Interview with Sir Peter Knight

Sir Peter Knight was born in 1947. The family lived in a village on the outskirts of Bedford.

Prior to the war his father had been a blacksmith but joined the Royal Engineers when war broke out. At the end of the war, he joined London Brick Company, and made bricks.  His mother worked on and off between looking after the children. During the war she worked at a site associated with Bletchley Park.

Peter’s parents were keen that their two sons had a good education and were supported and encouraged to pass their eleven-plus exams to be able to benefit from a scholarship to attend one of the local independent schools.

Peter says of family life: “We had a very modest background.  It was a little late Victorian cottage that we lived in, there was no running water in the house, but it was an incredibly supportive family environment. I wouldn’t have got through the eleven-plus without the encouragement of my parents.”

Peter attended his local schools in Bedford where he studied for and sat his eleven-plus and gained a scholarship to attend Bedford Modern School which Peter says “had the most remarkable science teaching”.

He studied nine O levels, three A levels and one S level. He says of his interests at school: “Although I’ve worked most of my life in physics, at the time I was actually really interested in chemistry experiments. We had a barn with water and a stove and I re-equipped it as my chemistry lab, making things that smelled vile and things that went bang.” Peter’s other interests included the Scouts, cycling and Youth Hostelling.

 Having completed his A levels, in 1965, Peter went to the University of Sussex.

He explains he chose the relatively new university based on a talk at school by one of the founders, saying: “It impressed me that they were attempting to do something, which was unusual in those days. They were supporting interdisciplinary or cross-disciplinary studies and thinking.” At a site visit Peter met with the science department staff, he adds: “They were quite an interesting group, many of them were kind of escapees from Oxford who could see that they could try to do something really different, and they were really quite inspiring. I had a chat with them, I thought this is the thing, I’m going to apply to this place, and I liked the idea that you could bring together disciplines that crossed and interacted. I went initially to do chemistry, but with physics and mathematics as minors.  But it was that ability to have that blurring of those boundaries that really impressed me, and at prelims at the end of the second term I immediately shifted across to physics because it was the atomic world that really interested me.  The place was quite glamorous in the 1960s and quite competitive to get into, but hugely supportive and enjoyable.”

After completing his degree, Peter remained at Sussex for his PhD. He says: “One thing you had to do at Sussex as an undergraduate was a final year project which was really a very demanding full year project working within a research group. I was working on a piece of atomic physics called optical pumping.  Relatively straightforward apparatus, where you could do things in a straightforward manner, it was sometimes fairly dangerous because the light source was a little bulb full of alkaline atom which you put in the top of a tank circuit. The whole thing had to sit in time in a magnetically shielded container that was light tight, and you had to jiggle the little light emitting bulb into the top of the tank coil above this thermionic valve until it the discharge struck.  Shocks used to throw me across the room on a daily basis, but it was great fun because it taught me that one could manipulate atomic systems in terms of coherence, a very strange concept, but it underpins what we do when we make atomic clocks.

“So I was doing these relatively simple experiments that uncovered some basic understandings right at the beginning and I had the most enormous fun doing this, and the theory around it. However, it completely misled me because I thought it was easy to do experiments – it wasn’t.  So I started doing both a mixture of theory and experiment as a PhD student in Sussex and it became evident that I was completely incompetent as an experimentalist. Nothing seemed to work. I had this enormous piece of apparatus which was really quite complicated, and I was the worst person to be let loose on it, but the theoretical stuff was really developing.  So towards the end of my first year I knew I had to be realistic, I’m not going to make an experimentalist, so let’s just focus on the theory side.  Everyone I was working with heaved a great sigh of relief that they’d finally found a way of getting me not to touch the bits of kit any more and so I became a theorist.”

Sussex University was the host of the Science Policy Research Unit (SPRU) which Peter became interested in because various people in the physics department held associate roles. He says: “Policy was actually interesting to me at the time, but what I didn’t know was how scientific insights translated into policy advice and moved things along.  It was really very much an amateur interest, as an evidence based scientist who was looking at decision making in terms of the evidence, you could see how that would that contribute to policy, and that’s always been an interest of mine.”

Peter’s first computer was at the University of Sussex. He says: “One thing you had to do as part of your undergraduate degree was to programme a machine in the computer centre. You wrote a simple programme as part of your coursework.  The one that I wrote was generating a string of random numbers.  It was on paper tape and you took your roll of tape and you put it in a tray. … It was a queuing system and you came back and if you were lucky they said it’s run and here’s a ream of printed paper that we’ve got your results on, or if you were unlucky, it didn’t run, and you had to do it again. It was a big machine room, you could see through windows that there were things in there that seemed to be revolving and so on, but you weren’t allowed anywhere near the place.” Peter wrote his programme he thinks in early Fortran.

