Nature Podcast

This is a transcript of the 28th April 2016 edition of the weekly Nature Podcast. Audio files for the current show and archive episodes can be accessed from the Nature Podcast index page (, which also contains details on how to subscribe to the Nature Podcast for FREE, and has troubleshooting top-tips. Send us your feedback to


Adam Levy: This week, the soviet internet that never was.

Benjamin Peters: Why, with all the motivations means and the mathematics to be able to build a computer network, do the Soviet scientists stumbled so badly?

Kerri Smith: And the year after a huge earthquake hit Nepal, the threat of landslides means it's still not safe.

Jane Qiu: What happens after the quake is just as important as what happens during the shaking

Adam Levy: Plus locating where words live in the brain. This is the Nature Podcast for April the 28th 2016, I'm Adam Levy.

Kerri Smith: And I'm Kerri Smith.


Adam Levy: Now, Shamini Bundell has become a bit brain obsessed this week. No, she's not turning into a zombie, she's been learning about language and where we keep all the words in our heads. Nature 532, 453–458 (28 April 2016)

Shamini Bundell: The internet is full of ways to procrastinate and this week I have found a new one. It's an online 3D model of a brain that Alexander Huth at the University of California, Berkeley, sent to me. It's interactive, brightly coloured, and you can zoom around and click on different parts to make colourful words pop up – a perfect distraction – but it's actually serious science too. I called Alexander to find out how his team came up with it and what it all means.

Alexander Huth: We wanted to map how the meaning of language is represented in the brain. So we took subjects, we put them in an MRI scanner and we scanned their brains while they listened to hours of natural stories. So we're recording, essentially, blood flow at different points in the brain and then we could use that to map out which parts of the brain seem to represent different concepts or different types of words.

Shamini Bundell: And that's how you ended up with this beautiful model that I've been playing with all week. You can click on different bits and see different words but what did it actually show you?

Alexander Huth: Right, it showed us that these broad brain regions are really composed of many small brain areas that seem to represent different categories of words. So there are some areas that respond to social words or words describing people and social situations and there are other areas that respond very strongly to words describing visual properties or numbers or many other things, and this is a much more detailed map than anybody had ever seen before of the semantic network.

Shamini Bundell: And are these all in particular language parts of our brain?

Alexander Huth: So, all of these areas are actually in a broadly distributed network that covers much of what we might call the association cortex. So, this is sort of, these high level areas that are probably doing many different kinds of functions. So they're not only doing language, they're doing all kinds of higher level cognition and so on.

Shamini Bundell: So once you got this model showing sort of different words at different places, what did it actually tell you?

Alexander Huth: So we found these very detailed maps, but then we also saw that the maps are very consistent across subjects. So they look really similar across people. This is something very high level. Language is something that is learnt so the fact that that is organised pretty much almost the same way in any subject that we studied was really interesting.

Shamini Bundell: But is there any logic to the organisation, if it's the same across people it seems like there must be some reason why numbers would be in a certain place.

Alexander Huth: Yeah, so for numbers, we know that one of the major representations of numbers is in an area that's sort of involved in like spatial reasoning, words describing visual properties of objects. There's a lot of representation of those words, sort of near visual cortex. So there is some organization in that sense.

Shamini Bundell: How does this study differ from previous research?

Alexander Huth: Most earlier research has focused on single words and to me that is a very limited way of looking at language. We don't communicate in single words, so what we've done instead is look at language in the natural setting, that's much harder to analyse which is why we have to use sort of fancy computational models but I think it gives us access to much more detailed information about how language system is organised.

Shamini Bundell: And when you say language in the natural setting, you mean stories. You used The Moth, a radio hour podcast, so quite complicated language.(extract from The Moth) She digs back in the front again deep, deep and she pulls out a pack of matches that have been laundered at least once.

Shamini Bundell: So compared to single words that must be activating a whole lot of more brain and that's quite important if you're not looking for just that one language centre area.

Alexander Huth: One of the things that make language really interesting to study for me is that language gives you access to all kinds of other cognitive processes, right. So when you use language, you can make people do all kinds of interesting things in their brain right. You can make them think about numbers. You can make them think about their childhood. You can make value decisions, any kind of thing that we can do with our brains, we have words to describe. So language gives you whole brain in a way, language gives you everything.

Shamini Bundell: And how confident were you that your map is correct that word does in fact activate that brain area in all circumstances and it's not just a random chance in this case?

