Interview with Chris Little

Chris Little was born in the north-east in 1949. His father was a  draughtsman and metalworker. His mother, after staying at home to look after Chris and his two brothers, went on to work in the Youth Employment Bureau. Chris says of his father: “He served his apprenticeship with Head Wrightson’s and the big companies on Teesside in the north-east of England. Somebody I met many years later said my dad could make anything in metal.” Later on, his father returned to university to study technical education and before the course was finished, was offered a role as a lecturer. He says of his mother: “After mum’s funeral, I talked to her boss, and she said mum was one of the best office managers she’d ever met.”

Chris’ parents encouraged their children to value their education. Chris, having gained his eleven plus, decided to go to the larger of the two grammar schools in Stockton-on-Tees. He says: “One was a small Church of England school and my parents said that I would stand a better job when you left there. However, they didn’t have a sixth form; you just left with O levels. I wanted to go to Grangefield, which was the bigger, state, school, which was where all my friends were going and they had a sixth form.”

In his third year, Chris was among a small number of students selected to be fast streamed to sit their O levels at the age of fifteen. Chris chose to study double maths and physics. and completed his A and S levels at seventeen, and in his eighteenth year sat the Oxbridge scholarships. He explains that his love of STEM subjects rose from his father and uncles, adding: “the kind of books that dad had on the bookshelf were things like Lancelot Hogben’s  Science for the Citizen and Mathematics for the Million. It was all part of that New Elizabethan era of electricity, atomic power, jet aeroplanes. I used to read The Eagle and it had centre pages with cutaways of the Queen Mary or the Comet aeroplane. Of course, I was a Meccano boy.”

In 1966, having completed his A levels early, Chris took the opportunity of a gap year before taking up his place at Oxford University.  It was during this period that Chris come into contact with his first computer, an English Electric KDF 9, having applied for a job at the Central Investigation Group of ICI in Billingham. The department was headed by Peter Price who wanted Chris to learn to use the computer and write a multi-dimensional, non-linear optimisation programme. Chris explains: “It was basically to work out the profit function, work out which is up, take a step upwards in that direction, work out the profit function again, determine the gradients and work out which is up and keep going up till you get to a maximum and then look around to make sure it’s not a saddle point. I understood the ideas quite well and I took to it like a duck to water, learnt to program in about two days after being given a manual and told to learn this: an autocode.”

He says of the computer: “It was a big machine, it had 4k of words, partitioned into four, so it ran four programs at once.  I never ever saw the machine because it was on the other side of the River Tees in Wilton. I used to go down to the punch room and the punch girls would take my coding pad and punch it onto five-hole paper tape.  The answers came back with compilation failure or whatever, on telex paper.”

After his gap year, Chris went to Oxford University to study mathematics.   He says: “I ended up being recruited to St Catherine’s College by Alan Tayler, who founded the Association of Industrial Mathematics, a bridge between the academic world and the practical world of trying to solve problems. … He was a fantastic teacher, he was really enthusiastic, he really persevered to explain things to you when you hadn’t grasped them. I owe a lot to Alan.” Among his lecturers were Michael Atiyah who became President of the Royal Society, Graeme Segal, Bryan Birch, who has the Birch-Swinnerton-Dyer conjecture prize named after him etc.

Chris continues: “Bryan Birch gave the first lecture I attended at the university; I thought this is really exciting and after a while I suddenly realised that all this new stuff is just going to keep coming and keep coming and keep coming, and there was the realisation that you’ve got to slog away. Michael Atiyah, was a small, lively character, who would bounce up and down on stage and really enthuse you and you’d come out and think, I didn’t understand a word of that, whereas some of the other lecturers were rather tedious and rather dull, but when you came out you understood it and you had a good set of notes.”

Having experienced two very enjoyable and adventurous long vacations of travelling and exploring , Chris was keen “to do good in the world and see more of it” when he finished his degree.  He explains: “I was really keen after my degree to go abroad to see the world and do good, as one does when you’re naïve and twenty-one, and I became a volunteer.  I worked with the International Voluntary Service (IVS). … I ended up going to Cameroon.  I spent two years there [teaching maths, science and sport in a secondary school] and I learnt a lot more about life than I did at university, though not a lot of maths. I had a motorbike and travelled to Timbuktu and back on my motorbike.”

Upon returning to the UK Chris decided to study for a Masters in quantum mechanics and general relativity at Nottingham University.

In 1973, following his studies at Nottingham, Chris started applying for a job within a 40 mile radius of Oxford where his then girlfriend, soon to be life-long partner, was working. He was recruited by the Met Office, he says: “They wanted somebody to do programming.  I knew a little bit more about computers. I’d not seen a computer for five years since the KDF 9 days, but while I was doing my Masters degree, I used to go for a drink in the evening with Nick Fitzhugh, who doing a PhD in machine independent job control language, or as we call it nowadays, Unix. I’d occasionally ask him about what he was doing, and that stood me in good stead for when I had the job interview at the Met Office.  The Met Office asked me the sort of questions about Assembler, first, second, third generation languages and I could answer all of that, and they gave me the job.”