In 1972, Peter finished his PhD and left Sussex University to take on a post doc fellowship. The fellowship was paid for by the University of Rochester but was based at the Stanford University with Joe Eberly, whom Peter describes as having played a major part in his scientific life.

He says of the experience: “That was a remarkable time. Art Schawlow, a Nobel Prize Winner for his work on lasers and so on, had labs on campus and I had an office up at the Stanford Linear Accelerator Center (SLAC).  SLAC ran an outreach programme for the kids in Menlo Park and they would run a science club on a Saturday morning in the cafeteria at SLAC.  Sometimes I’d pop in to talk to some of these kids from the local high schools.  I am told Steve Jobs and Steve Wozniak both came to that club.  I could see this emergent group of young people thinking about how you turn computing into something you could have on your desk at home.  … It’s a curious unaware intersection of lives.”

At the end of his spell at Stanford, Peter returned to Rochester which he describes as being at the heart of thinking about optics. He adds: “It was a small university, very highly research focussed, but had some of the most pioneering people in optics. Eberly was my boss, Emil Wolf was there, he was the grand old man of optics, the guru of the whole field, and Leonard Mandel, British, but had worked in Rochester for a while, and we would go off to lunch together.  So there I was, age 24, younger than most of the graduate students in the group, with these tremendous world-leading characters that were just transformational.  It was a great experience.”

From Rochester, Peter moved to Johns Hopkins University for a summer to take time to explore what his next move might be.

Having spent the summer at John Hopkins University, Peter returned to Sussex in the UK on a government fellowship which was part of the scheme to reverse the brain drain.  He says: “They pay for you to come back, paid your salary and your removal expenses, and so I seized upon that to get me back and to get a bit more re-established.”

He then moved to the Royal Holloway College, a small University of London college at that time.  He says: “It was a very small department, beautifully equipped with lots of facilities and the ability to use those facilities and to build one’s own group for the first time.  That was a great, I loved being there. As part of this attempt by government to support early career people, they instigated a scheme of fellowships for five years to do what you like and so I got one of these, it was called an advanced fellowship. After his interview for the fellowship position, he was contacted by the Head of department at Imperial College and invited to move to Imperial on the scheme with a promise of a tenured role at the end of the five years.

 Peter says of the Imperial: “Basically, it was a big engineering and natural sciences school.  I worked quite a lot on lasers and quantum stuff and so on, and in physics and in engineering there were a lot of people who worked on optics and information and so on, so there was a cross-talk between engineering and that physics.  So although I’ve always been interested in the applicability of light to do things, I’ve always been on the physics side of it all, but I’ve always had a close relationship with our engineering colleagues as well.”

Asked about his management style, Peter says: “I think I’m pretty good at the people side of it all, of getting people to work together for the common good, and I believe I’ve got the right sort of talents to engage with people. I’ve always needed somebody much more administratively competent behind me to support me on these things.  I’m not bad at the visionary part, I’m not bad at the people stuff, but I always need an operating officer right behind me to make sure that we actually deliver what we dreamt up.

 “Almost all of my career I’ve tried to make sure that we’re engaged with people who actually carry out the experiments and do the realisation of stuff.  I can come up with ideas and concepts, but the marriage of theory and experiment together is really the most advantageous thing.  Sometimes the structural mechanisms to do that have not really been in place in some places: in many universities in the UK, theorists sit in a different department to the experimentalists, and so on.  I was really lucky at Imperial that they never really had those constraints.”

On the impact of the UK leaving Europe, Peter says: “My colleagues in the rest of Europe are deeply saddened by our inability to continue to engage with them.  They really want us there, we really want to be there.  I’ve had years and years of wonderful engagement with the rest of Europe.  At the peak of my research group at Imperial College, half of the members were from the rest of Europe. It was enormously profitable in terms of an intellectual satisfaction.  We really made a difference.  The ability to move around Europe and engage with people was transformational and we hugely miss it.  Brexit has been enormously damaging to the scientific enterprise.  The scientists across Europe look back at those days with nostalgia, as we do, because it was the most perfect way to engage and it meant that we were part of an intellectual powerhouse that could rival the world.  Being on our own out on this distant island has really fragmented things. I really hope that we get better engaged and there are movements towards Horizon Europe.  But the damage that was done through the Brexit negotiations has generated big barriers to us getting better engaged in the future, so there is institutional damage that we need to repair, but the scientific willingness to get engaged has never gone away. I meet up with my colleagues from the rest of Europe and we often lament the good old days because it was of great benefit to the UK, as well as to the rest of Europe.”

