Ralph Merkle on Moore’s Law, Molecular Computing & The Singularity
Imagine molecular nanocomputers made out of diamond powering the computers of tomorrow and opening the door to atomically-precise materials & life-extending medicine. We’re joined by computing & nanotech pioneer Dr. Ralph Merkle, who’s developing the nanofactories that will make it a reality. Learn where his research is taking us, and why he believes the Singularity is imminent…
Ralph, you’re well known for your work in computing & cryptography, but you also have a true long-term passion for nanotechnology. I’d like to explore that a bit — let me start by asking your work in nanotech & molecular manufacturing as a senior research fellow at the Institute for Molecular Manufacturing (IMM). Can you tell me a bit about what your focus in that organization has been?
Well, we’ve been looking at a variety of applications for molecular manufacturing, and one of the things we’ve focused on is computation. If you look at the future of computing, it’s pretty clear we’re going to have a spectacular amount of computational power, and one example is the concept of a molecular mechanical computer.
Now, you might think that Babbage “is so very 19th century”, but the concept of the mechanical computer, if you scale it down, is really quite attractive because things move faster at the smaller scales, and energy dissipation goes down as well — so we’ve developed a concept for a molecular mechanical computer.
So for a molecular scale computer, the only parts are rods that are connected to each other by rotary joints — and the only energy dissipative element is rotation around the joints. This makes the energy dissipation of the computer very small, which means that you’re looking at 10 to 12 orders of magnitude less energy dissipation than existing computational models. So anyone who tells you that Moore’s law is coming to an end is unaware of the opportunities that are ahead.
What this means is that the generalization of Moore’s law to power computing is going to continue. There might be some little hiccups along the way as we transition from one technology to another, but keep in mind we’ve gone from relays to vacuum tubes to integrated circuits to large-scale integrated circuits so far — and if we transition to molecular computing, that’s just another technology along an exponential path towards ever more powerful computers.
At this point in time, the big limiting factor for computational power seems to be heat. So designing a molecular mechanical computer with very low energy dissipation for logic operations seems like the appropriate thing to do.
In addition to molecular computing, you’re also the co-founder of the Nanofactory Collaboration, which has the goal of creating a desktop manufacturing system capable of creating atomically precise diamondoid products. Can you tell me a bit about that?
Well, so far we’ve been talking about the theoretical benefits of building molecular computers, but from a practical perspective, the question is exactly how are we going to build these devices?
Today’s existing technologies for building molecular computers are in their infancy, but if you look at the kind of advances that are being made, in a timeframe of 10 to 20 years we’ll have the core technology to let us start building a variety of molecular devices, and to do them with atomic scale precision.
Nanofactories are a path forward to the actual construction of components on that scale, and then once you can build molecular machines, you can design them to build even more molecular machines, and then you’re off to the races.
So once we’ve got nanofactories working, it opens up wide variety of applications that you can build devices for. For example, there’s a whole line of research that says that if we had medical nanodevices, we could use them in your body to repair damage at the molecular and cellular level.
Today’s tools are simply too big to deal with this scale of problem, but if we begin designing tools now of the appropriate size and precision, then you could deal with the fundamental causes of disease, and cure cases that today would be considered simply hopeless. This would give medicine a whole new plane of capabilities to give us longer and healthier lives.
As I understand things, progress in nanotech has been difficult without having the self-replicating molecular assemblers that Eric Drexler envisioned. It sounds like your research is kind of trying to sidestep by using a different manufacturing process, is that right?
Well I think the focus with nanofactories is more on the overall manufacturing process to build products that are atomically precise. The assembler was an early approach towards these concepts, but it’s a concept that gives people the Heebie Jeebies, so we’re working on a different model which doesn’t make people feel nervous in order to move forward.
I think the nanofactory concept looks more like a box that manufactures atomically precise products, based on the concept of convergent assembly. This is the concept that small parts can be assembled into larger parts, larger parts can be assembled into yet larger parts — and in each stage of the assembly process you have a system that can take components of a specific size and put them together into larger parts.
I did a scaling study on this some years ago, and it turns out you can go from molecular scale parts up to meter-scale parts fairly rapidly, in the matter of a few hours. You start with molecular scale inputs, and build them up in size through a series of stages, and the result are outputs are macro-scale products built with atomic-level precision.
Being able to do this gives us some real advantages in terms of manufacturing — for instance, being able to create incredibly strong materials. For example, if I had a strut that was built with the strength to weight the issue of diamond, which is over 50 times that of steel, it would be very useful in the aerospace industry.
In aerospace, if you could improve the strength to weight ratio of the structural materials in rockets or the shuttle by a factor of 50, then you can reduce the cost of low earth orbit by a factor of a thousand to 10,000 — and you can do it using existing launch systems.
So that’s one area where it’s pretty clear a nanofactory offers great improvements to an existing industry, and developing it further is a very valuable project.
Now you’re also the Chair Emeritus for nanotechnology at Singularity University. Is that role also focused on computing side of things, or does it also include a lot of the materials-design goals you’ve just mentioned?
