Roy Johnston July 1999)

(Roy Johnston reflects on 4 decades of attempting to survive with a physics degree in a peripheral post-colonial situation.)

The Irish scene in 1951 was far from being a promising environment for a fresh physics graduate, even one emerging from a department headed by ETS Walton. The latter's Nobel Prize was still to come, and official governmental support for science as part of the national development programme was still some 2 decades in the future. There was however in existence the Dublin Institute for Advanced Studies, which had been founded on the personal initiative of de Valera, as a sort of abstract tribute to the idea of scientific culture. Linkage with the existing university system were tenuous; the DIAS was primarily conceived as a haven for scholarly refugees from Fascism, and had little impact on the university scientific environment, which consisted of mutually exclusive sets in the two Universities: the intellectual Partition of Dublin mirroring the political Partition of the nation.

The DIAS, as well as hosting Schroedinger, had a cosmic ray group, headed by Janossy, which included an energetic Yorkshireman McCusker, who knew about pulse electronics. In 1951, as a gesture towards linking with the university physics departments, McCusker put on a trial course based on the Los Alamos monograph by Elmore and Sands. A few of us went to this, as a supplement to our final year. This seemed to suggest a way forward, so I took up an opportunity offered by the French Government and went to work with Louis Leprince-Ringuet's cosmic ray laboratory in the Ecole Polytechnique, with a letter from Walton as introduction. I got in with the Pic du Midi group; we had 2 large Wilson cloud chambers, the upper one in a half-megawatt solenoid and the lower one with lead plates. With this we measured masses of the various short-lived exotic particles which were beginning to be classified with phenomenological labels.

In the French environment I found myself regarded as the expert, having absorbed Los Alamos pulse electronics via McCusker; Elmore and Sands had not been translated and pulse technology was not well understood, being dominated by technician witchcraft.

In 1953 there was an international cosmic ray conference in Bagneres de Bigorre (the location being selected in honour of the Pic du Midi, which at the time was a key site for cosmic ray work) and I encountered Cormac O Ceallaigh, who had just been appointed to the DIAS chair, succeeding Janossy, who had gone back to Hungary. O Ceallaigh was in high standing internationally thanks to his discovery when with Powell in Bristol of the heavy meson. This was my opportunity to get into DIAS and I took it; there followed an academically fertile seven years, with several papers per year in what was then a rapidly expanding and exciting field. This was the basis for my FInstP.

In retrospect, what I got out of it was a training in systems engineering and a feel for 'dirty statistics' and the problem of getting a signal out of a noisy background. In DIAS however we were never close to the experimental strategy; we contributed technical support to the success of experiments which were increasingly centred in places like CERN, Berkeley and Stanford, once the accelerators came in as the source of high-energy particles.

Cosmic ray people divided; some like McCusker stayed with cosmic rays, getting into higher and higher energies via extensive air showers. This involved money, so McCusker left DIAS and went to Australia, where Messel, who had been in DIAS in the Janossy epoch, had managed to persuade the Government that physics was worthy of support. The O Ceallaigh group stayed at accelerator energies, with nuclear emulsion, depending on Berkeley and CERN for exposure; we honed the experimental technology, becoming a source of precision measurements on the ionisation-velocity curve in emulsion. This however was a dead-end.

So by 1960 it was necessary to make a strategic decision: stick with physics and go elsewhere, or diversify with a view to finding a niche in Ireland. I took the latter course, primarily because of a gut feeling that politics in Ireland needed some radicalising before it would become supportive of science, and recognise the importance of the dynamic linkage between science, technology and economic development in the context of national development. I had picked up some feel for this process from the writings of Bernal.

(Why had Messel to go to Australia, and bring McCusker with him? Why should the DIAS only be of significance internationally and have no relevance in the Irish nation-building context?)

There is space to mention the highlights of what happened subsequently. A spell in a brewery (guess which one) spawned what I think was the first application of a photomultiplier in on-line industrial instrumentation; we measured yeast concentration in a dark fluid, picking up the back-scattered light, giving a linear signal. Conventional nephelometry measured absorbtion, giving a log signal and only worked for clear liquids. Several patents came out of this epoch.

A switch then into computing: in 1963 IBM planned their first incursion into Europe with commercial real-time systems. The test-bed was Aer Lingus the Irish national airline. I was recruited in support of this by Finbarr Donovan, whom I had encountered when in DIAS; we were both engaged in use of early computers in the analysis of results. (I had done some curve-fitting on the ionisation-velocity curve in 1958 using the only computer in Ireland, a HEC owned by the Irish Sugar Co in Thurles; the Bernal model of the State sector being the technological pioneer seemed to work in Ireland!)

