matter? Is it just that it's efficient?

C: Yeah, that's 100 percent why.

S: Otherwise, any food would still be a quote-unquote treatment for the amount of nutrition as long as it was nutritious, right?

C: It would, but it's kind of complicated. As I was looking at the medical protocols, like the routine medical management, it's pretty complicated because at the early stage, you have to attack the hypoglycemia and you have to treat with antibiotics and you have to test for malaria and TB. And you start putting on the calories and you start adding fluids, but you have to be careful not to overload.

S: Yeah.

B: Oh, it's a process.

C: Yeah, it's a whole process.

B: You can kill somebody. You can absolutely kill somebody by just giving them a whopper.

C: Yeah, because all of their nutrition and their electrolytes are really imbalanced and they're dealing with, like, really severe pitting edema. They're dealing with a lot of difficulty just taking the food. And so there's different inpatient versus outpatient treatment protocols, rehydration solutions, whether they're using NG tubes, whether they're just able to orally feed. Sometimes when these kids are so severe, they're too lethargic and they can't eat or drink.

S: Right.

C: They can't physically swallow it or they'll throw it up the minute that they take it. So it has to be delivered in different ways. But part of the reason that these specific nutritional ready-to-use therapeutic foods are used is because they are just incredibly nutritionally dense. I think I have here the description of what they're made of. Powdered milk, peanuts, butter, vegetable oil, sugar, and a mix of vitamins and minerals. So it's just very, very dense and it gives a lot of what's necessary kind of as an early or late stage intervention.

B: Sounds yummy.

C: And it probably is actually quite palatable for children too. And that is important.

S: Yeah. Interestingly, often when we talk about topics like this, we often get feedback from people saying that basically the problem is overpopulation and they really advocate strongly for population control. Although I haven't found their arguments convincing and that they generally ask them, so how do you propose doing that? You know, are you going to kill people? I mean, seriously, it's like, sure, it would be easier to manage if we had a lower population. But what exactly are you advocating? And they never seem to have a good answer to that question. But they just say, well, they just say, well, we shouldn't just keep feeding people because that just worsens the problem of overpopulation. So you are saying let them die.

C: It shows a poor understanding, and I see this a lot when we talk about hunting or trophy hunting or animals that are critically endangered. There's this assumption because we tend sometimes to think statistically, like there's equal disbursement across the globe. But I'm sorry, we're running out of these RUTFs and children are dying in Mali, in Chad, in Niger. Are these the areas that these people are talking about where there are too many people?

S: Well, often yes, yes, because they have a lot of kids. But here's the thing that I also, where I'm headed with this, is that if you're genuinely interested in population control, there's two things you can do that are clearly effective. One is women's rights and two is taking care of kids, reducing infant mortality or child mortality. People have more kids as a hedge against a high child mortality, and they end up overpopulating, if you will, having more kids than they otherwise would have. But if they were confident that their kids were probably going to live to be adults, they would have their two and be done with it. This is just what generally happens. So when in any country that modernizes or industrializes to the point that women have more self-determination and child mortality goes way down, their population tends to decrease over time. And it's the poor countries that are having the population explosion problem, because it's a kind of a survival strategy, ironically.

C: And I hear what you're saying, and I think that's a really important point, right? Empowering women is a –

S: And keeping kids alive.

C: And keeping kids alive. But the point I was trying to make, which I think we glossed over and I just looked up some statistics. The population density of the United Kingdom, for example, is 279 people per square kilometer. The population density of Mali is 20 people per square kilometer.

B: Oh, wow.

C: I think that often this misplaced and deep judgment is a bit xenophobic and is rooted in some of these difficult questions about poverty. Those people over there are doing that thing. We are all promoting or increasing or contributing to overpopulation. But the poorest people in the world are suffering the most from it.

S: Sure.

B: Disproportionately.

C: Exactly. They are disproportionately suffering and they are not disproportionately using resources and they are not disproportionately population dense.

S: Right. Although we're talking about two different things. You're talking about population density. I'm talking about population growth. And of course, there are some parts of the world like India that have both. But either way, the solution is women's rights and then fix poverty. You do those two things and that takes care of your population issue. Then we actually have the opposite problem where it's hard to get people to have enough kids to maintain a population.

C: Yeah, we don't have to worry about that problem.

S: Yeah, but yeah, we're probably population experts project that will probably peak at around 10 billion people on the planet and then population will start dropping. But who knows there's going to be so many cultural changes between now and then. But anyway, the population question is interesting, and I just always find it fascinating when I get into conversations with people on that, and the facts actually make it so that doing the right thing is actually the solution here.

C: Yeah, and I also think that sometimes I struggle when we talk about global health and we talk about these really difficult topics that are hard for people to sit with, right? It's hard to look in the eye of wasting disease and actually see its dramatic effects. But there is an important conversation to be had here about the difference between prevention and treatment right now. These kids are dying right now.

S: Oh yeah, totally. You don't want to do one instead of the other, right? Well, we need to fix the long-term problem. Of course we do, but we still need to treat it right now.

C: But yeah, that is not the response to this problem. The response to this problem is to make sure that no child on this planet dies because they don't have enough nutrition in their body.

S: Yes, I agree.

C: It should not happen.

B: Let's pony up, especially the billionaires.

S: That bit of poor logic comes up a lot when we, like for example, when we say, oh, we could use golden rice in order to treat vitamin A deficiency. And people say, the real problem is poverty. I'm like, good, you get on that. In the meantime, in the meantime, I'm going to give them vitamin A so they don't go blind. Is that okay? While you fix poverty, you know? So yeah, you don't, don't say, oh, no, we need to fix the long-term problem instead of doing something short term. No, you need to do both.

C: Yeah, and very often, it's also a distinction between I need to fix this systemic, multifaceted, multivariate problem versus there is a solution for this right in front of our faces. We need more of these foods. Let's figure out how to make them and get them where they need to go.

S: Yeah, and bias the time to fix these really hard, long-term, intractable problems.

C: Exactly.

S: All right, thank you, Cara.

News Item #3 - Latest Starship Launch (37:13)

• Latest Starship Launch - NeuroLogica Blog 

S: Guys, I'm going to give a quick update on the latest Starship launch. Have any of you guys seen that one?

E: Oh, yes. All over the news, that one.

S: Yeah, so this is impressive. This is the next in the series of Starship test flights. Starship is SpaceX's heavy lifter. It's a system, right? There's the Starship system, which includes the super heavy rocket booster and the Starship, which is the upper stage, which is the part of the ship that would go to the Moon or go to Mars, right? That would carry cargo and people. It is, when combined, the biggest rocket, I think, the most powerful rocket that anyone's ever developed. And this, of course, is following the SpaceX's pattern of reusability. So this test got farther than any of the previous ones. It successfully launched the stack with the Super Heavy plus the Starship on top, launched, completed the launch so that the Starship got into its suborbital trajectory, not into orbit. And then, this was the big test, was having the super heavy return to the launch pad and essentially land on the launch pad. So like we've seen Falcon 9 rockets do, in order to make that component reusable. However, there's a catch, and I mean that in two ways. The Super Heavy is too heavy to land on legs, so they created this tower with these mechanical arms that grab it out of the air. The rocket has to return to this tower, get into position so that the arms can grab it. And that whole procedure worked. It successfully grabbed the Super Heavy.

J: Remarkable.

S: The rockets turned off, and the Super Heavy was hanging there by the chopsticks, as they call them, by those arms. So it worked. And then the other piece of it, the Starship, also returned to its designated location in the Indian Ocean. It got all the way down to the ocean, the water surface. It then flipped on its side and blew up, which apparently was okay because they had no intention of recovering it. And they were like, we were done with it anyway, that was fine. But I don't know what it really means that it blew up, if that was a good thing or a bad thing. But that wasn't part of, yeah, they were never planning on recovering that stage in any case. So they're considering this pretty much a complete success, but I thought the whole thing, I wanted to talk for two minutes about the whole idea of why they had to sort of catch this super heavy booster out of the air. If we back all the way up, the main problem is that it's really hard to get out of and back into Earth's gravity well. One G is a mother. And so you need these massive rockets to get into even suborbital height, let alone into full orbit, or then to get to some other destination. Although by the time you've gotten into orbit, you're 90% to wherever you want to go, right? In terms of the energy expenditure. Although even still, if we're going to use the Starship to go to Mars, they're going to have to refuel in orbit. They're not going to be able to get there without refueling on just one booster. So anyway, could you name all the different ways that we've developed to bring spacecraft back down to Earth? How do we safely bring stuff back down?

J: We use parachutes.

C: Oh, to Earth.

J: We use retro rockets.

