Host: Shamini Bundell
Welcome back to the Nature Podcast. This week, we’ll be hearing about the logic of quantum computing…
Host: Nick Howe
And learning about how labs recover from unexpected disasters. I’m Nick Howe.
Host: Shamini Bundell
And I’m Shamini Bundell.
[Jingle]
Interviewer: Nick Howe
To me, computers are basically magic. I’m talking into a microphone now, my speech is going to be edited and then broadcast to the world through the air, all thanks to computers. But how do they actually work? Well, computers use logic, specifically logic gates. These gates take simple inputs and turn them into one of two outputs, a ‘1’ or a ‘0’. Putting loads of these gates together makes circuits, and circuits allow complex inputs and outputs. In the real world, that might be something like show me cute cat videos translating to watching a cute cat video. But computers do have their limits. Enter, quantum computers.
Interviewee: Michelle Simmons
A quantum computer is, at the moment, a theoretical idea that if you could encode information in quantum states, there’s a predicted exponential speed up in computational power for certain types of problems.
Interviewer: Nick Howe
This is Michelle Simmons, a quantum computer researcher. To get these benefits in a quantum computer, you need to make a quantum logic gate, composed of quantum bits or qubits. A quantum logic gate also has some advantages. Whilst it still takes simple 1 and 0 inputs and also outputs 1s and 0s, in between it computes everything up to and including 1 and 0, as the qubit can be both at the same time, thanks to all the, well, quantum-ness. Effectively what that would mean is certain calculations can be processed more quickly – ergo, exponentially faster computer. Now, making a simple quantum gate based on one qubit has already been done. But to get to really complex computations that we use daily, scientists also need to have gates based on two qubits. And that is a real challenge, as it requires them to entangle the qubits. Entangled qubits affect each other – the state of one impacts the state of another. But entangled states are delicate, and to further complicate things, in order for the gate to work properly, the entanglement needs to be able to be turned on and off. Whilst this has been achieved by some quantum systems, in silicon – Michelle’s material of choice – it’s proved difficult. The secret to making all this work, according to Michelle, is getting the qubits incredibly close together.
Interviewee: Michelle Simmons
And that really means, in our system, getting them down to about 12 or 13 nanometres apart.
Interviewer: Nick Howe
Whilst making a two-qubit gate is hard in silicon, the material has tightly packed electrons which can help ensure stable entangled quantum logic gates.
Interviewee: Michelle Simmons
If you look at all the electrons that are in silicon, they’re all used to form the silicon lattice itself, and so that means that there’s nothing in the silicon crystal that can interfere with your qubit’s states because all the electrons are tightly bound. So, what we’re getting here is these beautiful, fast qubits in a nice, stable environment.
Interviewer: Nick Howe
Moving these qubits around with mega precision is very difficult, so to get the qubits very cosy, Michelle adapted a technology used to image surfaces at the atomic level – scanning tunnelling microscopy (STM). Michelle used STM to draw a pattern onto the silicon with hydrogen atoms, creating a sort of stencil which in turn guides qubit placement with atomic precision. The resulting two-qubit logic gate is not only functional, but accurate and very fast. Previous two-qubit silicon gates have been less accurate, slower, and more likely to lose information. With this new gate, Michelle thinks quantum circuits are within reach.
Interviewee: Michelle Simmons
So, when you’re building an actual computer chip, you need to be able perform what they call a universal set of gates – you’ve got to be able to read out and initialise the information into the computer and then you’ve got to do what we call single-qubit gates and then two-qubit gates. And if you have the ability to do all of those things – read-in/read-out, single-qubit gate and two-qubit gate – then you can actually design the actual chip to run as a computer. And so, over the last decade or so, we’ve shown that we can read in and read out with very high accuracy. We’ve recently demonstrated a single-qubit gate and now we’ve demonstrated a two-qubit gate.
Interviewer: Nick Howe
What’s more, because all this has been done in silicon, it means that Michelle can lean upon the decades of experience the tech industry has making conventional silicon computer chips. She thinks this will allow the quick scaling up of quantum computers, with more and more qubits. But how much scaling is needed is still unclear. Right now, Michelle’s team are working towards a ten-qubit system.