Alexander Huth: We tested the models that we built by using them to predict the held out data set. So we collected second data set for subjects which has them listening to another story and we didn't use that data to fit the model at all, so the model hadn't seen that data in anyway and then we used the data to try to predict responses on that held out data set and when we did that we found that these models predicted quite well in a lot of these brain areas.

Shamini Bundell: So you're taking new stories and predicting what the brain will do. Could it work the other way around; could you take the brain scans and predict what the stories were?

Alexander Huth: Yeah, yeah, that's very interesting. So we call that a decoding model, in general, when we try to go from the brain responses to the stimuli that elicited them. Can we decode language that somebody is hearing and then even taking that further can we decode language that somebody is just thinking which would be very interesting both scientifically and medically for people who have communication disorders or say locked in, that might be a very useful sort of brain prosthetic device. So we're working on that.

Adam Levy: That was Shamini Bundell talking to Alexander Huth from UC Berkeley in the US. He's put an interactive version of the model online so everyone can procrastinate with science. You can find it at That's Huth with a U and we've also got a video giving you a guided tour through the Brain Dictionary which you can find on our YouTube channel,

Kerri Smith: Coming up in the Research Highlights the plant that bleeds nectar and the reason you might sleep badly when you're away from home.

Adam Levy: But first, it's hard to imagine life without the internet and it's almost as hard to imagine what the world would look like if the internet had developed in a different way or with different priorities. Kerri has a story of a kind of internet prequel ; Internet −1.0, if you like.

Kerri Smith: Science is full of positive results and not so good at remembering the null results; the experiments that were tried or failed or the ideas that were just too good to be true. The history of science has the same bias. Say you study the internet, you know it evolved in the US from a network called the ARPANET built by the military but used by scientists to share data, but the trouble is there isn't another internet for comparison, nothing to tell you if the properties of our current internet would always be the case for other internets or if they're unique… Except for this one other idea. Nature 532, 438–439 (28 April 2016)

Benjamin Peters: The OGAS project or the All-State Automated System in Russian that's the Obchegosudarstvennaya avtomatizirovannaya sistema.

Kerri Smith: That lovely bit of Russian was spoken by Benjamin Peters who is a communications scholar at the University of Tulsa. The All-State Automated System or the OGAS project is the soviet internet that never was.

Benjamin Peters: Probably the most ambitious attempt to network a nation in the world so far; it was to be a real time decentralised hierarchical network that would spread from one central computer in Moscow through hundreds of regional computing centres all the way down to as many as 20,000 local computing centres.

Kerri Smith: Yep. That sounds a lot like an internet. So why was it never built? That question has preoccupied Ben Peters enough that he had to write a book to answer it.October 4th 1957 and the world's press announces the miracle of the age. The Russians have successfully launched the first satellite ever to circle the Earth and Sputnik hurtles its way into space to make a date with history that heralds the dawn of a new era.

Kerri Smith: The Soviet Union in the late 1950s was not a place that lacked scientific prowess. They'd launched the first satellite into space. They were doing world class theoretical physics and maths. They had nuclear power.

Benjamin Peters: Under certain conditions, the Soviet Union is this extraordinary pro-science superpower. It just becomes an interesting question then, why at the height of the space race and the tech race and why with all the motivations, means and mathematics to be able to build a computer network did the Soviet scientists stumble so badly?(Music)

Kerri Smith: It certainly wasn't because they didn't think to build an internet. For three decades from the late 1950s, various teams of scientists invented and then tried to pitch a nationwide network project to the government to streamline the economy and make it ruthlessly efficient. Their aim...

Benjamin Peters: … Grandiosely put: to upgrade Soviet Socialism into a higher form of communism by networking the economy itself. This was a fulfilment of Marxists prophesy.

Kerri Smith: The architects of these systems saw each factory, each productive unit of the economy, getting their terminal, where they would log their productivity onto a computer, which would connect to other nearby factories and towns.

Benjamin Peters: All the way up to the local city centre, all the way up to Moscow itself. Think of it as like a giant tech upgrade for what already exists.

Kerri Smith: In fact that was the very least that the scientists imagined their network would be. One man in particular had a whole lot of other ideas that went along with his vision. His project was the one that Peters mentioned at the very beginning: OGAS. His name was Victor Glushkov and he dedicated his life to his beloved network idea.

Benjamin Peters: Some of the stories are surprising. It's true that, for example, Glushkov collapsed after working repeated 20-hour days as the new director of the Institute of Cybernetics in Kiev in 1962 and then he in fact insisted on finishing a prize winning book on computer programming while strapped to a bed in the hospital.