S: Yeah, so SpaceX has innovated using the rockets to literally land, or at least get down to the surface so you can get caught. So that's one method. Somebody said splashdown. So use a combination of parachutes and splashdown.

E: And then what? Are we including like a space shuttle landing, like a plane?

S: Land like a plane.

B: There's always lithobraking.

S: There's lithobraking, which I do not recommend, but yeah.

B: That's when you... lithosphere, which is the surface of the Earth. Lithobreaking means you crash into the Earth.

E: Oh, I see what you're saying.

B: It's totally viable. I mean, it's a way to get back. Not a good one.

E: It is a way.

S: Yep. And then the Starliner uses parachutes and airbags. Although they haven't had a chance to test that with actual people yet. And also what I found very interesting is that the Starship itself, because the Starship also was going to be grabbed out of the air. That's how it's going to be brought down to the Earth. It could land on the Moon, because there's much less gravity, there's one-sixth the gravity. And NASA is going to use it as the lunar lander, as one of their lunar landers for the Artemis program. But if we're bringing the Starliner back down to Earth, again, it's going to use the capture method, the chopstick method. The reason for that, for the Super Heavy and the Starship, is they're too heavy, they're too big to engineer legs that can support them to land on the ground. The legs would have to be too big, add too much weight to the rocket, right? So they figured it's easier, think about this, it's easier to snatch this thing out of the air than to engineer legs for it that aren't too heavy.

B: Wow.

J: That's so crazy, right?

B: That's awesome.

J: Talk about thinking out of the box.

S: Right? It's so counterintuitive though.

B: I love it.

S: Yeah, but obviously it would be better if it had just legs and it could land anywhere. But the downside of this is that you have to go precisely to the tower that has the mechanical arms to grab you.

E: Yeah, yeah. No room for error there.

B: Very little.

E: You lose your whole tower.

B: I would think it would be a matter of feet.

S: Yeah.

E: Yeah, the tolerance in there? Oh, gosh.

J: The question is, though, can they do that 50 times in a row?

S: Exactly.

E: Yeah, how reliable?

S: Yeah, and it's interesting to think about how even today, because I just recently watched a couple of Apollo shows, because we all love the whole Apollo era and everything, and it's so obvious that they were operating right at the very edge of our technology and our ability.

B: And luck.

S: Yeah.

E: Oh, they needed things like pen and duct tape and, you know.

S: But even just the tolerances, like we're going to have just enough fuel to make this whole thing work out. Like they couldn't really even bring extra fuel, right? We're literally at the at the very edge of our engineering. And here we are 60 years later, whatever is 50, 60 years later. And it's still true. We are still operating at the very edge of our engineering and our material and our ability. The Starship is made out of steel. Not that there aren't more advanced materials, but in terms of cost-effectiveness and everything, SpaceX went with steel. We're still using the same rocket fuel. The physics haven't changed. The basic trade-offs haven't changed. Really, the one big difference between our current rocket program and Apollo, what would you say the big difference is?

E: The Current Program and Apollo.

B: Computer power?

S: Computer. That's it. Computer power and all of the related electronics like sensors and blah, blah, blah, all that kind of stuff. That's incredibly different. But the rockets, it's the same fuel. It's the same basic principle.

B: It's the same rocket equation.

S: It's the same rocket equation. Nothing has fundamentally changed.

E: Do they still use slide rules to figure trajectories out?

S: They use computers. But the thing that has changed is computers.

C: Now they're not people.

S: But also like, you couldn't do the whole landing back on the ground again without computers, right? I mean, you need computers for that level. You're not going to manually bring that thing in. So it's interesting to think about. It's still really hard, really hard to get into space and to get back down again safely.

E: Damn gravity.

S: Yeah. That's why going to the moon, like if we really flesh out our cislunar infrastructure and have like a permanent functional base on the moon, the moon is a great launch pad to everywhere else in the universe. You know, because it has such a smaller gravity well. The Earth is a terrible place to launch from.

News Item #4 - New Metasurface (45:41)

• KAIST Unveils Janus Metasurface Responsive to Light 

S: Alright, Bob, tell us about the latest metasurface.

B: Alright guys, researchers have upped the ante with metasurfaces, creating for the first time a Janus metasurface. Janus or Janus, both are good, one's British pronunciation, one's American. They have created for the first time a Janus metasurface that can manipulate light differently depending on which direction it's going. This has never been achieved before in any optical system. It's kind of fascinating and the potential of this sounds pretty cool. Okay, this latest advance is from KAIST Institute of Researchers in South Korea. Hyunhee Kim from the Department of Materials, Science and Engineering and Jun Kyo-yong are the first, the co-first authors, and it was published online first in the journal Advanced Materials, and it's going to be published on the best day of the year for their October 31st issue. The paper's title is Bidirectional vectorial holography using bilayer metasurfaces and its application to optical encryption, blah, blah, blah.

E: Scary.

B: Don't worry about that. But what the hell was that all about? Alright, first we have to talk about the idea of a metasurface. I think we've mentioned this once or twice, obviously related to metamaterials, which we've talked about more than twice. Metasurfaces are essentially two-dimensional versions of metamaterials. Because they're much thinner than most wavelengths of light, incredibly thin. This thin surface is composed of nano-engineered components in specific patterns that can powerfully interact with and influence light. Now, in my opinion, this type of technology, this metasurface technology, is by far the most interesting and cutting edge in optics. They're really doing some amazing stuff here. But not only is it these metasurfaces lighter and thinner than conventional optics, it can manipulate light better than anything in nature or any other tech. And to prove that point even more emphatically, these researchers have created the first Janus metasurface. It's called Janus. Anyone have an idea why it's called Janus?

C: I don't know what that means.

J: Too many letters.

B: Janus or Janus is the two-faced Roman god.

J: Okay.

B: One looking to the past and the other looking to the future. A non-godlike Janus metasurface can manipulate light differently depending on which direction the light is coming from and entering the device. So one example that they give, the researchers say is that the light coming through the device in one direction will magnify the light just like a magnifying glass, but coming through the device in the other direction would give you, it acts like a polarization camera telling you information, polarization information about the light that's going through it. So basically it's one optical device that can do, that behaves like two different optical devices, two in one, if you will. So that, and that leads specifically to what the long standing challenge that these, that this breakthrough resolves. A Janus metasurface needs to selectively and minutely control three properties of light. The light's intensity, its phase, and its polarization. Those are the things that needed to be manipulated in order for a Janus metasurface to even exist. So the intensity of light is kind of straightforward. Essentially, the higher the intensity, the brighter the light is, right? More technically, it's the power per unit area carried by the specific wavelength of light, but it just means that's brighter. Higher intensity means brighter light. The phase of light refers to the specific position on a waveform. Are you talking about the top of the wave crest, the bottom somewhere, or somewhere in between the top and the bottom? And the polarization is a little bit more complicated. Imagine light. Light is composed of oscillating electric and magnetic fields at right angles to each other, right? And those two oscillating fields are at right angles to the direction of light travel. So that's fundamentally, at its basic elements, that's basically what light is. Now, polarization describes the orientation of that oscillating electric field component. It's really all about that. So polarized light has its electric field oscillating in one plane. It's up or down, left and right, or any angle in between. Whereas unpolarized light has the electric field oscillating in many planes. It's all over the place. So as you might imagine then that polarization types have many applications in telecommunications and optics. It's really important. So now creating a Janus metasurface that can minutely control the intensity of light, the phase of light, and the polarization has been sought for many, many years. This is something they've been wanting and trying to do with conventional optics, and now with metasurfaces, and they finally pulled it off. Professor Zhang Huixin from the Department of Materials Science and Engineering at CAST said, this research has enabled the complete asymmetric transmission control of light's intensity phase and polarization, which has been a long-standing challenge in optics. It's opened up the possibility of developing various applied optical devices. All right. Creating one device that can both magnify and determine the light's polarization may seem esoteric and just for labs doing inscrutable experiments. But this will likely lead to many different applications that will get your attention much more dramatically than what's going on here in the lab. So just for just a few examples in closing, what seems reasonable to extrapolate what a more mature Janus metasurface can do, it could lead to adaptive camouflage that changes depending on the surroundings. Now we've talked about metamaterials. I think the very first time we talked about metamaterials, God, 12 years ago, I don't know. The first time it came up, and it always came up in this context, it was with camouflage. And they have been using they have used metamaterials to to produce some type of camouflage invisibility, if you will, but it's very, it's for a very specific scenarios with very specific non visible frequencies of light. But there's these Janus metasurfaces, I think it seems like they could do an interesting job at adaptive adaptive camouflage camouflage. And who knows how far that will go. But this is something it seems like it's kind of made for that type of idea. Also, Janus Metasurfaces could revolutionize medical diagnostics and surgeries by doing something like real-time enhancement of image clarity and 3D image depth. Could have a huge impact there. Steve, you're like this one. Janus Metasurfaces could enhance solar energy harvesting efficiency.