Interviewee: Michelle Simmons
So, at the end of the day, people are still arguing theoretically about how many qubits do you need to prove that a quantum computer can outperform a classical computer, and the kind of general view at the moment is it’s between 50 and 100 qubits. But part of the challenge is if you have a small number of qubits, you can see how well they behave because you can measure them very well, but if you go to too many qubits too quickly, you can’t measure each individual qubit and prove that it works. And so, our approach over the next few years is to make sure that every time we add a qubit, we understand exactly what happens to the existing qubits so that we can maintain the quality of each one as we add the others. So, we’re focusing on ten qubits in the short term because we know that if we can do that well, then we should be able to expand that out to larger systems.
Interviewer: Nick Howe
Of course, according to lots of researchers, quantum computation has been just around the corner for quite a while now. But Michelle is confident that in the next couple of decades we could be seeing a whole world of unknown quantum opportunities arising.
Interviewee: Michelle Simmons
One of the fascinating things that I’ve enjoyed is looking back through the evolution of classical computing, and seeing how we started off with a single transistor which was used pretty much as a hearing aid. Then we got the integrated circuit which is eight components on a chip, but it was another five to ten years before we actually got commercial products from that, and they would be things like calculators and transistor radios, where we started to use those integrated silicon chips at the time. I think in the quantum computing world, it’s kind of plotting a similar pathway but it’s also different, and so we can see how we can build larger and larger quantum systems, but the applications that we expect to have will be much greater than we have now because we’re still learning them as we go.
Interviewer: Nick Howe
That was Michelle Simmons of the University of New South Wales in Australia. You can find her paper over at nature.com. If you’re eager to learn more about all things quantum, then head over to the Nature Video YouTube channel and type quantum in the search box and you’ll find some videos just for you.
Host Shamini Bundell
Coming up at the end of the show is the News Chat, where we’ll be finding out the latest dinosaur news. Coming up now though, it’s the Research Highlights, read this week by Noah Baker.
[Jingle]
Noah Baker
For the first time, a computer has taught itself to efficiently solve a Rubik’s cube. Solving the colourful cubes has been a favourite pastime since their invention back in the 70s and there are various well-known algorithms that people and computers alike have been using to beat the box for years. But building a computer which can work out how to solve the cube on its own, well, that’s a much tougher challenge. A Rubik’s cube has over 43 quintillion possible combinations and only one solution. Most AIs develop this kind of puzzle-solving strategy using trial and error – essentially, they try lots of options and learn which approaches lead to success. But the cubes have so many options, it’s really unlikely the algorithm would ever hit upon the right solution by chance, which rather scuppers that plan. Now, a deep neural network is taking a different tack and working backwards. By starting with the solution and messing up the cube into a random configuration, the AI can observe how closely that configuration and the solution are related. It can then repeat that process for a vast number of configurations and fold all that information in together to come up with optimal moves to solve the puzzle. After training, the machine could solve every cube it was presented with, and in more than 60% of cases it could do it with the smallest possible number of moves. The programme could be applied to other problems with mind-boggling numbers of solutions. Decode that study over in Nature Machine Intelligence.
[Jingle]
Noah Baker
Anthropologists have used the ancient teeth of ritual sacrifice victims to explain a historical population boom in the Mayan city of Chichén Itzá. Chichén Itzá Is located in what is now Mexico, and was founded somewhere around 800 AD. In the following two centuries, it’s population grew to nearly 50,000 people – a veritable metropolis at the time. To find out more about Chichén Itzá’s population, researchers used teeth found in a nearby sinkhole known to contain the remains of adults and children ritually killed in the city. Chemical analyses revealed that many of the victims were born outside of the city, some up to 1,000 kilometres away. This hints that immigration from distant lands could have helped fuel the ancient population boom. Read more on that study over at the American Journal of Physical Anthropology.
