In this video I want to explain a little bit about the background of my Frith story series. But first, a quick recap.
In 2009 my protagonist, Auron Blake, was working on an AI system intended for scientific work when he had the idea to ask it to design a more intelligent version of itself.
It succeeded beyond his wildest expectations.
His new machine was able to terraform distant planets and open interplanetary portals.
Unfortunately this led Auron and his friend and colleague Jer Darby to become marooned on a terraformed alien planet, which Auron named Frith, Frith being an Old English word meaning peace, protection, safety, or sanctuary.
Auron had to build a new AI from only mud and rocks in order to escape, and Frith turns out to be no sanctuary.
Eventually he succeeded, but Auron and Jer discovered that while they had been away, a nuclear war had devastated the Earth.
In the process of trying to return to the Earth, Jer was killed.
Auron decided to continue to improve Frith and to rescue people from the Earth and take them there.
He built a new civilisation.
But he was consumed with guilt over the death of his friend, and eventually decided to travel backwards in time with the aim of preventing both Jer’s death and the nuclear war.
His time machine didn’t quite work properly and he found he had arrived on the Earth three years too early, in 2006.
Now he’s living somewhere near York, England, essentially omnipotent and immortal.
But a strange signal is attempting to interfere with his machines, and a blogger called Daisy is becoming suspicious over who Auron actually is.
On top of this, he has one particularly thorny problem. Auron isn’t sure what will actually happen when he finally attempts to interfere with the Earth’s past. Is it even possible to change the past?
For the past three years I’ve been writing one story a week, almost every week, and turning it into a YouTube video.
Originally my idea for the channel was to create sci-fi horror stories somewhat in the vein of Victorian or early 20th-century science fiction horror stories, with plausible science and plausible protagonists, and by and large I think I’ve mostly stuck to that idea.
In cases where I make use of hypothetical scientific principles, for example involving time travel, I at least want to try to use a little logic in exploring what these imagined innovations might or might not be capable of.
It bothers me when I encounter science fiction where the scientific principles involved seem utterly implausible and the characters involved with scientific innovation just don’t seem like the kind of people who might actually be involved with scientific discovery.
These days you can find all sorts of people calling themselves scientists and claiming all sorts of discoveries, but when you look back at history, the kinds of people who made genuinely impressive discoveries; testable discoveries, they were, I think, generally pressed from a certain mould.
Most of them were a little unhinged and generally they weren’t interested in much other than science.
Most of them, although there are certainly exceptions, weren’t chasing glory or wealth. They were chasing knowledge.
I try to reflect this in my stories.
Most of the time, serious scientific innovation has also not been without its downsides. It’s almost hard to think of a really big scientific discovery that hasn’t led to something horrific somewhere along the way.
We live in a time when there is a lot of talk about an AI singularity. The idea behind the AI singularity is as follows. At the moment we’ve got chatbots that, however prone to mindless stupidity, are sometimes capable of making interesting discoveries. For example, a system called AlphaFold can successfully predict the structure a protein will fold itself into on the basis of the chemical structure of the protein. This is a problem that biologist were puzzling over for fifty years.
The question arises, what if AI advances to the point where AI can tell us how to build a more intelligent version of itself? Or what if a machine of some sort, controlled by AI, can actually fabricate a new, more intelligent AI?
It’s completely possible that we may actually be faced with this eventuality in the near future, and it seems as though this situation might very well lead to a kind of runaway intelligence.
The machine devised by AI could create an even-more intelligent version of itself, and so on, until we end up with machines of staggering, unfathomable intelligence.
This is something I wanted to explore in my Frith series, but with a twist. To explain the twist I need to tell you about a little-known relatively unexplored avenue in the history of computing, and we also need to touch on a closely-related set of interesting questions that span philosophy, mathematics and physics.
In 1928 a man named Gordon Pask was born in the county of Derbyshire, in England. In fact, he was born in Derbyshire’s main city, Derby.
There are no monuments to Pask in Derby. In fact, you could easily live in Derby without ever realising that Henry Cavendish, the discoverer of hydrogen, is buried in Derby Cathedral. Derby has a street named after Lara Croft, the fictional character from the game Tomb Raider, which was created in Derby, but Derby doesn’t seem particularly keen on remembering people from earlier times.
Having said that, there is a street named after Frank Whittle, inventor of the turbojet engine; Rolls-Royce in Derby built Whittle’s earliest engines.
I find this relative lack of history in Derby a little strange, because it was from Derbyshire and the surrounding areas that the industrial revolution began and spread, changing our world almost beyond recognition.
But Gordon Pask was not an industrialist. In fact, he has often been described as something of a dandy.
What he discovered isn’t now regarded as even particularly all that significant, but here’s where I wonder if computing couldn’t have taken a different path at a certain point in history; a path that it might still end up taking, even now.