S: Yeah, that would be great.

B: By taking into account the angle of the light as it hits the solar panel. So it could actually in a sense change its structure depending on the angle of the light that's hitting it. Is it noon coming kind of straight from the zenith or is it down at an angle later in the day? And it will actually adapt to those angles and maintain a high level of efficiency throughout the entire day. Also, as we robotically explore and colonize the solar system and beyond, these systems can optimize signals that have to travel really long distances in space through different cosmic environments. And imagine if some of our most distant probes, Pioneer, had something like this, the signals that we would be getting could be much more optimized for efficiency and robustness and that we would have a much better signal. For example, that's one example. But there's so many other things that these systems can impact. Holographic displays, augmented reality interfaces, environmental sensors. It's all over the place. So I'll be tracking this tech. It's really interesting. Of course, I'm always tracking metamaterials and metasurface technology. It's just so fascinating in so many ways. But this one was interesting. Check it out online if you're interested.

S: All right, thanks, Bob.

Who's That Noisy? + Announcements (55:03)

S: Alright Jay, it's Who's That Noisy time.

J: Alright guys, last week I played this Noisy. [plays Noisy]

S: It's uncomfortably long.

E: Oh, and it's something that's stretching and it's going to break and the tension in both ways is like unbearable.

J: Alright, ready guys? A listener named Jeremy C wrote in and said, Hi Jay, I'm likely wrong, but I guess this week's noisy is the sticky foam backing to a scotch or 3M wall mount hook being pulled out and stretched until it releases from the wall or surface.

C: It does sound like that.

J: I will say this is a very good guess and you'll find out why. Going on to the next one, Shane Hillier. Okay, it for sure sounded like a kind of shepherd tone. Maybe it's a guitar player trying to mimic a shepherd tone by dragging multiple picks down multiple strings at different times or intervals. I thought that was a very interesting guess, but that is not correct. Candice Dennison wrote in and said, my first instinct was balloon curling slowed down, right? This is when you run the blade of a pair of scissors over the ribbon of a balloon. You know how you do that? You curl it. Sure, there's a little element of that in there, but that is also not correct. Chris W. wrote in and said, hello, I hear a few things. My guest will be pulling the last really long piece of tape off the end of a roll. Chris is representative of about 30 or 40 people who wrote in something similar, and he says he's hoping that he will make it to NOTACON this year.

S: Nice.

J: Yes, that'd be cool. Visto Tutti wrote in, he said, this one sounds just like the sound of zip lining. You know, zipping down a cable in a harness through the canopy of a forest. And then he said, I'm going to kick myself if finally it's actually a bird. This is not a bird, and it is not zip lining, my friend. We have a winner. I was very excited when I finally found someone who won because I got through about 100 emails before. I was about to be like, OK, I can't do this anymore. And then Albert Flores. I hope that one of those two is the way you pronounce your last name. He said, hey, all longtime listeners since at least 2012. First time, who's that noisy guesser. This is the first time I am almost certain I know what the noisy is. It must be the sound made by the adhesive that holds an iPhone battery in place when you pull the release tab, stretching the adhesive and releasing the battery from the phone assembly. That is exactly what that is.

E: Is it really?

J: Yeah, he said I've personally done this hundreds if not thousands of times. So I think he might do this professionally. So let me play that noise for you one more time. [plays Noisy] Yeah, so what happens when you want to disassemble an iPhone and you want to take the battery out of an iPhone? They have an adhesive tape in there that I think is like essentially the release mechanism to get that battery out of the chassis. It's really remarkable how many tiny little screws and doodads and whosey whatsers go into an iPhone. And this is just one of the many, many things that are in there. Anyway, very cool. If you're interested, look it up on YouTube.

E: Quick question. Do you guys remember when you could replace a battery inside of a cell phone? I remember. I always had Android, I never had iPhone. But why does that seem to be entirely gone?

C: Well, it's similar to laptops. I think it's just that the more integrated it is, the tighter they can compact everything and they can make it smaller, sleeker.

J: I think it's also kind of bullshit. There should be one button you press just like I had a computer. You press a button, the battery comes out. And you could get a new one in there. I bought a, Steve and I, when we bought laptops for the SGU, because we needed them for travel and all that, and my battery came and it was a dud, right? The battery came and I'm like, tried to replace it. I began the process and it's like a four hour process.

E: Oh my gosh.

J: Like ridiculous.

S: Yeah. With laptops, they basically decided that they were going to make engineering and design options, decisions that were optimized for making it a small laptop, not upgradeability or even replacing the freaking battery.

E: Right.

C: That's why the phones look like that too. It's how they can make them small and sleek.

S: Yes, right, by making it impossible to replace anything.

E: Bad for the consumer, right?

C: Well, it depends on what you prefer. I think probably it's not—I mean, it's easy to say it's bad for the consumer because that's what you want, but people buy them like this, and they love them like this.

S: It's a tradeoff.

E: What's the choice?

S: People want the slim profile, so that's the tradeoff you make.

E: I don't know. Are they manufacturing both of these side by side, and it turned out to be a victory for the non-replaceable battery? I don't think so. I think they forced us all into this choice.

J: I don't ever remember getting a memo about this.

C: No, that may be the case, but I have a feeling, and it would be interesting to look at market research, that people are going to go smaller, slimmer. They're going to choose the sleeker, sexier phone over the replaceable battery. No, I know.

S: But usually for things like that, the industry is going to follow consumer demand.

C: Yeah, they're usually market testing this stuff.

E: Okay.

S: And if it was like a big missing demand for laptops with replaceable batteries, even though they're bigger, then somebody would make that laptop and sell a lot of them. But apparently people don't care.

C: Now your iPhone or any phone really has the magnetic charger on the back and you can put an external battery on it. You just slap it to it and you can take it right off. So you have the big battery that's got all this power, but it's not part of your phone.

E: Does that extend the battery life inside the phone?

C: Well, it makes the battery – you mean the life of the actual battery? Probably not.

E: Okay. All right. So you're still forced to buy a new device every three years. There you go.

C: Totally, yeah. These things are built to be disposable at this point.

S: Jay, what's the new Noisy?

J: I have a new Noisy this week, and this Noisy was sent in by a listener named Patrick. [plays Noisy] If you think you know who this week's Noisy is, if you think you've heard it, you know it, you can email me. If you heard something cool, you email me at WTN@theskepticsguide.org. And one quick note, don't forget, if you're going to send me a Noisy, make sure you remember to attach the Noisy to the email.

S: I'm assuming that people forget to attach it.

J: It happens all the time.

E: They get excited, you know.

J: And then I have to email them back, and then they email me back, and it's this process, and I don't have the time.

E: Oh, that's how they get them to interact with you, Jay. Yes, they do. Clever audience.

J: All right, so guys, there's things going on. SGU, this is a podcast, right? We make podcasts in this podcast. We like to talk about science and critical thinking, and we also like to make each other laugh. And when we're together, we like to have meals with each other. That's just the way we roll. Anyway, we would love it if you would think about this and consider becoming a patron of the SGU. We are in our 20th year. We are over 1,000 episodes. Every single one of these episodes has been free to the public. We do have ads in our show. That's one way that we pay to make this thing happen. The other way we do it is we have a membership on Patreon. You can go to patreon.com/SkepticsGuide and you can help support the work that we do. We would really appreciate it if you would consider to do this and help us expand our reach and to continue doing what we do. You could also join our mailing list. We have a mailer that goes out every week. We talk in there. We have a lot of fun. We tell you everything that we've made and produced the previous week, and there's other stuff going on in there. I have been working on an SGU weekly puzzle that I'm hoping that I can finish soon. I hope you're interested. You could just join our mailing list by going to the homepage. Now, we have shows coming up. We have two shows in Washington, D.C. on December 7th. We have a private show. This is a private show plus. It's a two-hour live podcast recording. And then we have an hour of fun and other stuff that we do. It's unique every time we do it. And that night, if you're still in D.C. that day and you want to see us again, you can come to the Skeptical Extravaganza of Special Significance. This is our stage show. We've been developing this show for 10 years. It's constantly evolving, and we have been having a wonderful time doing this. It's always a good time. We always have an incredible time with the audience. The big gun, the one that I am the most proud of and the most excited for, this is NOTACON 2025. This is the SGU's conference. The idea of this conference is socializing. It's our community building conference. It revolves around meeting people, socializing, and of course there is a ton of stuff that we do to entertain you over the course of those 2.25 days. So go to theskepticsguide.org or notaconcon.com. Ian, what the hell? Anyway! Please join us. If you don't, Cara will get really angry.

