In March 2004 I received an enquiry from Brendan Donovan, a son of Finbar, who was interested in his father's life and times. I told him what I remembered, as follows:

When I was in the Dublin Institute of Advanced Studies in or about 1960, or perhaps 59, we had a visit from JFD, who came at the invitation of Michael O'Connell who was then working with us; he was from UCC and I expect they were contemporaries.

At the time I was engaged in developing a small dedicated computing system which would relieve us of the tedium of calculating by hand the sequence of many 1000s of second differences of co-ordinate readings, taken by microscopic measurements on high-energy particle tracks in ionographic emulsion. These we required to total in groups without regard to sign, as a measure of particle momentum (multiple coulomb scattering).

We had prior experience of using the HEC computer, then the only computer in the country, which was in the Sugar Co in Thurles, in a context which involved doing a least-squares fit of experimental results to a theoretical curve. This was a less trivial calculation than the second differencing, and worth the trouble of data preparation and programming. The experience suggested however that the data preparation involved in feeding the microscope readings to the Thurles computer for a relatively trivial calculation would be an unproductive effort. So we developed a system for which we just tapped in the readings, as into an adding machine, and the total of the second difference moduli came out. This was fun, innovative, and generated a paper in Electronic Engineering.

Word of this must have got to JFD, and he decided to look in on us when he was on his way to Liverpool, where he had decided to go to get some cutting-edge computing experience. I don't think what we were doing was relevant to what he had in mind, but he paid a courtesy call to us in DIAS, where we were a source of some minor computing knowhow.

Why Liverpool? At the time it was a significant centre for computing development, and was paving the way for commercial computing, at a time when Britain was still in the forefront. There was a commercial computing operation called Leo around at the time, which (I think, better check this) had roots in Liverpool. When I left DIAS at the end of 1960 I considered joining it, but in the end went to Guinness in Park Royal, and into process instrumentation and control work.

What JFD did in Liverpool I never discovered, nor how he ended up as Aer Lingus Sales Manager. It will be necessary to piece this together with inputs from any Aer Lingus colleagues who may still be around (I am researching this). But one way or another he head-hunted me in or about June of 1963 to join his real-time reservations system project, and enhance the ALT team to a status of being able to talk across to the IBM people in the context of the proposal, with some level of knowledge-based understanding. Others recruited by him at the time included Pat O'Regan, recently retired from UCC DCS, John Quill, another Cork man with some IBM programming experience, and a couple of people from the North who had some computing experience in Shorts, whose names escape me. We became a team of 'senior programmers' so-called, working with junior programmers who were recruited internally from the junior staff, most of whom had joined with leaving cert.

There were 2 layers of internal bureaucracy between us and JFD; there was one Paddy Byrne, internal middle management, and Tom Clarkin, who knew all about reservations practice. These reported to one Niall Darragh, who reported to JFD, who sat remotely in the Sales Manager's office.

We talked across to Fred Kennedy, Paddy Doyle and , later, Brian Rothery. There was also Brendan Byrne, who was a source of programming and system analysis skills. The latter subsequently worked in UCD for a while. There was also John Moriarty, who ended up in charge of the TCD computer laboratory; he could be a useful accessible source. Paddy Doyle is also perhaps accessible, in Duncannon.

Brian Ennis was a junior programmer who evolved with the system; in 1985 after having had a spell with Core Memories he started his own firm Irish Medical Systems which now is a supplier of the NHS in Britain, employing about 100 people. I have been with him since 1988.

Fred Kennedy runs Captec in Malahide which is into satellite imaging and other space work with the ESA. System Dynamics is another spin-off, associated originally with Paddy Doyle. Cara was another spin-off. So the real-time project was an important seed-bed for development of Irish-based innovative enterprise.