Kerri Smith: Glushkov had a bold blueprint for OGAS full of futuristic ideas.

Benjamin Peters: Alongside the OGAS network, there were almost killer apps such as the following: online banking, a system of electronic virtual currency receipts, natural language programming, paperless office was repeatedly promised, mind uploading for people, individuals and collectives to be able to upload their memories and thus achieve a kind of virtual immortality.

Kerri Smith: Okay, so you're thinking no wonder nobody bought that. Even to a well organized, socialist would be utopia that sounds completely unrealistic but Glushkov was a practical man, happy to scale down to more doable ambition. At one point, instead of a nationwide scheme, he proposed a set of local area connections and he was able to realise some of those, but the forces against Glushkov's larger plan were too strong.(Music)Peter sees an irony when he compares Glushkov's defeated project in the Soviet Union where the success of what was to become the modern internet in the West.

Benjamin Peters: The first global computer networks take shape, thanks to collaborative institutions and state funding in the West, well, this contemporary project fall apart due to unregulated competition among bureaucrats and institutions in the Soviet Union In other words, you have the capitalists behaving like socialists and the socialists behaving like capitalists.

Kerri Smith: The reasons that the Soviets never got networked are not glamorous or cataclysmic.

Benjamin Peters: I think the simplest answer is institutional infighting. There was an unwillingness to collaborate at the level of agencies, across military industry and within the state itself and as a result Glushkov's project, however grandiose or scalable or pragmatic it might have been faced genuine and constant opposition at a number of different levels.

Kerri Smith: But the things that overcame Glushkov, self interest, clashing personalities, still defeat big projects today says Peters.

Benjamin Peters: I thought this work was about computer networks and it turns out to be about social networks. His story is humanising in the sense that I recognize in him, you know, something of the modern researcher or scientist who is seeking to bring about some big science, ambitious project and is simultaneously standing by it, not by the ideas, but by the institutions.

Kerri Smith: There's one question you really shouldn't ask historians, they really don't like it but I asked Ben anyway. Would anything have happened differently if the Soviets had invented the internet?

Benjamin Peters: I will venture a response. I think yes, it would've changed things. It seems to me that the Soviet Union collapsed largely due to internal economic reasons and if this central attempt to reform the economy technocratically would have succeeded then there would have been reason to imagine that the history would have been different.(Music)

Kerri Smith: That was Benjamin Peters whose new book is called How Not to Network a Nation. You can find it on Amazon and other good book selling places. Benjamin also writes blog at and you can tweet mailto:@bjpeter. Find a review of the book in this week's Nature,


Adam Levy: On the 25th of April 2015, a devastating earthquake hit Nepal killing thousands. Now, a year on, work is still underway to rebuild and repair the damage, but just because time has passed since the earthquake doesn't mean the danger has. Nature 532, 428–431 (28 April 2016)

Jane Qiu: Earthquakes can actually have quite long lasting effects.

Adam Levy: This is reporter Jane Qiu who spent March this year travelling in Nepal. I called her up to find out how the earthquake's effects are still being felt today.

Jane Qiu: So they can trigger a lot more landslides, years after an earthquake and sometimes even decades. So in China for instance after the 2008 Wenchuan earthquake, a lot of the new houses that were built after the earthquake were simply in the wrong locations. They were very beautiful, they were all quake proof but some of the houses were built right below some of the fragile slopes, so several years, probably even now, houses are still hit by landslides at a higher rate than before. So I think a crucial lesson from Wenchuan is that what happens after the quake is just as important as what happens during the shaking.

Adam Levy: So is that something we're currently seeing in Nepal, that there are still higher rates of landslides than there were before the quake?

Jane Qiu: Yes, I think the people are experiencing more landslides, especially during the monsoon season. So in the last monsoon for instance, in 2015, the landslide rates was about ten times higher than the average and it was quite a weak monsoon and it's a dry season, so it's not that bad now but in the trip I took to Langtang in Northern Nepal, along the valley so we keep hearing songs of rock falls and shifting slopes. So I think it's quite obvious that there's lot of instability in the system and things are coming down.

Adam Levy: When we talk about these rock falls, how big are we talking? Are we talking about a few boulders that could demolish maybe a house?