C: And you don't want to see me angry.

J: No, you trust me, you don't.

S: All right. Thanks, Jay.

J: Bye.

Emails (1:04:56)

S: One quick email. I'm just going to basically just read the email. This is more of a public service kind of announcement. This is a response to our discussion about myopia. You guys remember that? This comes from Randy, and Randy writes, Hey there. I just caught up on the recent episode where you talked about the myopia epidemic. I'm an optometrist, so I'm dealing with myopia every single day. As a public service announcement, I thought you might mention a few points. One, there are methods and treatments available now that help control, slow down, the progression of myopia. Spectacle lenses, contact lenses, eye drops, and combinations thereof. Other than the treatments noted above, other recommendations are kids should be outside a minimum of two hours per day. Sunlight seems to have a positive effect, as we said on the show. Frequent breaks from screens. Kids should avoid the use of LED lights while studying reading in the evening. One final point, Cara mentioned that it's not so bad to be myopic. But which I suppose is true until you become a strong myope. An increased degree of myopia is directly correlated with medical eye health problems such as retinal detachment, glaucoma, myopic degeneration, etc. So everything should be done to minimize the degree of myopia. Cheers, Randy. Cara, what do you have to say for yourself?

C: My prescription's like negative 1.5.

S: So mild myopia, yeah, it's not a big deal. But he's saying it's not that myopia itself is such a big deal, but we want to minimize severity because as it gets more severe, it increases the risk for these serious eye problems. Yeah. Okay. Well, thank you, Randy. Guys, let's go on with our interview with Christian Hubicki.

Interview with Christian Hubicki (1:06:30)

S: We are joined now by Christian Hubicki. Christian, welcome back to the SGU.

CH: It's great to be back. You're such a delight to talk to about science and robotics, so thanks for having me back.

S: Well, thank you. It's always fun to have you on the show. We're going to get right to the meat here because you're on for a specific reason. You're a roboticist, I understand. Is that right, Christian?

CH: That is correct.

S: I have that correct.

CH: And you can pronounce my title correctly. I'm very impressed, Steve. Very impressed.

S: And there's some recent robotics news out recently, and we're like, hey, let's get Christian on. He's a roboticist. So tell us about Elon Musk's recent announcement about his robotics.

CH: So yeah, so Elon, as many people know, is the CEO of multiple companies between SpaceX and all the cool SpaceX news recently. But what we're here to talk about today is his work at Tesla and what Tesla has been doing in terms of their pivot to robotics. So every year in October, they've been having some kind of AI/robotics event. In the past, it was called AI Day. And this year it was called We Robot. And the idea is they were going to roll out a bunch of technologies that they've been working on, or I should say products that they've been working toward. And one of the things that those headliner was a robot taxi and all the various accoutrements that they're making to help the robot taxi. That's its own subject, but. What piqued my interest was the humanoid robot news. So for those who don't know, Elon and Tesla et al. have been making this optimist humanoid robot, because of course you have to name it after some kind of pop culture robot. Apparently that's a law. It's a law run. I think Bob knows this law. That's Isaac Asimov's fourth law of robotics. And they've made some pretty fast progress, basically going from not having any of these humanoid robots to having what seems now to be some kind of a fleet. And so as my specialty is in humanoid robotics, legged robots, that's where I cut my teeth in the research field. And so I always have to watch these events, see what's going on. And I was seeing everyone talk about it. And one thing that was really interesting is they walked out the humanoid robots, and they had them mingle with the crowd. There was something like roughly about 20 of them at the reports I've seen. And mingle with the crowd, pouring drinks, playing like Rochambeau with people and charades. It was incredibly impressive. But the thing is, as roboticists, we looked at this and we're like, oh, this is teleoperated. There's a human in the back with in a motion capture suit that that's making it do these things, which is not a non-trivial, which is a non-trivial task to get a robot to do it. But of course, the misinformation started to spread almost instantly that these robots were all autonomous. And this became a bit of a controversy because the people at this event, they're not necessarily roboticists, a lot of tech bloggers and types who are maybe not particularly familiar with the technology here. And so I thought it was so fascinating to see a kind of misinformation spread so quickly. And one that's sort of born out of a lack of information, I would say even a intentional lack of information from the company. They just trotted these robots out, they're doing stuff. And I don't know about you all, if I'm at a robot event and people say, oh, look at our awesome robots and they're doing stuff, you don't expect necessarily that there's a person in the back that can do those things, right?

S: Because there's a person on stage right next to the robot controlling it transparently, right?

CH: Exactly. Exactly. And I think – so there's a lot of – and I know I've talked to you all in the past about I've always been hopeful that one day I get to talk about a robot story on the SGU. And so this is – because what I'm starting to think is that as robotics becomes more popular, much more salient in the public eye, people are doing more things. And there's a lot of incentive to have people believe things it's doing that it's not. You're going to run into these misinformation snafus, if you will.

S: It's amazing that this is basically a mechanical Turk-con.

CH: What is Turk-con?

S: The mechanical Turk. So if you're not familiar with this story, in 1770, a guy, a German created, and a Hungarian, created an automatic chess player, an automaton chess player. And the robot, if you will, had the turban on, so he was called the mechanical Turk. And he would beat, go around as a show, beat people at playing chess, right? It was a strong game of chess. And this went on from 1770 to 1854, 84 years.

B: What?

S: 84 years. This same machine was going around, obviously operated by different people. And it turns out there was a person inside the box playing chess.

B: Yeah, it's a little person.

S: And operating the machine. I don't think it had to be that little of a person. Just not a huge person.

B: What I read is, yeah, it was a little person.

S: Apparently there were a lot of really good chess players at the time. It was not hard to find a sequence of basically chess masters that would get in the box and beat people at chess.

B: What's in the box?

S: Pretending to be this mechanical Turk. And so that kind of that term has represented whenever you have a person pretending to be an automaton of any type. And here we are, 2024, and Elon Musk is pulling a Mechanical Turk con.

CH: Yeah. And I feel like I should point out here that, like, there's a good reason to have the ability for these robots to do teleoperation. And just to make it clear, what I'm talking about is that somewhere, somewhere else on the premises, there is a person wearing a motion capture suit. Now, you might think of motion capture suit, you might think Andy Serkis in Lord of the Rings doing some kind of like inside of a motion capture volume with a bunch of cameras. You don't need to do that anymore. There are actually suits that you can wear that have inertial, basically motion sensors in the suit, more or less, tracking the acceleration of your joints, and therefore doing motion capture, which is great. And we'll be wearing a VR headset that's also tied to a sensor, so it actually knows how you're actually tilting your head. And then the robot mimics those movements. And the reason you'd want this as a robotics laboratory is for imitation learning. And the idea is that you can try to get a robot, basically train a neural network to do a bunch of tasks, so long as you have examples of robots doing those tasks that you then record the data and you train the network to do those things. People sometimes call it behavior cloning. It's cool, and it's one of the leading contenders for how people are trying to make robots useful in homes and things like that. But right now, as promising it is, it's still quite limited. Like if you go to like the most recent papers on this kind of technique, like really cutting edge labs working on it, they will put out a paper which will have like maybe two arms doing like 12 household tasks and they'll be like, we got the reliability up to 80%. And we're like, oh my god. That's so good. Congratulations. You're moving the field forward. I mean –

B: Yeah.

CH: But you wouldn't put this out in a live demo in front of a bunch of people expecting to pour four out of five beers correctly. There's a good reason that they have this technology for teleoperation. It's just – and it makes sense that they would want to show it off. But it's exactly as you say, Stephen. The way you show that off in a way that's maximally transparent is you have a person right next to it doing the teleoperation.

S: Right.

CH: So people see what's going on. That was – it's obvious. And I think that when we look at this, we're like – you can infer an intent to mislead especially when this is a company that has investors that are going to want to pour money into the company based upon whether or not they're impressed or not. I mean you can infer those kinds of things even though we don't know. And only in the last – I think in the last day, finally some information is leaking out from inside the company saying, yeah, yeah. These were teleoperated. But that's the gist of it.

B: Yeah, I mean, it's devious and slimy are two words that come to mind.

S: Well, it is. I do think this is the Silicon Valley culture of fake it till you make it kind of thing. And you're always trying to, whether you're looking for investors or whatever, you're trying to seem like you're farther ahead than you are. In this case, that was pretty deliberate deception. It seems, from what I'm reading, that the Tesla people were saying, we're like years ahead of the competition. They're saying that explicitly when, in fact, they're years behind the competition. So the deception was to make it that very specific misdirection, misleading people as to where they were compared to their competitors. And in fact, from what I'm reading, the competitors have started to label their demonstrations no teleoperation, just to make sure like, yeah, we're not cheating like the Tesla guys have been cheating.