It’s all over the board , because Singularity University is interested in the future of pretty much everything. I’m a component of that because I’m interested in molecular manufacturing, medicine, and computing. However, Singularity University also wants to know about the future of the internet, 3-D printing, etc. You name it, they are into it.
Singularity University all also looks at the challenges facing the world — like the list of major global challenges that the UN keeps. For example, how do we provide clean, potable water to be the world? This non trivial problem in a lot of places around the globe, so what’s the best way to do that?
As it turns out, nanofactories are a solution to that problem as well. If you can build an atomically-precise membrane which is very thin, with pores in it that pass water but not contaminants, then suddenly you have the ability to purify water at a large scale. What we’re talking about here is building up a device that allows seawater or grey-water to enter one side, and pure water to come out the other, and it can be done at very low cost.
What are your thoughts about the Singularity itself? I guess I’m asking we’re all really just guessing to one degree or another, but based on your experience & knowledge, do you think it will actually happen?
Well, just take Moore’s law, along with what I’ve said about computers getting 10 to 12 orders of magnitude more efficient — and then draw out some straight lines to 12 orders of magnitude on a semi-log paper for Moore’s law, and then ask yourself, “okay, if we’ve got computers with this much more powerful than what we’re currently using, what does that let us do in terms of general AI processing capabilities?”
I’m not even going in to where the is software is going — whether it be neural nets, deep learning, or anything else. I’m just pointing out that we’re going to have a lot more hardware capability which will lead to a lot more computing power for all the advances in software that continue happening.
If you want to call that a singularity, then okay, where does that leave us — as good old biological humans. Well, I’d like to be in a situation where we have these very smart computers provide us with intelligent guidance and help along the way, if we can structure it that way. That still leaves a lot of debate about whether or not we can structure it like that, but I like to think we can.
Do you believe artificial intelligence and the Singularity are a threat to be avoided, or an opportunity to be pursued? It seems like futurists are to be polarized on this subject, at least at the moment — with Musk, Hawking and others being very pessimistic, and others like Kurzweil promoting the benefits that these technologies can bring. What’s your opinion on this?
Well, it’s coming. I mean, what, you’re going to announce that people aren’t going to build more powerful computers? How do you turn this process off? Go to Intel, Samsung, Nvidia, and all the rest and tell them to quit building more powerful components for computers, too?
We’ll also have to tell other nations that they should stop building more powerful computers, too. We’ll go to the Chinese and say, “Hey, you guys — you’ve got stop building faster machines. We want you to slow down.”
We’re on the rollercoaster and it’s moving, and it’s going to carry us forward. At this point we need to figure out how to guide the process and move the most sensible direction — and I think we can do that.
There’s a lot of discussion right now about whether we should or shouldn’t develop all these technologies that will lead us to the Singularity, but the decision has really already been made. We’re going to do it. So then the question is how to do it in a way that makes the most sense for humanity. Basically this is what Ray Kurzweil is saying — we’re heading towards this technology, let’s figure out how to benefit from it.
Are there risks? Sure there are risks with any new technology, but let’s figure out how to develop this technology to avoid the risks and claim the benefits, which is what humanity has always done.
I mean come on, fire, is a risky technology — it can burn down whole villages. Fire is not something you want to deal with without understanding the risks. But we have fire stations, we have fire extinguishers, and we’ve developed a whole set of technologies for dealing with the risks. We need to do the same thing with AI so that we understand the risks, benefits, and have a plan in place to benefit from it, because it’s not going away.
Ralph, thank you again for your time, and let me close by asking you what’s coming next for you personally — what are your plans & vision for the future?
Well, I try to stay happy and keep things rolling along. We’re going to have a very interesting future. I think one of the things that I find very interesting at the moment are some of the issues we’re seeing with governance — I think that’s very important right now.
I think everyone will agree the governance that we see in the world today is perhaps not as good as it could be. That raises the question of how we create a model for governance of higher quality — both for nations and organizations, and it gets into what I’ve been calling DAO democracies, where DAO stands for Decentralized Autonomous Organizations.
For quite some time now, lots of folks have said that democracy as we know it is the worst form of government except for all the others, right? So I started asking what it would look like to develop something better than democracy — taking advantage of everything we’ve learned over the past century.
So a DAO democracy actually does maintain the benefits of traditional democracy, but also blends things we’ve learned from the wisdom of crowd knowledge into a larger governance model aimed at creating the best solution, rather than falling to the least common denominator.
I look at this idea in particular as something that might provide some useful, building blocks for how people can more effectively organize into groups and make decisions. The way we do it right now isn’t optimal, so this is definitely an idea that may offer insights into what what’s coming next.
About Our Guest
Dr. Ralph Merkle has been a Distinguished Professor at Georgia Tech, and is currently a Senior Research Fellow at the Institute for Molecular Manufacturing, in addition to being a Chair Emeritus at Singularity University, a Director at Alcor, and Co-founder of the Nanofactory Collaboration. He has also been honored to be a recipient of the 2010 IEEE Richard W. Hamming Medal for outstanding achievement in information science, in addition to a 2011 Fellow at the Computer Science Museum for his work on public key cryptography.
Learn more on his website at: http://www.merkle.com/