[Jingle]
Host: Shamini Bundell
It’s been almost a year since a fire in Brazil’s National Museum destroyed one of the country’s largest and most valuable scientific collections. Countless precious artefacts were lost in the blaze and the lifework of dozens of researchers went up in flames. Reporter Anand Jagatia has been speaking to scientists affected by disasters like this one about their experiences and how they managed to rebuild their work after the dust settled.
Interviewer: Anand Jagatia
Paulo Buckup studies the evolution and diversity of fishes at the National Museum of Brazil in Rio de Janeiro, where he’s been a professor for the past 25 years.
Interviewee: Paulo Buckup
On the night of September 2nd, I was at home when I started to have news through the social networks that there was some smoke and fire in the museum, and the first thing I thought was I must get there and try to help as much as I can.
Interviewer: Anand Jagatia
When he heard that the museum was ablaze, he jumped into his car and drove straight to the scene.
Interviewee: Paulo Buckup
The fire went through the roof and was spreading to the adjacent hallways. It was a kind of very eerie, odd experience because it was night and the fire was just yellow, red, illuminating the sky with lots of sparks. Book pages were flying to the sky. The smell was the worst part. It was stuck to my skin and hair for several days and probably on my imagination for even longer because being there and what it meant made it even worse.
Interviewer: Anand Jagatia
Brazil’s National Museum is more than 200 years old and housed a collection of over 20 million objects, including mummies, animal specimens, fossils and archaeological artefacts. Several factors made that night particularly devastating. The floors of the old building were made of wood. It had no sprinkler system, and when the firefighters eventually arrived, they didn’t have any water because the hydrants were dry. Tens of thousands of the priceless items inside would eventually be consumed by the fire, but Paulo and his colleagues broke into the burning building through a closed entrance and tried to salvage whatever they could.
Interviewee: Paulo Buckup
We saved quite a bunch of molluscs, including specimens that are like the identity of described species, and the computers that had lots of data about the collections, some high-end equipment. But the most devastating thing was the impact on people that work in the museum, especially those scientists that used to work in that building.
Interviewer: Anand Jagatia
Some of those scientists lost their entire life’s work.
Interviewee: Paulo Buckup
Whole lives of collections amassed but it was even more than that, some of them lost their personal documents, documenting everything they’ve done. Even their diplomas were burnt there, and that’s a devastating experience for a researcher. I’d say that everybody I’ve seen here cried, from the doorman to the most tough scientist. Everybody cried at one moment or another. Some cried in front of the flames. Some cried several months later.
Interviewer: Anand Jagatia
Nobody when they begin their career as a scientist envisions having to deal with a disaster on such a scale. Catastrophes like this one are rare, but they do happen. Back in 2012, 4,000 miles away from Rio, Kathryn Moore was researching at NYU Medical School. She had an active lab of around 14 people who were studying cardiovascular disease and obesity in mice.
Interviewee: Kathryn Moore
In the midst of all this, we were hit by Hurricane Sandy. We lost power. There was also flooding of the building which resulted in the loss of the animal facility. Initially, we couldn’t enter the building to continue working and then once we could, we realised that we had lost a lot of our samples that were in the freezers and all of our mice, and this had a major impact on all of the projects that had been ongoing.
Interviewer: Anand Jagatia
When Kathryn and her colleagues were eventually allowed back into the lab, it wasn’t in a good way.
Interviewee: Kathryn Moore
Things were a mess. You could still smell the residual smell of diesel fuel that had come from the generators in the basement and there was a lot of clean up that had to be done, and that initially was good. It took our minds off of the fact that we had lost so many mice and ongoing experiments and samples and it was really after the first few days of busywork that it began to sink in that we were going to be facing a major uphill challenge. And so, several postdocs went to work in other labs of friends of mine or colleagues of mine that were generous and made offers to have people come and work there.
Interviewer: Anand Jagatia
Some of her students went to Massachusetts General Hospital, others to nearby Columbia University, so they could continue with their research. Similarly, in Brazil, some of the students at the museum were able to relocate to other institutions who offered their support during the crisis, and despite everything they had lost, the staff were able to get up and running in some capacity almost immediately.