CH: That's exactly right. That's been happening for a while because when you're in the robotics business, you know the ways that you could potentially massage a video to make it look more impressive than it is. But thankfully, at least in the academic space, you report on that in the paper. You say, actually, this is what we did, and you try to make the videos as clear as possible. You say, this is how they operated. Or another trick that people will do is they will speed up the video and it makes it look a lot more fluid than it actually is. This is something that actually Tesla kind of got a little bit burned on about a year ago. They had this really nice demonstration of actually showing seemingly behavior cloning. And it was just picking up a block and putting it down on another side of the table. And it was really – and using – it was autonomous. I can confirm because I actually might have talked to the person who coded that or one of the people involved with that research project. And I believe it. It looks like it. But it looks so smooth and that's because when they put together the video, it was at 50% increased speed. And so there are tricks like that. And I think for me, the thing that's a little bit of a shame is that they are making progress. I mean are they years ahead of the competition? No, I don't believe that. But one way that they are doing a very good job is that they have a lot of these robots. And it shows that they are moving forward with improving the fabrication. And the fact that they had something on the order of 20 of these robots doing a live demonstration, even if it's teleop. I mean I would expect a lot of these robots just to keel over because of loose wire. But apparently, the story that's being told from the event, there's only one fall out of these 20 robots, which I consider a pretty big success. I mean now that's just a robotics research standpoint. But from a product standpoint, less so. But I mean that's impressive that they worked at least that well. And very few companies have like a fleet of these robots. I mean I can think of one or two that actually has a fleet of robots doing live demonstrations. And normally, they're in factories. So there's stuff to show. But I think it's just not as glitzy and glamorous as trying to show off like it's bartending for you.

J: So Christian, when you look at these robots function during the event, and I watched all the video, Were you impressed with the robots, the way that they move, the hands, the articulation? What are you perceiving when you look at that? Is it well-constructed? Does it look like it could do what they want it to do in the future?

CH: I'd say from a hardware standpoint, I am pretty impressed, like the amount of articulation in the hands. And again, just the fact that they're well made enough that half of them didn't keel over for no apparent reason shows that they have some good engineers working on this stuff. I'm also looking at the algorithms beneath the surface. I mean, it sounds a little silly to say I'm looking at the algorithms, but I'm looking at the motion and inferring the kind of control that they're doing. And as far as I could tell at this event, there were three types of control, sort of three modes of control that were being employed. One thing that the engineers who have sort of leaked out off the record have said that, well, the walking wasn't human-assisted. I'm like, well, yeah, you do not want to tele-operate walking. You want the robot to be balancing per its own sensors. You know, it's actually a bad idea to tele-operate walking. That makes a lot of sense. There's also a mode where they were standing and performing a pre-recorded routine, it was a dance routine, and they're all doing it in sync. And that is actually, that's also pretty standard robots have been doing that for decades. But again, they have a lot of them, so it's actually, so I gotta give them credit for that. And the third is this teleoperated mode where they are standing, you'll notice they're not walking while doing teleoperations, at least I've not seen in the videos I've seen, they're standing still and the upper body is being captured by the motion of the people. And you can see the lower body is adjusting for balance, kind of shimmying around to make sure it's not falling over. So only the top part is being teleoperated with the arms and the head and the shoulders and such.

S: That is interesting and it makes perfect sense, although I never thought of it, but now that you mention it, of course it has to be that way because when we are maintaining balance, there's a moment-to-moment feedback loop where our brain's getting a lot of sensory input, including where we are with relationship to gravity, vestibular function, visual information, tactile information, muscle information. And then we use that to generate balance. A teleoperator is getting none of that feedback from the robot. So they can't create balance for the robot using teleoperation.

CH: Exactly. That's exactly why. And those algorithms for not falling over, those are pretty standard. They've improved over the years. And you can even see on the way that it walks, and this is not a knock, it's a pretty safe way to walk if you're on a flat surface, but you put your foot flat on the ground, then you quickly shift your other foot over, and then you shift your weight over. Therefore, there's not a very long period where you're falling. But the adage with human walking is we're always falling. We're always falling forward and catching ourselves.

S: Interesting.

B: But now that said, it doesn't sound too unreasonable to me in this scenario to have a robot walking itself, but then having a tele-operator using the arms.

S: That's what he's saying. That's what was happening.

B: No, no. I thought you – Christian, I thought you were saying that the legs were basically locked down when the teleoperator was doing their thing.

CH: So, typically – yeah, as far as I could tell, the legs were standing still while the teleoperator was doing their thing probably because they didn't want it to perturb the balance while it was walking. Now, in principle, you could do the teleoperation while walking. It would be – it would – and not even in principle. It's – that's – I could see it as a doable thing. It's just an unnecessary risk when falling over. I will say it's also a bit brave. I mean I'm not there. I don't know what the engineers are doing in terms of safety. But having a robot around people in close proximity is – can be a safety hazard if you're not careful. I mean not because the robot is going to reach out and strangle them or anything, not to give them any ideas. But like they just fall on people. They fall over.

B: Yeah. Heavy machinery.

S: They're probably heavy.

CH: I mean like they're at least probably on the order of 100 pounds or more, these things, and that's not – and they're not made of foam, these things. They're made of metal and stuff. There's actually a video that came out recently. There's a company called Unitree which actually makes a lot of really, really good robots on the cheap. They're out of China and a lot of research labs buy them because they're cheap and you can just – and you can break them and who cares? You buy another. And they have a humanoid robot series and they – and they're – that they have out that they were doing a demonstration in China where the robot was going to conduct the orchestra. And the robot raised its hand to start the orchestra and all of a sudden – I think it had a sensor error is my guess and just lost its balance. And then the way – and so the arms were moving around like it was conducting but it was basically losing its balance and charging toward the violins. And so it looked like a monster going –

B: What the hell? That's epic.

CH: So I don't know whether or not to applaud them or not for having robots in such close proximity playing rock, paper, scissors with people. But that's one thing I always look out. And I think that one thing that people also clocked – I think that clued a lot of people in that this was a human behind this – that there was a literal man behind the curtain is all of the robots, they would talk with different voices of different operators.

S: So different operators had different voices.

B: Yeah. There you go.

J: Did they sound like robots? Did they make the voices sound robotic?

S: No, not from what I saw in the video.

CH: No, not that I can tell either, Jay. So to the point where I was like, did I miss an announcement at this event saying, by the way, these are all teleoperators? To me it was so obvious, but apparently not.

B: We're not denigrating teleoperation because it's an amazing thing.

S: Just be transparent about it.

B: Yeah, exactly. I mean, for me, the most sublime manifestation of that is a doctor on the other side of the world teleoperating surgical tools performing an operation. That's the kind of stuff that happens through teleoperation and it's incredible. And also from the point of view of the person teleoperating, if you've got like vision and maybe some tactile feedback or just being able to see yourself in the robot in the environment that you're controlling, from what I hear that the feeling that you are there is so powerful. You are transported because a lot of your senses are there. And it's an amazing experience to do that. I would love to do it sometime to really experience some sort of telepresence using teleoperation. So obviously, we're not knocking that. That's amazing. But this is a different beast when you're trying to trick people into thinking that it's an autonomous robot when it's not.

CH: Yeah, and I and just sort of put like, and just build off that point one more time, Bob, like there is a legitimate case you could have made as a tech company that's saying we are making a big play towards autonomous robotics. And we think that that play is to gather as much data as possible from human robot from human teleoperation of robots. And that's very clearly the play that they're making. There was a reporter that contacted me a few weeks ago that was saying that they asked for the comment that say they found a bunch of ads trying to – of Tesla trying to employ dozens of people for 24-7 data collection of these robots. And so if you had 20 or 40 or more of these robots with people 24 hours a day giving examples, you would probably have the biggest trading dataset out there in the whole dang field. So there's a play you could make with that. Now, I personally am skeptical whether that's going to scale up to good autonomy because I still think there's a reliability gap that that won't cross. But there's a case that you could make that's honest and I – but it's just not as sexy, Steve, as I always say.

S: Yeah.

B: Yeah. Well, what about this, though? What about instead of doing meat space, if you will, training or robot space training, could can't you just do some some types of training virtually, have virtual environments where they were in an environment or virtual environment that has gravity and has friction and has an atmosphere has everything that could help train it more quickly because if it's virtual, I mean, you could spin up a thousand virtual servers and with the environment and having these robots try to navigate a virtual reality that's similar to Earth in terms of these things that I mentioned.