Jane Qiu: Yes, we're talking about massive, massive boulders. So we went to this town called Kodari, it had really bad rock fall during the earthquake. We saw gigantic boulders probably 10-15 metres high, they're everywhere, I've never seen such large landslide before. We saw amazing landslides that ripped entire slope open all the way from the top of the ridge to the bottom of the valley, so we actually tried to cross them and then we thought you know there's no way we can walk around it. I was quite nervous. I thought, oh I hope I'm not going to slip and sometimes because we keep having rock falls every now and then, a couple of times we had to run because we saw a few rocks coming down onto us, we had to run very quickly.

Adam Levy: What's actually the the mechanics of this? What causes this increased rate of landslides in the wake of an earthquake?

Jane Qiu: So before an earthquake, the subsurface materials are quite tightly packed and after a strong shaking they become quite loose, so they're expanding with holes and cracks in between, so now they're in a state that we can have more water penetrating into those holes and cause the slope to fail.

Adam Levy: What research is been done to look into these landslide risks in Nepal?

Jane Qiu: So, immediately after the earthquake a group of researchers rushed to install a series of seismometers and also other sensors to measure water flows and also weather stations. So they want to combine those data, trying to understand how the landscape evolved after the earthquake, so that they can also monitor the potential dangers.

Adam Levy: In the story you describe that they're really mapping the landscape in quite a lot of detail. How do you actually begin to do that?

Jane Qiu: So one thing scientists really need to know is how the landscape will change over time. So one way they can study this is to take high resolution pictures of the landscape every now and then so they can trace its evolution. So when I was in Nepal I was with a group of researchers that use drones to map the landscape.

Adam Levy: What are the questions that are still open that researchers in Nepal are hoping to answer?

Jane Qiu: So I think scientists are trying to understand how much rainfall can cause a slope to fail. So now they can actually monitor the propagation of rainfall into and through a slope so they can understand, they can trace actually, how the water is moving underground, they can link that process to the stability of the slope and then they can also link that to some of the precursor signals that they can detect that precede a landslide. They are trying to link all these processes so that they could forecast when and where a slope may fail.

Adam Levy: That was Jane Qiu who's based in Beijing and whose trip to Nepal was supported by a grant from the Pulitzer Centre. Read her full feature at

Kerri Smith: Still to come in the News Chat: dodgy duplicate images in biomedical papers and watching Darwin's finches evolve. Before that it's time for the Research Highlights with Noah Baker.

Noah Baker: A plant has been found to bleed nectar from its injuries. If a plant is getting munched by a herbivore, say a slug, it'll normally heal its wounds pretty quickly but the bitter sweet nightshade has a different tactic. It keeps its wounds partially open and secrets nectar. This recruits sugar hungry ants to the area which fight off the plant's pests. Unlike other plants that secrete nectar, the bitter sweet nightshade doesn't have special organs for the job. This could point to how these nectary organs first evolved. Check out the paper in Nature Plants. Nature 532, 416 (28 April 2016)Struggle to get a good night's sleep when you're away from home? Blame your left hemisphere. Researchers scanned people's brains while they were sleeping somewhere unfamiliar. They found that signals of deep sleep was stronger in the right hemisphere than the left. Playing sounds into the right ear of study participants which is rigged to the left hand side of the brain disturbed their sleep more, but these differences between hemispheres went away when their subjects got used to their new bedrooms. This suggests that their left hemisphere keeps watch in new and potentially dangerous environments. That study is in Current Biology. Nature 532, 416 (28 April 2016)


Adam Levy: Time now for our News Chat and Dan Cressey joins us in the studio. Hi Dan.

Daniel Cressey: Hello.

Adam Levy: So results have just become available from a big bio-study looking at a whole raft of different papers and what they're actually looking at was dodgy images of some kind. Nature

Daniel Cressey: So, this is a study showing that one in every 25 biomedical papers seems to contain an inappropriately duplicated image. This is in some ways the easiest kind of inappropriate image to spot in paper because you look at it and go that one looks like it's same as that one and they claim to not be the same, so there's clearly a problem there.

Adam Levy: So when you say inappropriately duplicated images, do you mean, people are copying each other's images?

Daniel Cressey: No this is within the same paper. So the same paper from the same team will have two images within that paper that are actually the same although they're listed as not being the same.

Adam Levy: That seems like quite a strange activity. It seems almost like plagiarizing oneself. Why is this happening?