Just before he joined the Met Office, they had replaced their KDF9 with an IBM 360/195. He says: “It was the biggest and fastest machine in Europe when it arrived.”

Chris goes on to describe the pen plotters and graphics, adding: “We had Calcomp pen plotters which were superb. They ran at an acceleration of 6G so when they were plotting you couldn’t see the pens, they were moving that fast, and they plotted A0 size; one square metre. IBM also supplied a Calcomp microfilm device that plotted on 35-millimetre black and white film. The film was exposed in front of a cathode ray tube, which was a vector screen.  They exposed the camera to the things being drawn and it would draw a 16,000 by 16,000 equivalent resolution in two seconds. The Met Office was producing 100,000 charts a day, most of which weren’t looked at, but you didn’t know which ones you needed to look at until the weather was sorting itself out.”

 The Met Office’s continued need and demand for faster, bigger machines was, says Chris, “driven by the mathematics of meteorology.  You’re trying to solve simultaneous, non-linear, partial differential equations in five unknowns. You approximate them by discretizing, by doing things on a grid; they’re not analytic solutions in any sense, it’s on a grid.  But the closer, the finer the grid is, the more closely it approximates the continuous solutions that the atmosphere behaves by.  So you know that if you’ve got a bigger computer, the weather forecast will be better.  That’s a mathematical kind of certainty, as it were.”

In 1950, meteorology was working with grids that had a spacing of 600 kilometres, just covering North America.In 1973, it was down to three hundred kilometres for the northern hemisphere. Chris says: “They were running a model down to fifteen degrees north, it’s not quite even hemispheric.” In 1991, the measurement was ninety kilometres for the full globe, 2017, it was down to ten kilometres and today it’s pushing seven, with a sub-grid of one point five kilometres over the UK.

Chris says: “In research now people are studying down to 200 metre resolution for urban areas. When the Olympics were on in 2012 we did run a third of a kilometre model for the Dorset coast for the yachting.  That was relatively easy to do because it was at sea and there was not much orography. The sea is relatively flat, so the numerical problems you get with instabilities in the equations weren’t there.  But as soon as you get to that kind of resolution, whether you have a tree or not in a valley affects the forecast. It’s that sensitive, so in that sense it really is chaotic.”

Chris continues explaining the three level model of grids used by the Met Office in the 1960s when it used the KDF 9, he says: “The very first models were run on ENIAC by John von Neumann and colleagues, it was  a single-level model.  Then it went to three levels in the sixties.  At the beginning of the seventies on the IBM mainframe it’s a ten-level model.”

In 1974, Chris learned Fortran when he was given the manual by his then manager, Ann Jackson, and tasked with writing a programme to print out the first hundred or thousand prime numbers. He says of the experience: “It was fine because my first program worked. I wrote a program, and she put the appropriate JCL around it and she took me down to where the card readers were and showed me how to punch cards and how to feed cards into the card reader, and then later that day the output came back on standard fanfold paper. Ann was quite surprised that the program had worked first time.  I was quite pleased when my next job failed, because it was a bit of a reputation to maintain.”

 Chris goes on to explain how the Met Office team has found ways to improve computer performance over the years. He says: “When I arrived in the Met Office in ’73 and then ’74, the forecast program, the model of the atmosphere was written in Assembler. The Assembler had been modified by the Met Office to get performance. They thought that the subroutine handling, where you’re saving the registers and then moving into an Assembler subroutine and then restoring the registers was much too slow, so we tweaked it so that it ran much, much faster.  It required a bit more discipline from the programmers and when things went wrong it was much harder to work your way through the dump.

“The other thing we discovered on the IBM 360/195 was that if you got your loops down to less than 32 bytes, it ran like a bat out of hell, because it stopped doing instruction ‘fetch’es and there was a very small buffer and it did everything from there and it was extremely fast.  The engineers couldn’t believe that we were getting 99.9% CPU utilisation.  We also tweaked the channel I/O programs. I didn’t do that tweaking, John Gibson [known as Rex] was one of the people that did that.”

At this time IBM had produced their Fortran H compiler. Chris says: “An H compiler is they’d put some optimisation in and when you wrote something in that straightforward H compiler Fortran it was as good as, but not much better than good Assembler.  It was starting to beat Assembler.”

After a few years, Chris took a role at the European Weather Centre which had just been established. The role was a three year secondment.  He says: “It was a pan-European organisation set up by all the national met. services in Europe, not just the European Union, it included Norway, Iceland, Turkey, and Cyprus, for example. ECMWF was tasked with doing one-week weather forecasts, nobody had ever done that before. You can’t do a one-week weather forecast without actually doing a very good six-hour or twelve-hour forecast, but they take a day to do that forecast, so actually that’s not useful for a lot of customers.  We can give you a really good forecast for twenty-four hours, but it arrives in twenty-four hours.  The ENIAC took twenty-four hours to do a twenty-four hour forecast, and it took three months to program the ENIAC to do a forecast. [For a one week forecast] You’re going across the life cycle of a depression which only lasts three or four days and then it dies and another depression appears somewhere. So you’re going across the generations of the meteorological systems. It was not at all clear that the science would enable that.”