CH: Your intuition is spot on, Bob, as a competing method for robot control. It's called Sim-to-Real reinforcement learning. So in a simulation, you do training and then you take what you learn in simulation and you put it on the robot. And where that's been really successful is in actual leg and locomotion, particularly quadrupeds with robot dogs. And part of the reason is that people have figured out how to simulate the environment of a robot dog on a flat surface or even a rough surface pretty darn well, even to the point where the feet are just spheres contacting a plane. And we have those physics down pretty well. Now where it becomes more tricky is what if you want to fry an egg? What if you want to squirt ketchup on a hamburger? Then the simulation, the physics involved in that simulation are a lot more complicated and difficult to model. And so trying to train a simulation to do that actually works a lot worse. So it's no surprise that the legged people tend to do a lot of reinforcement learning in the simulation where the physics are better and the manipulation people, the hand people doing kitchen tasks with all kinds of weird, viscous stuff like beating eggs. They like the behavior cloning, real-life training data.

B: It makes sense. Interesting.

J: Christian, how far along are there robots?

CH: I mean that's tricky to say, Jay. I mean I'd say where they shine the most is very clearly they've done a lot of iteration and they've seen how these things work when you've built dozens of them. And there's only like one or two companies I think that – other than Tesla who are doing things like that. Unitree comes to mind but I don't think they're doing anything too practical with those. And Agility Robotics which is a company that I have my friends and my friends work at and they're trying to do more warehouse tasks. So their robots are I think are going to be very well tested. In terms of their algorithms, it's hard to point at anything in particular and say, aha, Tesla is doing this way better than their competitors like at Figure AI or Sanctuary Robotics or some of these others. That's hard to point at. And I really would encourage your audience to look up Hondas Asimo, A-S-I-M-O. I mean that's a robot that is a humanoid robot built by Honda and it was walking and running, picking up objects autonomously, playing a little bit of soccer, lots of really cool demonstrations. And you might say, hey, when was the last time you saw that? Well, you haven't seen it in a while because it came out in the year 2000.

B: So I thought I was so excited back then. And then it just seemed to plateau relatively quickly. I think they came out with one or two versions after the initial ASIMO release, and then that was it. I think they just lost interest in in taking that any further.

CH: It's an incredibly expensive showpiece for Honda to have done that. I mean, for them, it was always a showpiece in a way so is the Boston Dynamics Atlas. I mean, they never Boston Dynamics humanoid thing. There is never any intention to sell that thing. But it was always to show, hey, we know how to make robot control algorithms. And it was, for my money, the most famous real-world robot for many years. But now the difference is people are trying to find a good market for it, whether it's in a factory, that's what Elon was saying like last year or earlier this year, and now he's saying, no, it'll be in your home. I'm like, pick your poison. They're both two different kinds of hard.

S: They're very different and also completely worthless as a tele-operated, right? Because if you're replacing a worker, having a worker tele-operated is – just have the worker do the thing.

CH: Exactly. So again, so clearly it's a play for training data for an algorithm. You can see though the way it's being sold. In fact, I think – I wish I had the quote in front of me. But it was something that Elon Musk was saying to the effect of two years ago, we had a dancer in a suit. That was our robot. That was famously – and every year – and now we're here. So if you extrapolate that progress, you will be very impressed where we will be and I hope I didn't make this entire panel just break out into hives by just raw extrapolation of progress.

B: Wow.

S: I'm still trying to figure out how to think of Musk as a technologist, right? You know, his political shenanigans aside and just thinking about, like, I think he genuinely advanced the state of the art for battery electric vehicles. You know, he made a great product and he pushed that market forward. Then he squandered his lead with the freaking Cybertruck, which is, I think, just terrible, terrible. But anyway, then same thing with SpaceX. I mean, he invested a lot of his time, a lot of his money into that company, and they definitely pushed private rockets into space significantly.

B: And look what they did recently, catching a landing rocket.

S: Yeah, we're talking about that as we are talking about on this episode, absolutely. Then he has these other projects where I know it feels like he's just throwing a lot of crap against the wall, or it is this kind of shenanigans where he's basically just taking existing tech, trying to pretend like it's his own, not really pushing it forward, and then pretending like he's more advanced than he is. So, I'm thinking like Neuralink is the same exact thing that feels to me like the same as these robots. Is he really doing anything different with Neuralink than any of the other research outlets are doing? He's making all these claims about redefining the boundaries of human, whatever. No, you're not. You're not redefining anything. You're just sort of duplicating what other people have already done and not really pushing anything forward. I'm always trying to figure out, is he actually doing something here or just pretending to? Because he has billions of dollars to invest. If he really wants to put $10 billion into making robots come to market, he could do it. Or in advancing brain-machine interface, he has the money to do it. But for those projects, I don't really see how he's actually pushing the field forward.

CH: I think that my take on him is that when he's focused, he does so much better. When SpaceX was his primary focus, I mean, he's still doing great stuff, to be clear. But he's got his fingers in so many pies. I'm sorry. I think he's just spread too thin. This is my personal opinion. And so it's hard to truly advance in so many things at the same time. And I think what you end up doing is taking in like the gestalt of where the field is and be like, oh, I'll do that, but better. The but better is I'll maybe pay for the best roboticist. But they're all coming from the same laboratory. I think that's the thing that you've said many a time. It's not like you crack experts out of a hermetically sealed bottle that know something other people don't. These are people who work in our laboratories. They go and they intern over the summer. Then they go work there, and that's what happens.

S: Yeah, the cynical view is that he's just sucking up talent and calling it his own.

CH: Yeah, I mean, that would be the cynical view, but that's what all companies do. I mean, like, in terms of, like they want to hire the best from the laboratory, and as a person who runs the laboratory, we want our students hired by the best companies.

S: That's fine if that's what you're doing transparently, but when you're doing that and claiming that you're revolutionizing the field, when you're just doing what everybody else is doing with your own people. I agree. I think he's spread a little thin. A number of years ago, he said, I'm about cars and Mars. That's the Musk that I like the best when he was about cars and Mars. He was focused. I love Tesla at the time. I love SpaceX. And I just haven't been impressed, got to say, overall, with anything he's done since then.

CH: I would love to hear more of your opinion on the Neuralink. I'll have to go back to some of the episodes, probably to find you talking about that. But that's my opinion as well. And I think that when it comes to the tactical strategy of robots, of humanoid robots here, there's maybe some percentage, I'll just guess, like 25% moonshot, this will actually solve all the things. And let's just say 75% making Tesla continue to look like the future when it's hard to make an electric car look like the future anymore. So a mix of those two things improves the visibility of the company. I mean, look how people are talking about it. These tech influencers nominally would be saying, oh, look, humanoid robots, the future, and they don't necessarily have the historical context that people have been working on these for decades.

S: Well, Christian, this has been exactly as fascinating as I was expecting it to be.

B: Slightly more, slightly more.

CH: Well, I appreciate that. This was a lot of fun. And again, this is all because ever since I was listening to the podcast, so you know, just many moons ago, I was like, one day, I know, I know they are going to come from my field, and I'll be ready for that day.

S: Absolutely. Well, definitely, we'd love to have you back on the show whenever there's any big robotics news out there, or just to have you on, because you're a cool guy.

B: Yeah, man.

CH: Well, thank you.

S: All right, Christian, take care.

E: Thanks, Chris.

CH: Take care.

S: Thanks, buddy.

Science or Fiction (1:35:52)

Theme: Good News Item #1: The US Geological Survey released a report that conservatively estimates there is enough commercially accessible geological hydrogen to meet current global demand for 1000 years. Item #2: A new review finds that aquaculture reduces demands on ocean fishing and therefore benefits local communities dependent on pelagic fish. Item #3: Researchers have demonstrated bedside rapid quantitative blood biomarker detection using an acoustic pipette to purify and label biomarkers in just 70 minutes.

Answer	Item
Fiction	Item #2
Science	Item #1
Science	Item #3

Host	Result
Steve

Rogue	Guess

Voice-over: It's time for Science or Fiction.

S: Each week I come up with three science news items or facts, two real, one fake. And then I challenge my panel of Skeptics to tell me which one is the fake, just three regular news items. Although they do have a little bit of a theme. The theme is they're all good news. But one's fake.

C: Good news, everyone.

S: OK. Here we go. Item number one, the US Geological Survey released a report that conservatively estimates there is enough commercially accessible geological hydrogen to meet current global demand for 1,000 years. A new review finds that aquaculture reduces demands on ocean fishing and therefore benefits local communities dependent on pelagic fish. That's fish in the pelagic zone which is basically where the fish are. Item number three, researchers have demonstrated bedside rapid quantitative blood biomarker detection using an acoustic pipette to purify and label biomarkers in just 70 minutes. Jay, go first.

J: Steve, the US Geological Survey released a report that conservatively estimates that there is enough commercially accessible geological hydrogen to meet the current global demands for a thousand years. Damn! I mean, where would they be finding this hydrogen? Enough commercially accessible geological hydrogen.