Around 1958, working in his flat on Baker Street, London, together with another man named Stafford Beer, Pask created something quite remarkable.
Pask and Beer were able to grow a device that was capable of learning. One iteration of this device became known as Pask’s Ear: it could distinguish between different frequencies of sound. Other devices grown by Pask could detect magnetic fields or differences in acidity in a solution.
One of Pask’s principles was that his devices should decide for themselves how to interact with the world. He didn’t tell them how they should perform the tasks he set them; he only provided feedback to tell them whether they were on the right track or not.
In this respect, Pask’s devices resemble modern digital artificial neural networks, which are trained via feedback. But Pask’s devices weren’t digital. They were grown in tanks.
Unfortunately most of the precise details of how Pask’s system worked have been lost. It seems as though Pask was as much interested in art as in science, and he never seems to have regarded his discoveries as particularly important, and perhaps they weren’t. Or then again, perhaps they were.
Pask’s Ear seems to have consisted of a solution of ferrous sulphate—a chemical easily purchased from almost any decent garden centre—through which he passed electric current. This caused iron wires to grow in the solution, and it’s these growing and developing iron wires that enabled the device to learn.
The chemical basis of this is well understood. An electric current passed through a solution of an iron salt will tend to pull iron out of the solution, depositing it at the cathode.
To get actual wires to grow is a little more tricky. I have carried out some experiments myself and, at low voltages, ferrous sulphate doesn’t easily form wires or dendrites, but rather just clumps.
In this context, by the way, dendrites are metallic growths named after their resemblance to tree branches, déndron meaning tree in Greek.
Some metal salt solutions do form interesting dendritic patterns at low voltages, notably tin and silver salts, but not iron.
I suspect Pask used fairly high voltages, but that information appears lost.
When Pask’s systems did what he wanted them to do, he rewarded them with more electrical current. When they failed, they received a lower current.
The process was a bit like rewarding a dog with treats when it does what you want.
Considering this took place in 1958, when computers existed but were certainly not ubiquitous, you might wonder why Pask didn’t persist with this interesting research.
I think several factors were at work there.
For one thing, ferrous sulphate solutions, or electrochemical systems in general, aren’t very convenient to work with. Consider another kind of electrochemical system, the battery. From its invention in 1800 by Alessandro Volta, to the creation of the modern lithium-ion battery in 1985, we have a span of nearly 200 years of development and experimentation.
In the case of the battery, the internal growth of wires or dendrites is considered a big problem, leading to many terrible fires, and vast amounts of time and money have now been spent not on growing and training dendrites, but on suppressing them.
Another factor was perhaps Pask’s personality and interests. Pask doesn’t seem to have had any kind of passionate drive towards creating intelligent machines. His Ear was just one of many of his projects, spanning art, science and philosophy. He was a kind of “philosopher mechanic”.
It wasn’t as though no-one thought this kind of technology promising. In the 1950s and 1960s, there was quite a bit of talk about self-organising systems, although I can’t find any evidence that anyone apart from Pask and Beer actually created such a system. Most of the work done on self-organising systems was theoretical.
An exception to this was the iron-wire model of the neuron, which was popular for a time, but this idea was never developed into any kind of neural network, capable of performing calculations.
Let me read you a quote from R. M. Stewart’s 1969 paper, Electrochemically active field-trainable pattern recognition systems.
“We may as well start with the notion that with 10 000 000 000 parts per cubic foot (approximately equal to the number and density of neurons in the human brain), there will be no circuit diagram possible, no parts list (except possibly for the container and the peripheral equipment), not even an exact parts count, and certainly no free and complete access with tools or electrical probes to the ‘innards’ of our machine or for possible later repair ..... We would manufacture ‘logic by the pound’, using techniques more like those of a bakery than of an electronics factory.”
Here you can see that R. M. Stewart among others was looking forward to a future where intelligent machines were grown like brains. The parallel was explicit: after all, the most intelligent entities in the universe that we actually know of directly are human brains, and not, as yet, digital computers.
And yet, at the same time that the first very tentative steps were being taken towards growing self-organising intelligent systems, another revolution was taking place, and it simply eclipsed everything that was happening with analog computing at every stage.
The transistor was invented in 1947. At the time, vacuum tubes, also know in Britain as valves, were the backbone of electronics. Within 20 years, semiconductor technology had taken over.
The world already had a working technology for amplifying and modifying electrical signals: vacuum tubes, and this technology was simply displaced by semiconductors, which were far smaller and far more energy efficient.
The idea of building an analog computer by growing wires in a solution of metal salts was something that was just never really explored.
There have continued to be periodic forays into the idea of growing computers over the years.
In 2010, Alfred Hübler’s team at the University of Illinois discovered that ball-bearings floating on oil, under the influence of a strong electric potential, would self-organise into wire-like structures.