Peter has been a visiting professor at the University of Louvain la Neuve in Belgium and at the University of Konstanz when he had his Humboldt Prize. He says: “Humboldt Prizes are lovely, they’re six months where you’re a guest of colleagues in Germany.  In my case it was with Jürgen Mlynek in Konstanz, a beautiful town on the border with Switzerland, doing the most innovative work in my field.  It completely freed me up from administrative responsibilities and so on and they allowed a lot of my students and postdocs to come over on visits and that built relationships that continue now.”

Asked if he prefers research or teaching, Peter says: “I hugely miss teaching.  I was enormously enthusiastic about teaching, so it’s not either/or, it’s and.  And uncovering new ways of thinking is often stimulated by doing a course and getting some bright young person say, ‘Why is it like that?’  It’s been a joy to teach and I miss it enormously.  I paint a rather sad figure, I wander round the department hoping that someone will ask this Emeritus Professor to give a guest lecture.”

Asked about why we do not yet have optical and  quantum processors, Peter explains: “In a sense we probably will.  There was a lot of fuss back in the eighties about optical computing and the way that one could use non-linear optics as a switch.  It turned out to be really difficult because you could turn these switches on really quickly, but you couldn’t turn them off really quickly and that became a bit of an obstacle to all optical computing using non-linear optics.  But it didn’t go away, and in fact, not only did it not go away, it started to come back with a vengeance. One of the things that one can do, especially in data centres, is to start to link up the ability to move information around using light with the ability to switch and couple.  People are using integrated optics in terms of chip-based processors within data centres.  The fusion of data centres, processors and so on, has already been transformational because you can’t use conventional switching techniques in a big data centre and hope to control the power budget. There’s a big investment in data centres in terms of integrated optics and silicon-based chips, so silicon photonics with very fast switches, the ability to modulate and so on is right now at the heart of what we’re doing in terms of big data centres.”

Asked about quantum computers, Peter says: “If you look at the way that a quantum computer is supposed to work, compared with a classical machine, what you have to do is to exploit the very peculiar quantum concept of being in a superposition of states at the same time.  It can be in a bit state one and a bit state zero simultaneously. These things are quite rare and very fragile and the slightest engagement with the environment disturbs that lovely superposition.

“If you have one of these things at superposition it’s fairly straightforward to keep that superposition going, that’s how an atomic clock works, and we’ve been exploiting it for years. Quantum technology has already given us GPS.

“If you put two together in the right way, that’s an entangled state, and that’s also quite fragile, and if you couple more and more of them the fragility gets worse and worse.  So the ability to get screwed up by noise in the environment is infinitely worse than in a classical machine.  It’s called the decoherence problem and that’s been the great challenge for all of us.

“However, you can turn that into an advantage, because the ability to sense the outside world is what you want in a sensor. A lot of what we’re producing in the quantum technology programme is actually a by-product of this fragility.  We can build quantum sensors that can sense the gravitational field around you.  This matters because half the holes dug up in London are in the wrong place, because they don’t know where the buried infrastructure is.  If we can improve by a factor of two that ability to sense below your feet, think of the impact.  That’s what we’re doing, we’re building those gravity sensors.

“You can imagine a situation where the ability to measure magnetic fields from the brain is really straightforward. That’s called magnetoencephalography.  If you tried to do that sort of brain imaging, you currently get shoved into one of these big superconducting surround that is deeply scary, but what our quantum programme, led by this wonderful group in Nottingham, can do is to replace all that stuff by a thing that looks like a cycle helmet. The little sensors sit in the cycle helmet and once you’re in the shielded room, you can move around. That’s hugely important if you’re trying to diagnose children, because you can’t easily put a child into one of these big superconducting devices unless you sedate them. This has already generated a new product and a new company called Cerca Magnetics.  They’re the only one of our start-up companies I believe that’s turning a profit at the moment. They are selling machines to Great Ormond Street and to SickKids in Toronto and so on, and it’s being used by surgeons treating juvenile epilepsy.”

“So, if we’re going to build a quantum machine, we’ve got to inhibit this ability to talk to the environment and make it quieter and quieter and that means we have to do much better in terms of the quality of the fabrication of the bits. It wasn’t until about 2015 that we could get that noise level down in such a way that we could be relatively confident we could build a fault-tolerant machine.  If you look at a classical machine, it’s extraordinarily fault tolerant, you build those error correcting codes in and the ability to actually control what’s going on in a chip on a classical machine is superb.  We’re orders of magnitude away from that at the moment in a quantum machine.  So the best machine that exists in terms of doing something only has of the order of a hundred fairly noisy qubits at the moment.  But our educated guess is that we will have a machine of substance, of consequence within about a decade.”