The offices were located near Bracknell and the computer in use was a CDC 6600. Chris was involved in the data assimilation program together with Andrew Lorenc and Chris Clarke from the Met Office, Gorm Larsen from Denmark, Stefano Tibaldi from Italy and the head of the assimilation group, Ian Rutherford from Canada.

Chris continues: “We were using a new technique for doing the data assimilation, and that started on a CDC 6600, which was fun, because it was in the office next to mine and I could see my program running. I could talk to the operators and for fun we could run the Moon Lander programme,The 6600 had two big circular Tektronixs storage screens, which were very, very high resolution, and ideal for the Moonlander program, which was one of the first games that was programmed on it.”

A year into his secondment, the machine was replaced  with a Cray-1, which entailed several trips to Minneapolis St Paul and Chippewa Falls.   Chris says of it: “What was different about the Cray-1, serial number 1, was, as it was the first one, a  first working prototype, it didn’t have parity checking on the memory.  There was no single error correction, double error parity checking on the memory, so if a cosmic ray went through you may get a wrong result.”

After his secondment, Chris returned to a new  role at the Met Office, became a senior programmer and was involved in the replacement of the IBM mainframe with the next supercomputer; a contract which was won by CDC with a Cyber 205. It had a speed advantage over the Cray because it could do floating point calculations in 32 bit precision, rather than the Cray’s 64 bit. 32 bit precision is adequate for most well conditioned meteorological calculations.

On his management style, he says: “I like to let people do what they’re good at, it is actually by demonstration. I expect the same high standards of them that I apply to myself, which is not necessarily very good management.  I think I’m kind to people, though I can actually take hard decisions.

Asked why he has stayed at the Met Office for nearly fifty years, Chris says: “It’s intellectually interesting.  Simple as that. In a sense I’ve had four or five careers inside the Met Office. My first career was data assimilation and did the change to a different maths technique, so when I first started in the seventies I was actually using the maths I was taught at my Masters degree in Nottingham. When I went to ECMWF it was a completely different mathematical technique, it was optimum interpolation.  When I came back to use the CDC 205, doing my third operational data assimilation program,  a third technique which was variational analysis[was being developed].  So that was intellectually very interesting.”

“I still had this inkling that maybe I’ll be a scientist. I want to publish a paper, and having written all this code, I’d like to do some science about how you assimilate those observations and the stuff about the impact of satellite observations, running the forecast with and without satellite observations.”   However, it was while he was on this third operational data assimilation scheme on the Cyber 205 in the Numerical Weather Prediction research team, that Chris realised he was best placed in IT.

Asked whether Y2K was difficult for the Met Office, Chris says: “In some ways it was very straightforward.” He goes on to say that other software within the Met Office was changed and updated, Chris adds: “I remember just after it all happened and it worked and there wasn’t any disaster in the Met Office, meeting one of the directors coming springing up the stairs, He was over the moon, but I was tired, it was a lot of hard work.”

He continues to highlight a project designing data formats for the World Met Organisation which included binary formats, and which saw the team invent GRIB, which stood for gridded binary data, and also invented BUFR, which is Binary Universal Format for Representation data, which did the observation data, and had a very flexible data format, hinged around 16-bit structure, rather than a 8 or 7-bit structure.

In 2002, the Met Office relocated to Exeter, a move that involved no loss of service and was the biggest ever IT move in Europe. Chris, who was involved, says: “There were ninety separate projects and I was given one of the pear-shaped ones, which was we’d identified thirty customers that had their own telecoms links to us.  … Lots of people had leased lines and we had a policy of owning our leased lines. which have an A end and a B end.  You own the line, you pay for it, you pay BT or Cable & Wireless, or whoever, who run the cable.  And if you’re paying for it, you own the A end and the customer has the B end.  But about thirty customers who had the same policy, so British Airways, for example, owned the A end, they owned the link to us. I had the job of going round to all of them and explaining, we are going to move down to Exeter.”

Initially, there was reluctance to move and many of the thirty customers did not take the deadline seriously. That changed after Chris had a letter drafted and sent to each of them from the Met Office CEO.  He continues: “Basically it said, we are moving, this is the deadline, we would like you to move the B end of your cable down to Exeter, we suggest you have a one-month overlap so you could transition, and if you haven’t done it by then we’ll terminate the contract and we’ll forego the money from you.  It’s much more important to move.”

Asked about his biggest mistakes, Chris says: “Some of them were fuelled by not trusting my own confidence. …

“In my third generation of data assimilation I tried to make things too complicated.  I really thought I could write an all-singing, all-dancing, thinking of all possible cases and program it.” In doing the exercise, Chris learned that doing things simpler is sometimes better, he concludes: “Doing things simpler and hard coded sometimes gets what you want.  I seem to remember somebody once said that if you’re trying to make a piece of software fool-proof, you have to remember fools are very, very smart.”

Interviewed by Richard Sharpe

Transcribed by Susan Nicholls

Abstracted by Lynda Feeley