The IBM project stalled when the prototype system in the US saturated at about a third of its planned capacity. There was consternation, and much analysis was done to see why. It turned out that the design had been done on the basis of average values of stochastic variables; the real-time people had joined IBM from the rocket scene, where data-flows were steady. Here the cosmic-ray training came in useful: I had a feel for stochastic situations, and immediately recognised one, though I hadn't encountered it before. It turned out to be possible to model the real-time system in terms of queue theory. The system involved a hierarchical series of queues within queues, but all the theoretical work had been done, and it was simply a matter of applying it in the context of a structured model of the system in the environment. We were able to explain the US pilot saturation result to within a few percent.

We went on to produce a model into which we could plug the parameters of any real-time system proposed by a vendor and predict its performance, and we did this with the next generation of proposals, which involved 2 other vendors.

This work was done in the period 1964-66. Having discovered that what I was doing was 'Operations Research', some of the results were published in the proceedings of the annual conference of the Airline Group of IFORS. This was probably too obscure, as when the academics got hold of the idea of modelling real-time systems with queue-theory (this didn't happen until the 70s) I never got cited. I even discovered that Gordon Foster in the TCD Statistics Dept had published a proof of a theorem about the number of messages in the system which I had taken as obvious and used in the model.

I think this gives some insight into the way discoveries take place. Pythagoras got his name on the paper, but his theorem had been used for possibly a millennium before by the land-surveyors. The current 'experimental mathematics' people have identified this process and are building on it.

We went on to model the airline as a whole, with a view to plugging in possible aircraft purchases. In the course of this we modelled the various supersonic projects on the Atlantic: it was obvious that no way could any of them come near being economic.

There also emerged a model for predicting overhead costs in a management system: this I regard as unfinished business, as we never applied it in full, though some aspects found their way into the mainstream, like giving overhead costs 'volume-dependent' and 'variability-dependent' components. The full model involves open-system thermodynamics, with the firm as a heat engine, the main output being reduction of entropy. The TS term in the analogue Gibbs function, having the dimensions of energy, is the overhead cost. It is possible to define an entropy in information-theoretic terms, with Shannon as starting-point. So what is the temperature analogue? Consider Maxwell's Demon: the temperature he can maintain across the barrier is the measure of his ability. The manager is an entropy-pump, and the ability of the manager is the temperature-analogue.

Put high-ability managers in a high-entropy situation and the initial cost will be high, but if they do their job the entropy will come down quickly. There is scope here for some interesting dynamic modelling, with a crucial role for the differentials. Perhaps some theoretical person with a practical feel will take up this insight, and run with it. I would be pleased to collaborate as a constructive critic, and perhaps this time get my name on the paper!

I left Aer Lingus in 1970 and have been in consultancy ever since. I spent some time trying to develop models for creative activity at the interface between Dublin University and industry, with some though not total success. I ran a science and technology column in the Irish Times from 1970 to 1976; I emphasised the socio-technical aspects, and the creative aspects of innovative entrepreneurship; in its later years it was on the financial page. I find people tend to remember me now for this more than anything, and I enjoyed doing it. I stopped because I found it difficult to combine with being a practitioner.

Currently I am involved in a project with interactive video and a knowledge-base; I find a new label has been invented called knowledge engineer' and perhaps this is what I am. The firm is run by one of the 60s Aer Lingus real-time team; there are many other high-tech firms in Ireland with similar roots.

I think the Bernal model (this was consciously adapted to the Irish conditions in the 60s by the authors of a seminal White Paper 'Science and Irish Economic Development', in response to an OECD assessment) for introducing advanced technology worked in Ireland because there was mobility of labour and it was possible for people to drop out and 'do their entrepreneurial thing'.

In the now-defunct 'centrally planned economies', although the scientific and technological Establishment received massive State support, and was nominally influenced by Bernal's thinking, the 'mobility of labour' aspect was missing, and the nomenclatura-dominated bureaucracy stifled the process of transition science -> technology -> utility via the entrepreneurship process.

The 'enticing foreign capital' model also has worked where the Irish-based units employ scientists and engineers in significant numbers.

The mobility process appears to thrive through informal networks. I have never got a job or a contract by answering an advertisement. All my numerous switches have taken place as a result of informal network ontacts, often among people whom I knew before in an earlier existence, and who have themselves moved on. Perhaps this is a feature of a small open fringe economy.

Since the 1960s the split between the old colonial science tradition and the new Irish-oriented one has vanished; the latter has learned to appreciate the former and to build on it. There are aspects of this process which may be relevant in other post-colonial situations, especially in Africa. I wonder how many South African white scientists and engineers will stay and support the development of the economy under a non-racist democratic government?

(c)Roy Johnston 1992-1999

(Note added July 1999: the question posed in the final paragraph can now be answered positively, but questions remain regarding how science is to be applied in an African-type developing situation. RJ 7/99)