Peter continues: “There is a double Moore’s Law working to our advantage in quantum technology.  The ability to actually build better and better quantum bits has got a Moore’s Law with about an 18-month time, and at the same time, the people developing algorithms that are going to run on this thing have become less and less demanding.  So there’s a negative Moore’s Law running on the demand from the algorithm developers.  That’s why I say it’s a double Moore’s Law. I’m reasonably confident that quite soon we’ll be able to find ways in which we exploit these noisy intermediate scale machines for doing something in terms of the simulation of things, an optimisation.  Some of that is already underway, so we’re beginning to see something of economic importance quite quickly from those noisy machines.  Building a large-scale machine is harder and that will be at least a decade away.”

Peter also speaks about the impact of a quantum machine on internet security, adding: “People often talk about the fact that a quantum machine will undermine our entire internet security, because our encryption protocols wholly dependent on the inability of a classical machine to conquer what’s technically a hard problem which is factoring.  But checking you’ve got the right answer is dead, easy, it’s called multiplication, and it’s only polynomially dependent.  But the factoring is exponentially dependent on the size of the problem. A quantum machine changes all that, because of Shor’s algorithm, which shows that you can, with a quantum machine, turn what was an exponentially demanding problem into a polynomially demanding problem. For example, within about ten years we have to assume that everything that we use to secure the internet has to be replaced by something which is immune to a quantum attack.

“I’ve been working quite a lot on this issue, along with many others. … Part of our problem space is to work out how we can make quantum resilient or quantum resistant architecture in the internet. We know in principle how to do it, but it takes a while, you have to stress test it, road test it, red team, blue team it to make sure that you haven’t left any vulnerabilities in and so on.  The second part of it is to start working with all of the sectors of the economy that depend upon this to make sure that they’re alert to the need to be quantum ready.  Part of what we’ve been able to do is build this thing called a quantum readiness programme, where we are talking to the banks and to others and to the data centres and providers to make sure that they have an alertness to what is necessary to engineering this.”

On the subject of the potential future impact of artificial intelligence on society, Peter says: “I think it’s worth considering seriously about the dangers of AI.  I was in Washington two weeks ago where we had a couple of presentations on the dangers of AI and the ability to undermine societal responsibility for decision making.  I am thinking really quite hard about it.

He highlights that AI and quantum may well work together, adding: “AI and quantum are not alternatives, they could well come together. Quantum machine learning is something of real substance that’s worthwhile.  The thing about AI is that because you need to develop these large learning sets and then to grind through the whole learning process is that it can actually be quite demanding on power. A quantum machine may actually supplement parts of those algorithms to say, maybe you don’t need to grind it that way, you grind it this way.  So the ability of a quantum machine to do stuff in parallel may actually be advantageous in reducing that power demand.”

Asked about the UK’s ability to compete and make successful companies in the IT sector, Peter says: “We should have been good at it. If we go back to the late sixties, early seventies, we still had a number of companies of size and substance compared with our international competitors.  My personal view is that a lot of the electronics companies became so used to cost-plus contracts with the Ministry of Defence that they stopped thinking of mass markets and innovation and they became suppliers rather than innovators. I honestly do think that that was the seed by which we started to decline. Some of the work that was started in the UK, for example optical fibre work, not only the ability to carry stuff down fibres, but the ability to build, for example, amplifiers that go into the fibres, was very much British thing.  At the same time we had people of extraordinary competence working on semiconductor systems at Plessey and so on, but somehow with the mergers and bringing together of companies that had started to think Marconi-like, and when Arnold Weinstock started to think that the big surplus he had was much more profitable used as a bank, moving money around, rather than innovation, we lost our way.  We really seriously lost our way.  Of course we did have innovative companies, if you consider Acorn, the BBC Micro and then ARM, it gives you an example of what you can do if you’re imaginative and bold and innovative.  We just didn’t have enough of them in the seventies and eighties, I’m afraid.

So my feeling is that we lost our way in terms of the ability to see things through at a time when our companies were diminishing in their appetite to work on new product.  There were good companies that maintained a research activity to strike out to new areas; British Telecom is an obvious one.  But it was a rare phenomenon.  We had the most remarkable government laboratories at the same time that could have been the seed for lots of things.  Liquid crystals came out of the work in Malvern combined with Hull, but we lost the ability to exploit them.  A lot of compound semiconductor work was launched, for example, by Cyril Hilsum and colleagues in the UK, but we didn’t scale it up.  So we lost the ability to be innovative and scale for a while.  We’re trying to re-create it now.”

On his biggest mistake, Peter says:  “I think it was being so single-minded about understanding this whole area that I’ve been working on with the quantum stuff, that to some extent there’s been a bit of neglect of family life. My kids, and my wife especially, have been extraordinarily patient with me, but it has been pretty single-minded.  I think my biggest mistake is not to think harder about life balance.”

Interviewed by Richard Sharpe

Transcribed by Susan Nicholls

Abstracted by Lynda Feeley