S: I mean, under the ground, I'll give that to you.

J: It's cased in the ground, and it's hydrogen. Christ, that would be awesome.

S: They call it gold hydrogen.

J: You see that? That's good stuff right there, man. All right, so okay what do I know about this? What could I, how do I pick holes through this? I mean, that's a lot of hydrogen and that shit is flammable. I mean, that's dangerous stuff right there. A thousand years worth, oh boy, I don't know, a thousand years worth of a gas. It's probably embedded in a rock somehow. Okay, let me put that on the back burner. A new review finds that aquaculture reduces demands on ocean fishing and therefore benefits local communities dependent on pelagic fish. What's pelagic mean?

S: Yeah, so as I said, that's in the pelagic zone of the ocean, so not by the shore and not in the deep ocean.

J: In the middle ocean.

C: Yeah, it's just the open ocean.

S: The open ocean.

C: It's not like the floor.

S: It's where, yeah, where a lot of the fish are.

J: All right. Great. SO people who are breeding fish that they're not in the ocean, like fish farms and all that stuff. And okay, this one makes perfect sense to me and that would be wonderful if that's true because we would very much like fish populations to increase in the ocean. The third one, researchers have demonstrated bedside rapid quantitative blood biomarker detecting using acoustic pipette, which I know what it is now, to purify and label biomarkers in just 70 minutes. Steve, could you please explain this?

S: So this is a technology for detecting like antibodies, right? Biomarkers, protein biomarkers. That's both rapid and quantitative. So right now we have one or the other. So you know how you can do a COVID test and you get an answer right away? But it's not quantitative. It's yes or no. It doesn't tell you how much. Or you could do a quantitative detailed test, but it takes a couple of days, three, four days, whatever, and it's expensive and everything. So, this is a technology for doing it rapidly in 70 minutes at the bedside with a small amount of blood and quantitatively. So, that's sort of the first time that we would have this technology.

J: Okay. I've made up my mind. I think that the hydrogen one is fiction and I'm going to tell you why. First, to have a thousand years how much is it, like a thousand years worth, whatever, it just basically means an incredible amount of hydrogen that is embedded in the rock or whatever. I think it's possible that it could be there, but it could be unobtainable. That's possible, but I just think that's a lot of hydrogen. I don't know if I would never have heard about this happening before with these giant pockets of hydrogen. It seems shaky to me, so I'm going to say this is the fiction.

S: Okay, Evan?

E: Yeah, hydrogen. I meet current global demand for a thousand years. I wish I knew what the current global demand was. I don't, but a thousand years.

S: Would it really help you if I told you what the current global demand is? If I put a number on that, do you think that would help you?

E: Maybe. How is it measured even in like gaseous tons?

S: One hundred and five million metric tons per year.

E: Okay. That's impressive. So, times one thousand of that, or scaling it up going forward, I'm assuming. Okay.

S: No, that's at current usage.

E: At current usage. Okay. That does help. It helps me a lot, actually. But moving on to the second one, about aquaculture. Sure, I suppose so. I mean, new review, sure, nothing wrong with that. Showing that, okay, it's a study. I'm not seeing anything here that really throws me a curveball or something I can hang my hat on. A number or something that seems out of sorts, so I think that one's science. And then the last one, whoa. So the last one here about the blood, right? The quantitative blood biomarker detection. This has a whole, I don't know, Elizabeth Holmes kind of vibe to it, doesn't it? Like Theranos almost promised sort of we'll do your blood and we'll have it back to you in an hour. And so it sounds almost like that. I'm sure it's not that. But 70 minutes, that's kind of what makes it. Oh, gosh. So I think it's either that one or the gas. All right, Jay. I'm going to follow you. I'm going to sink or swim with you this week. And I'll say that the hydrogen is the fiction. I don't think we have a thousand years of demand there.

S: Okay, Cara.

C: That's the one that I felt the most science-y about.

E: Oh, no.

C: I mean, it's like how much oil is in the ground. There's so much. That's not the problem. We have a ton of natural resources. The problem is what happens when we extract them. So, I don't know. It's like, sure, there's probably enough geologic hydrogen. That doesn't mean we should go and get it. I don't know. That one doesn't bother me that much. I'm torn between the aquaculture and the rapid quantitative blood biomarker detection. And my spidey senses are tingling because of exactly what you said, Evan, that this feels very Theranos. But then I was like, wait, in the hospital we draw blood all the time and we look at things in the blood very fast. And we go, hey, look, this is your glucose. It's not good. Or, hey, check it out. Here's a protein that's not looking great. So then I was like, wait, why is that weird? Also, researchers have demonstrated rapid quantitative blood biomarker. I think these keywords matter. OK, so they showed that it could work. They're not showing that it's actually happening. They haven't integrated this. This is not a functional thing yet. When I look at the aquaculture one, it's so vague. Okay, a new review, so again, operative word here, a review means they looked at a lot of studies, and they found that aquaculture reduces demands on ocean fishing and therefore benefits local communities dependent on pelagic fish. The thing that's bothering me about this, and my guess is that it's the opposite. And the reason I guess that is because I think this has been the narrative for decades. If we just farm these fish, people will stop fishing in the ocean. But I think what we're seeing is that when we farm fish, people go, oh, look, cheap fish. Let's get more fish. It's so cheap, we can keep eating the fish. And so I have a feeling that this one is the fiction.

S: Okay, Bob.

B: Yes. So I see a disparity here. The hydrogen one seems like, really 1000 years, where did where did this even come from? And then the the quantitative blood biomarker detection, that sounds like a really cool advance. But the aquaculture one seems to be like, oh, yeah, it seems to be an outlet seems to be different than the other two, it seems more reasonable. So that's one of the reasons why I'm going to pick the aquaculture.

C: Yeah, Bob!

S: Alright, so we're split between one and two, so we'll start with three.

C: Oh no, he sounds so happy!

E: Sometimes reasonable sounding things turn out to be reasonable.

S: Researchers- That's true.

B: We shall see.

S: I try to mix it up. All right. Here we go. Researchers have demonstrated bedside rapid quantitative blood biomarker detection. You say that ten times fast. Using an acoustic pipette to purify and label biomarkers in just 70 minutes. You guys all think this one is science. And this one is science. This is science.

C: Yes.

B: Cool, man.

S: Yeah, it's pretty cool. This is a big deal. But Cara, you kind of nailed it. They've demonstrated the technology. They don't have a commercializable device.

B: Yeah. Scaling it up is different. But hey, man, they did it.

S: It's like a proof of concept kind of thing.

C: But that's good. That's the first step.

S: Yeah, not bad. And it is kind of a big deal. I mean, I know it's kind of a little jargony in there, but as I tried to explain, this is new. Like, we can't do it both rapid and quantitative. And if you're going from days to 70 minutes for like a bunch of antibody and protein biomarker things, yes, Cara, we could measure things like blood sugar and nitrogen and all that stuff relatively quickly. But this requires a fundamentally new technology.

C: Oh, OK.

B: Jay's been on the toilet longer than 70 minutes.

S: Yeah. The acoustic part of it separates out the cells from the proteins. And then it also then labels the protein so that it could be detected by a fluorometer.

C: No way. That's cool.

S: Yeah, it's pretty cool. It uses the standing wave, it shows like a standing wave where we're pushing the proteins out to the side and pushing the cells up into the middle so that they're separated out. Yeah, it's pretty cool.

B: It doesn't sound like a cheap test, though.

C: No, yeah, I was going to say this one's expensive.

S: No, no, the whole point is to make it cost-effective as well. Because the equipment that we use now is expensive and big, so this would be cheaper than the current option. And they said it has a greater than 99% efficiency.

C: What are they actually looking for? Biomarkers of what?

S: Like just any protein or antibody.

C: Oh, okay.

S: Yeah. So this might be like you could do like a COVID test and not just positive, negative, but here's your titer. Like here's an actual number of how many antibodies you have. Yeah, it would be cool. So yeah, hopefully it pans out.

C: But you might also be able to do like some early Alzheimer's detection.

S: Sure.

E: Yeah. Very handy.

S: Yeah. Yeah. A lot of applications because we'll just keep going backwards. A new review finds that aquaculture reduces demands on ocean fishing and therefore benefits local communities dependent on pelagic fish. Bob and Cara, you think this one is the fiction and Jay and Evan think this one is science. And this one is the fiction. Because Cara is correct, it's the opposite. But I don't think for the reason you were saying, though, Cara.

C: I'm always right for the wrong reason.

S: People were just eating more fish because fish is cheap now, but it's not that. So what's another hypothesis? What might you think? So it actually increases fishing and worsens the situation for local communities because it's taking all their fish away. And what are they doing with that fish? They're making it into fish food for the aquaculture fish.