Daniel Cressey: Well one of the big questions with this is what is actually going on here and what's the intention. So in some cases, this might be a researcher just straight out trying to commit fraud, not having found what they claimed they found and trying to fake that by saying that one of the images they have got is something else entirely to prove their point. Equally, it might just be a simple error; someone has just put the wrong image into their paper. So this study doesn't tell us about that particularly but it does mean that there are around 4% of biomedical papers, at least the ones that were looked at in the study, have a problem and that is maybe something that the community needs to address.

Adam Levy: Is this something that affects the entire community or are there some journals which are kind of doing a good job of stamping this activity out?

Daniel Cressey: There's certainly differences across journals, this study found. The authors looked at really quite a massive number of papers. They looked at over 20,000 research papers and in some journals there were 12% of papers that seemed to be problematic. So one in ten papers have some sort of issue with them whereas in other journals, some of which have screening programs to try and weed this stuff out before they publish, it was less than 1%, much less than 1%.

Adam Levy: It seems a bit like there shouldn't need to be a screening process. It sounds like the kind of thing you should just be able to see it, oh figure 3 is the same as figure 1. How come it's slipping through the net as it is?

Daniel Cressey: Well in some cases these things might be really obvious but in other cases, they're not necessarily and this is an ongoing issue and there's lots of discussion in the journal community and amongst researchers themselves about how much you can stamp this stuff out and how much you are ever going to be able to detect this. The other issue is if people really want to commit fraud then they're going to be really clever about faking images, you might not find it even in an analysis like this.

Adam Levy: So one in 25. I don't really have that much sense of how to place that. What's the reaction been to this? Do some people think it's very high or lower than they expected?

Daniel Cressey: The reaction depends to some extent on how much you've looked at this already. I think some people are very surprised by this and they say how can there possibly be so many problems that can be spotted in such a huge number of research paper and we're talking hundreds of papers just in the sample. Other people, the kind of people who have been saying for years the journals need to do more to deal with issues like this will be unsurprised and some of those people say that actually they think the true figure might be even higher.

Adam Levy: On now to a slightly less controversial biology story regarding Darwin's finches. Now before we look at what this study was concerning, just a brief history lesson. What are Darwin's finches? Nature

Daniel Cressey: Darwin's finches are a really iconic example of Darwin's theories. They're not actually as important to Darwin's creation of those theories as a lot of people believe. He barely mentions them in most of his writing and the current thinking is they really didn't shape his beliefs, but they're a really great example of what he was trying to say with his work because they have loads and loads of different sized beaks on different species, different sizes, different shapes and these beaks allow them specialize in eating different things. So what happens is the ancestor of these birds arrives in the Galapagos and as they spread out across the islands, they find different things that they can eat, be that insect or seeds and they adapt to those food sources by evolving beaks of different sizes and different shapes.

Adam Levy: Now none of that is really contentious. It's just a nice example of evolutionary history but in this paper though looking at something a bit more contemporary.

Daniel Cressey: Yeah this paper is great because researchers are now being able to not just observe these changes in the birds but to look at the actual genetics underlying them. So previously a team has identified the gene that controls the shape of these bird beaks and last week they announced that they have discovered the gene that controls the size of the beaks.

Adam Levy: But are the birds' beaks actually currently changing?

Daniel Cressey: Yeah and this is the other great thing that's described in this paper which is that they see this happening in action, they see this gene actually working. So they looked at a particular population which experienced a drought and what seemed to happen is that in one particular species, the birds with larger beaks, were less likely to survive. The birds with smaller beak can find really tiny seeds or something else to allow them to keep going and get through and afterwards the researchers looked at the genetic makeup of this particular group and they found a signature for this gene which they now think is important in controlling the beaks and they can see that change.

Adam Levy: So have they actually discovered anything new about the genetics of these finches or is it just our ability to watch these developments in real time?

Daniel Cressey: Well they have pinpointed this specific gene that they think is involved and as one of the people we quote in the story says, on the one hand it doesn't change anything but on the other hand it changes everything, because we're not just saying anymore we see this evolution happen but we're saying we can see it happen genetically and this is another great example of some people who say that evolution isn't true, you can never see it happen. It's like we do, we see often quite a lot and now we can even show you the genetic basis behind that change.

Adam Levy: Dan, thank you very much for joining us in the studio. If you want to read more about those stories and others of course, please go to

Kerri Smith: That's all for this week, tune in next time as we search out the dark reactions of chemistry.

Adam Levy: And if you enjoyed the show we'd love it if you dropped us a review on iTunes. It would do wonders for my self esteem. I'm Adam Levy.

Kerri Smith: And I'm Kerri Smith.