In 1963 JFD's project, when set up, was totally separate from such data- processing work as then existed, which was punch-card based and set in its ways, under a manager whose name I forget, but I think he was Scottish. The sales system was planned as a great leap forward, with different personnel. This was dominated by internal company politics, of which we below decks had little understanding. We simply knew that there was an 'us and them' situation between Sales and Data Processing..

The 1963 concept was based on the 7010 (7090? I am not certain of the number) which was in process of being implemented by IBM with American Airlines, the Sabre system it was called. This was '2nd generation' equipment and depended on random-access mass storage, disk-based. Then in early 1964 a crisis developed: the Sabre system saturated at 1/3 of its planned capacity, and all hell broke loose. It turned out that those who designed it had assumed steady state average values, and had not reckoned with the build-up of queues for access to the mass storage, consequent on the environment being stochastic. The IBM had a team of 10 or so people working on a simulator which demonstrated the effects of queuing. In parallel with this, working on an idea suggested by Paddy Doyle, I made a mathematical model of an abstraction of the system, using queue-theory results from the Saaty text-book. I wrote a programme in Fortran which gave message-time distributions, coming up with the same result as the IBM simulation. This I think was a first use of queue-theory in real-time stochastic systems, for performance prediction.

JFD presided over all this, somewhat remotely. He must have had heavy political influence to keep it going, despite the first-round disaster; the latter was compounded by an analysis of the down-time distribution of the Sabre system, which was taken care of by the Sabre system being duplex. Aer Lingus were considering a proposal that involved a single system, for which the down-time would have been prohibitive. The first- round proposal was thus stone dead by about mid-64.

What saved it was the 360 being on the horizon; third-generation stuff. Much re- thinking was done, and in the meantime, they got a 1440 to service an interim semi-manual system, and this also replaced the punch cards, using an emulator to take over exactly the punch card logic, a short-sighted decision, dominated I suspect by a somewhat cautious and conservative data-processing management.

The 1440 had a Fortran compiler, and I went into economic planning, where various models were developed to help plan the route system, at strategic and tactical levels. We plugged in the various supersonic jets that were on offer for the transatlantic; no way could any of them have been economic. We used the model to decide about aircraft purchase. So this was a spin-off from the as-yet-unborn real-time system which put us in the forefront of computer applications in airline investment planning. This is all on record in the AGIFORS conferences. I was now out from under JFD and working with Oisin O Siochru for Niall Gleeson, who was Economic Planning Manager.

When the 360 proposal was complete, we got to analyse its projected performance, using the queue-theoretic model, against that of similar systems from ICL (which used drum storage for some data) and from a 3rd source, in the US, a non-runner, whose name I forget; Rothery might remember. It was decided finally to go for the 360 system, duplex, with the routine data-processing (still in emulated punch-card mode, as far as I remember) taking up some marginal time on the reservations system. The management of the sales system and the DP system was in the end unified into a Systems Department, and this was the platform from which David Kennedy, later to be CEO, emerged. He had previously headed Operations Research, which was an applied-maths group working on various quantifiable operational problems. My role in Economic Planning was somewhat similar, at a more strategic level.

JFD had I think by this time faded out, but I have no idea; I simply lost track, leaving Aer Lingus in 1970. David Kennedy would perhaps be a good source for getting a feel for the politics of how JFD steered the system; he might however have axes to grind; he was a shrewd political operator. I think he is still active. I would be interested to hear his story, if you manage to get it out of him.

It will be interesting to assemble insiders views and I will try to find some contactable people. The IBM people would also have useful outsiders views, and the key contact there is Rothery. It will be important to get views from various quarters, as I gained the impressions that it was a 'blood on the carpet' scene, at lease on some occasions.

My impression was the JFD was never much into the detail of the technology; his Liverpool experience was by then quite obsolete. So his skills must have been in the politics of innovation management, which is an important and little-understood area, in fact a minefield.

I would be interested to see what you come up with; you may jog my memory and stimulate more which may be useful..

Some navigational notes:

Copyright Dr Roy Johnston 2004