Interviewee: Paulo Buckup
In less than a week, we were able to receive our colleagues so that they could have minimal conditions to continue to work and to restart the collections and research, and that was really important because we need to avoid a tragedy that would be even larger than the loss of the building and what was inside the building, and this tragedy would be the loss of the brainpower of our scientists.
Interviewer: Anand Jagatia
State representatives set up a fund to ensure the reconstruction of new buildings and Paulo says that they have received a lot of help in the wake of the tragedy, including generous donations of collections. But the hardest work is still to come, as the museum tries to painstakingly rebuild an inventory that once contained two centuries of science and culture which is now lost forever. In New York, almost seven years on from Hurricane Sandy, Kathryn reflects on how her lab managed to survive.
Interviewee: Kathryn Moore
I think I tried to focus on the silver linings. I like to think that Hurricane Sandy also provided us with an opportunity to change research directions. It made the team much stronger and more closely-knit and it also taught me about how I want to run my research group going forward.
Interviewer: Anand Jagatia
And back in Rio, Paulo says that nearly a year after the fire, things are going better than expected given the magnitude of the tragedy. The main immediate difficulty is the need for new buildings so that scientists can have a space to work and the new collections have a space to grow. But even so, there are reasons to be hopeful.
Interviewee: Paulo Buckup
Well, the main thing is the museum is alive – that is our motive since day one. We have our hashtag and everybody carries it in their heart, and in Portuguese it’s #museunacionalvive, and that translates, we’re still in business, we still keep our working capacity, we still have our staff, we still have lots of collections, we still have our main library, so I’m optimistic.
Host: Shamini Bundell
That was Paulo Buckup of Brazil’s National Museum and the Federal University of Rio de Janeiro. You also heard from Kathryn Moore of New York Medical School in the US. To read more about how Brazil’s National Museum is doing one year on, head over to nature.com/news. And if you want to make sure you’ve prepared your own lab for disaster, there’s an article over at nature.com/careers that you can check out.
Interviewer: Nick Howe
Finally then on this week’s show, it’s time for the News Chat. This week, I’m joined by Richard Van Noorden, Feature’s Editor here at Nature. Hi, Richard.
Interviewee: Richard Van Noorden
Hi, Nick.
Interviewer: Nick Howe
Thanks for joining me. Richard, our first story this week is about one man’s campaign to open up the world’s research. What can you tell me about this?
Interviewee: Richard Van Noorden
This is an American technologist called Carl Malamud and he’s spent his whole life, really, trying to liberate information that’s locked up behind paywalls and copyright, and he’s spent most of his career doing that for copyrighted legal documents and government codes that he says should be made open. He’s been quite successful in arguing that. Now he’s turning his attention to scientific research, and he thinks he has a way to make the world’s copyrighted paywalled research papers open legally, in a way. What he’s planning to do is to allow computer software to mine the text and images of research papers to pull out insights, but not that the papers themselves be able to be downloaded or read by people.
Interviewer: Nick Howe
Right, well, I suppose that is the first and most obvious question – is it legal?
Interviewee: Richard Van Noorden
Well, lawyers don’t know. What he’s doing is he’s building this data store in India. He’s got 73 million papers. He’s currently extracting the text and the images from these papers. They’re held on a 500-terabyte server in Delhi and he is going to allow scientists to write software to extract information from this text. Now, what he’s relying on here is two things. First of all, he’s saying this kind of data mining is called ‘non-consumptive’, a technical term meaning that you can’t read the original papers, and he’s saying well, in the case of Google Books, Google scanned a lot of library books and then they presented snippets of those books to the public, and there was a lawsuit about that. Publishers said hang on a minute, these books are copyrighted. But in the end, Google won that case and as a result, we are now allowed to search books through Google Books and we’re allowed to see snippets but not read or download the whole book, so that’s an example of successful non-consumptive mining under American law. Secondly, this data store that he’s building is in India, and India has a very interesting copyright law and there’s one particularly inspiring example for Malamud, which is the case of a photocopy services shop which was photocopying text books so that students could get what they needed for their courses. And publishers sued this shop but it won the lawsuit, and the court said this is because of a provision in India’s copyright law that says it is allowable to photocopy snippets of work for educational purposes, and there’s another bit in the same law that says for research purposes. So, Malamud is saying well, if that’s allowable then this should be allowable, again, it’s snippets of research papers for research purposes. Thing is, lawyers are a little bit unsure about both of these arguments. For one, in order to create this enormous data store, he’s had to get these papers from somewhere and he’s being a bit cagey about where he’s got them from. We don’t know. None of this has ever been tested in court and we asked publishers about this and most of them that we asked said that this was the first time they were hearing of this, and they basically said it’s their view that researchers need to ask for permission to do this kind of text mining and Malamud says he doesn’t have this kind of permission, so he’s really testing the boundaries of what’s possible.