In 2021, a team at the University of Nanjing claimed to have created a system using silver nanowires that could recognise handwritten digits from a standard AI training set.
But on the whole, research into the potential of electrolytic systems similar to Pask’s was just never really pursued.
In my story, Auron Blake’s conventional AI system tells him to create such a system, and it turns out to be capable of far more than the digital system that invented it.
Obviously this is fiction, and almost pure fantasy, but there are reasons why I thought this to be more plausible than it perhaps initially sounds.
The fact is that a digital system cannot fully replicate, or predict the behaviour of an analog system.
Consider for example, the three-body problem.
Isaac Newton showed back in 1687 that mathematics can be used to predict orbital motion.
Orbital motion involves two bodies: for example, our sun, and a planet. So this is a two-body problem.
But if you add another body in there, so that three bodies are all orbiting each other, you get mathematically chaotic behaviour; that is, complex behaviour that depends on tiny variations in the initial state of the system; especially the positions and speeds of the bodies in question.
The evolution of this system can’t be predicted over the long-term by any computer.
The problem comes down to one of measurement. To predict what the system is going to do, you have to know the speed, direction and position of all the bodies in question at a certain moment in time.
But to measure the speed of a thing, you have to measure it over some period of time. It turns out that speed and position are incompatible measurements; there’s a limit to how precisely you can measure both at the same time.
The precise behaviour of a three-body system quickly becomes largely unpredictable.
It’s often claimed that free will cannot exist because the human brain must follow the laws of physics. What’s omitted from this assertion is that the laws of physics don’t prescribe how a complex interconnected system should actually behave. The laws of physics aren’t adequate for even solving the three-body problem, never mind to figure out what a human brain is going to do.
And to me, that’s not a gap or a flaw in science. The best use of the known laws of physics is to use them where they work, and to acknowledge their current limitations. After all, that’s precisely where we have the opportunity to make further progress.
To me, this raises an important question. Could an analog computer potentially do things that a digital computer could not? After all, the human brain has something that digital computers don’t have: namely, it has an inner subjective experience.
But my story doesn’t rely on Auron’s computers being conscious. It relies only on the idea that a computer need not necessarily be fabricated in a factory. It relies on the idea that, just maybe, a computer can be grown.
When I first started to think seriously about this possibility, I began to wonder what I’d actually do with a hyper-intelligent computer. Could I even trust myself with that much power? Could anyone be trusted with that much power?
I quickly realised that solving problems on the Earth would be fraught with difficulty. You can’t, for instance, cure all disease, without handing over incredible power to the world—and look at what we’ve done with the power we’ve already acquired. It hasn’t all been good; let’s just say that. It’s been a bit of a mixed bag, although few of us now would actually want to be without technology.
Then I began to build a new world in my mind. I imagined terraforming a distant planet, and building a new, almost-utopian society.
At first my daydreams were blissful and pleasant. But over time, the problems inherent in building such a world increasingly intruded into my dreams. Gradually my dreams turned to nightmares as I began to encounter one seemingly-insoluble problem after another.
The problem we end up being confronted with in constructing a technological utopia is nothing less than the same problem Catholic scholars have been wrestling with for two millennia: the problem of evil.
And I daresay Hebrew scholars were wrestling with it even before that, because what happens in the Garden of Eden pretty well sums it up.
If we have free will, we have the capacity to do terrible things, and some people will do terrible things.
Without free will, our lives are meaningless.
Apparently, this is a circle that even God could not square.
The Frith series is, I think, now approaching its conclusion, and for the remaining episodes, we’re going to get into the question of whether the past is fixed.
It’s a question that’s not unrelated to the question of whether free will is real.
Suppose a time traveller really could travel to the past.
Could he then change the past?
If the answer is yes, what happens to the past that has already taken place? That past was crucial to the time traveller getting his hands on a time machine, after all. Could he, for example, travel to the past and stop his past self from developing or acquiring a time machine?
This seems paradoxical, and there are at least two good possible solutions to it, leaving aside the strong possibility that travel to the past is simply not possible.
One possibility lies in the multiverse proposed by some quantum physicists. Perhaps there are many different pasts, and many different futures, all somehow co-existing.
Another possibility is that it’s just impossible to change the past. Our time traveller cannot stop himself inventing a time machine, because that’s just what happened. That raises the question of whether he really has free will. Is he doomed to do what he has already done, in a past that has already happened?
Perhaps the time traveller was always there in the one and only past that actually occurred, and he cannot undo what has already occurred.
Which possibility is going to pan out in my story?
Auron is trying to prevent a nuclear war that has already occurred, and he’s trying to prevent the death of his friend, which he feels responsibility for.
Is he attempting the impossible? And if so, what will actually stop him from intervening in the past?
Or can he indeed change the past, perhaps in effect creating two different pasts, and two different futures?
I hope you’ll join me for the few remaining episodes, and we’ll see what happens.