C: Oh, Jesus. Are you serious?

S: Yeah. Yeah. That's the thing. Most people don't realize that aquaculture fish are fed fish. They need to be fed fish oil and fish protein.

C: They're fed smaller fish.

S: Yeah, because that's what fish eat, right? They just eat other fish.

E: Yeah, they eat each other.

C: And so you have these local communities that are eating bait fish as a main food source, but now they're taking the bait fish and they're feeding the bigger fish that, like, the inland rich people want to eat.

S: That's pretty much exactly the point of the study. So there's something called the FIFO, fish in, fish out, right? And that is a way of, it's a metric to say, like, how efficient is this agriculture? Like, how much fish are we putting into the system? And how much fish are we getting out of it? Obviously, we want to be getting out a lot more than you're putting in. Otherwise, what's the point, right? And one of the points they were making in the study was that some of the industry had been hiding their true numbers. And the way they do that is to take an average of their carnivore species and their herbivore species. And then presenting that as their average FIFO number. But actually, the carnivore numbers are much higher because they need to eat a lot more fish protein and fish oil. And that's things like salmon, which is what most of the fish are, most of the aquaculture fish are. So the whole industry is a lot worse than they made it seem by sort of combining those numbers into an average and kind of hiding the bad carnivore numbers. Does that make sense? So what's actually happening is that the majority of fish oil that's produced in the world is going to feed agriculture fish. And a lot of this pelagic fish that people would just eat directly are now being turned into fish food for upscale salmon for people in the West. And it's actually happening like in local communities in Africa are losing access to their pelagic fish, which is being shipped overseas for agriculture salmon to be sold in wealthy nations is basically what's happening. Some of the authors say that agriculture in many cases is not a food production system, it's a food reduction system, because it's just adding this layer of inefficiency, and it's shifting a lot of these fish calories from poorer countries to more wealthy countries. Not a good thing. Agriculture is a very efficient and sustainable system of producing food at its best, but like everything, it starts to follow the money. The industry is basically just following the money rather than optimizing things for sustainability. So that was basically the point of the study. Not good. So this actually turns out to be not good news. But that means that U.S. Geological Survey released a report that conservatively estimates there is enough commercially accessible geological hydrogen to meet current global demand for a thousand years is science. Now, the commercially accessible is the key there, right? So we've talked in the past about the fact that there is this underground hydrogen in pockets, right? When the conditions are right, the hydrogen can be trapped for a long enough period of time that it'll actually collect in rather large deposits or pockets. The older thinking was that, well, hydrogen's too light an element and it would just sort of seep through the ground and then go away into space and it's not collecting under the Earth. But it turns out it is collecting. If it's deep enough and the ground conditions are good enough, it can actually contain these pockets of hydrogen. The US Geological Survey tried to estimate how much hydrogen they think is trapped in the crust of the Earth. They came out with an estimate of 5 trillion metric tons of hydrogen. However, most of that is not commercially accessible because it's too deep. Or it's in too small a pocket to be commercially viable, right? It would cost too much to get to it. You wouldn't get enough hydrogen out of that deposit. So they also estimated, well, how much of that five trillion metric tons is commercially accessible? It's drillable and in a big enough deposit that it'd be worth doing. And they came up with an estimate of 2%. Which is a conservative estimate. It could easily be twice that, or three times that, but conservatively, if they said it's 2%, that would meet current hydrogen usage for a thousand years.

C: That is bananas. That's so much hydrogen.

S: It's a lot. It is a lot.

C: I think if I had completely overlooked that commercially accessible, I may have picked a different answer.

S: Yeah, I know. I noticed that, and I always have that dilemma, like, do I point it out or am I giving it away if I'm pushing them towards the wrong answer? You know what I mean? So I just didn't say anything.

C: But the number of non-commercially accessible is far beyond what I would have expected.

S: It's 50 times.

C: That's insanity.

S: It's quite a bit. And the reason this is important is because, again, I'm not a fan of hydrogen in general for a lot of uses because 98% of our hydrogen is gray and it actually releases more CO2 to make it than just burning whatever the fossil fuel source of the hydrogen in the first place. So it's not a good process and we basically have to do it for certain industries and we shouldn't be burning green hydrogen or more environmentally friendly hydrogen for fuel when batteries are better and more efficient. Whatever hydrogen, it's not all green, like it's color coded, right, but any greenish hydrogen should be going to the industries where it's going to do the most good. That's my opinion, at least for now. But the whole game changes if there's just this massive amount of hydrogen sitting under the ground. And again, it's color coded as gold hydrogen, this geologic hydrogen, because it's already in the hydrogen form, right? It's already in the burn. It's already a free gas, like all the energy has already been put into it to make it into free hydrogen. So where does it come from? Essentially, it comes from water trickling down to molten like volcanic, if you will, like magma. Magma, right, Jay?

J: Magma. Magma.

C: Magma.

S: So it's ultimately geological energy that heats up the water. It breaks it apart into oxygen and into hydrogen. The hydrogen often leaks away, but it can get trapped in pockets. But that also means That it's a renewable resource, right? Because it's essentially, there's a cycle to it. If we take out the hydrogen and then burn it with oxygen in the atmosphere, it turns into water. Water eventually leaks back down to the magma, gets turned into more hydrogen and oxygen. There's a circuit there, right? That's a sustainable kind of thing. And this happens on a human timescale. It's not like oil, which is made on a million year geological timescale. There's still so much to learn, so they don't know like how much is made over what period of time. We're still not really sure how much there is. Again, it may be a lot more even than this thousand year estimate. And of course, I would assume that if, as Evan said, if we can get a, if there's a source of environmentally friendly and commercially viable hydrogen, that's going to increase our use of hydrogen, right? But that's a good thing. We want to do that. We want to take this, what would be a completely or mostly green source of energy and displace other less green sources of energy.

C: So long as the extraction is also green.

S: That's what I'm saying. The only thing that's not green would be the extraction itself, like whatever energy we have to expend in that kind of process, which is mostly you drill and cap a well, and then you've got a source of hydrogen. Yeah, then you've got to pipe it around and everything. I mean, hydrogen's kind of a difficult gas to deal with because it is so light. So there's still some inherent problems to it technologically, but still, it would be worth it if we had that much of it, of a green energy source, like carbon neutral energy source, just sitting under the ground. It'd be certainly a lot better pumping hydrogen than gas, you know what I mean, than oil and crude, as far as the environment is concerned. So yeah, this could be a game changer. This is potentially really good news. I just don't know how long this is all going to take. Is it going to be something like we're going to start seeing these projects popping up in 10 years or 50 years? I just don't know. I don't think anybody knows at this point. They're still in the early days of research. But definitely something I want to keep my eye on.

E: I'm going to go with the fish as the fiction. That's my answer.

C: Final answer.

B: You had your chance.

S: All right. Well, what you can do, Evan, is give us a quote.

Skeptical Quote of the Week (1:56:58)

"One of the beautiful things about science is that it allows us to bumble along, getting it wrong time after time, and feel perfectly fine as long as we learn something each time."  – - Martin A. Schwartz, (description of author)

E: I will give you a quote. All right. The quote is suggested by a listener, John from Tallahassee, Florida, suggested this quote. "One of the beautiful things about science is that it allows us to bumble along, getting it wrong time after time, and feel perfectly fine as long as we can learn something each time. That was written by a gentleman named Martin A. Schwartz. Martin Schwartz is or was, unsure at the moment, with the Department of Microbiology at UVA Health System, University of Virginia. And the reason I say I'm not sure is because that was written back in 2008 in a paper that he wrote titled The Importance of Stupidity in Scientific Research.

C: I love that.

E: Which was published in the Journal of Cell Science.

S: I think he's at Yale now. And I think he's still going.

E: That's great. Maybe he's a listener as well.

S: Professor of Biomedical Engineering and of Cell Biology.

E: Most likely the same person.

S: Martin A. Schwartz, cell biologist.

E: Martin A. Schwartz.

S: Yep, probably him. I understand there's some smart guys at Yale. All right. Thank you, Evan.

E: You're welcome.

S: And thank you all for joining me this week.

J: You got it, Steve.

B: Sure man.

E: Thanks, Steve.

C: Thanks Steve.

S: —and until next week, this is your Skeptics' Guide to the Universe.

S: Skeptics' Guide to the Universe is produced by SGU Productions, dedicated to promoting science and critical thinking. For more information, visit us at theskepticsguide.org. Send your questions to info@theskepticsguide.org. And, if you would like to support the show and all the work that we do, go to patreon.com/SkepticsGuide and consider becoming a patron and becoming part of the SGU community. Our listeners and supporters are what make SGU possible.

[top]