Interviewer: Nick Howe
So, publishers may not be too pleased about this, but what’s been the response from researchers themselves?
Interviewee: Richard Van Noorden
Well, the researchers we spoke to are really excited. In the US, in the UK, in India, they’re all planning to use this data store to build up new databases of scientific content and even to mine this data depot just to map how academic publishing has evolved over time.
Interviewer: Nick Howe
I guess we’ll have to wait and see what happens with this then, but moving on to our second story, this is about new research looking at fossilised dinosaur eggs. So, Richard, what’s new in the world of dino eggs?
Interviewee: Richard Van Noorden
Well, what’s exciting here is it looks like we’ve got proof that dinosaurs essentially were quite gregarious. They came together and they made nesting colonies that together protected all of their eggs. Now, you might think don’t we know this already – palaeontologists have unearthed lots of fossilised eggs and nests that cluster together – but the problem is, how do you know that all the eggs were laid at once and the dinosaurs were all there as a colony, like some modern birds and crocodiles, rather than being quite solitary, sort of nesting on their own, burying their eggs and then going away again. But this new find seems to nail the case.
Interviewer: Nick Howe
And how does it do that?
Interviewee: Richard Van Noorden
Well, it’s a nest site in southeast Mongolia in the Gobi Desert and essentially, the fortuitous thing about it is it appears that a flood from a nearby river covered all of the eggs that are found at this site in one go because there’s a sort of 300-square-metre formation of rock, there’s layers of orange, layers of grey. In between these bands there’s a streak of bright red rock with about fifteen clutches of fossilised eggs, some of them hatched and filled in with the red rock, some of them not. And the bright red line appears to have come from the flood, so essentially, it seems like all of these eggs, all of these clusters, were laid and hatched at the same time, that the dinosaurs laid them in a single breeding season. So, pretty impressive, and it looks like the dinosaurs involved, based on what the eggs looked, like were theropods, which could be anything from velociraptors to Tyrannosaurus. So, they haven’t really narrowed it down that much, but it does perhaps suggest that this colonial nesting idea arose during the time of the dinosaurs and that this has then carried on over to modern birds and crocodiles.
Interviewer: Nick Howe
So, why might they nest like this? Why nest together?
Interviewee: Richard Van Noorden
Modern birds and crocodiles build their nests and lay their eggs in a communal area when they breed, and essentially, that’s just a behaviour that deters predators because the predators see a large group of adults so they’re less likely to try and sneak in and take an egg. One slightly unclear thing is how successful were these hatchings, and they reckon that about half of the nests had at least one successful hatch, looking at the number of broken eggs, and that actually mirrors the success of modern birds and crocodiles that guard their nests. But that’s still a bit uncertain because, some people point, out it’s quite hard to tell the difference between a hatched egg and an egg that got broken by some outside source like a predator.
Interviewer: Nick Howe
Well, Richard, I’ll be excited to hear about more updates from the world of dinosaur eggs. Listeners, if you want to hear more on those stories, head over to nature.com/news.
Host: Shamini Bundell
And that’s all for this week. If you’d like to get in touch with us, don’t forget you can tweet us (@NaturePodcast) or if you want even more than 280 characters to get your point across, then send us an email – it’s podcast@nature.com. I’m Shamini Bundell.
Host: Nick Howe
And I’m Nick Howe. Thanks for listening.
Explosions, floods and hurricanes: dealing with a lab disaster
The plan to mine the world’s research papers
