Seán Connolly worked his way up the ranks of the Revenue organisation from systems analyst to information and communications technology director. Along the way he contributed to a radical transformation of Ireland’s taxation system.
Like its counterparts in several European countries, the tax authority ran large mainframe computers from Honeywell Information Systems and its successors. Its long relationship with this systems vendor made its computer centre unique in Ireland.
“I know he is a good general, but is he lucky?” is a quotation attributed to Napoleon Bonaparte. When reflecting on my career, which began in the general civil service, I have to admit that I was exceptionally fortunate to have spent most of my time as an IT specialist and manager in the Office of the Revenue Commissioners. The path was not always smooth but the satisfaction was immense. As my career was spent in the civil service, the opportunities and barriers were very different from those working in organisations where competition played a major role.
The civil service held an annual written examination for recruiting EOs, based on the Leaving Certificate syllabus augmented by a special test in mental arithmetic. The candidate had to add columns and rows of figures to produce a final total. The examination was held one month after the Leaving Certificate. Candidates were placed on a panel in order of merit. There was no interview apart from an examination by the chief medical officer after starting work.
I was appointed to the Department of Defence in 1962 and spent the next six years there. This was not my first military encounter. I had joined the FCA while in school and it trained me how to use the Vickers machine gun which was used during the First World War.
The Department of Defence was a classic bureaucracy which allocated work according to grade. It had strict rules and offered little challenge. Precedent was sacred and change was anathema. Pencil, pen and ink were used for writing. Biros were limited to making a copy using carbon paper. A jacket and tie were mandatory for men and women were forbidden to wear slacks or trousers. They also had to resign on marriage. There were different pay scales for married and single men. When we sent a letter to a civil servant in another department, the first name was not used. “Dear Murphy” was the standard form of address.
Promotion to HEO in that department was based on seniority and normally took about twelve years. I attended UCD in the evening and gained a diploma in public administration to improve my prospects.
In 1969 I applied for training as a systems analyst in Revenue – a position that offered promotion to HEO. By a stroke of good fortune I was selected along with Pat Daly, who had served as an EO in the personnel branch of that department.
Revenue was more like a business than a normal civil service department. It had almost two million “customers”. In contrast, the Department of Posts and Telegraphs had “subscribers”. The efficient collection of taxes and duties was critical for the running of the country. There were annual published targets to be met. Each budget prepared by the Minister for Finance brought urgent change. We dealt with almost every citizen at some stage of their lives. We collected and stored vast quantities of data and performed repetitive processes.
The Revenue Commissioners reported to the Minister for Finance but, in contrast to a normal department, there was no political interference in the day-to-day running of the business. Every action was subject to public scrutiny by the Comptroller and Auditor General and the Public Accounts Committee of the Dáil.
Civil service grades from bottom to top
CO Clerical Officer
SO Staff Officer
EO Executive Officer
HEO Higher Executive Officer
AO Administrative Officer
AP Assistant Principal
PO Principal Officer
Revenue was fortunate to have a manager of the calibre of Sean Bedford (1916-1989). As collector general of taxes he had acquired the first computer in the Irish civil service in 1963. He saw that the advances in computer technology could be exploited and was willing to take the risk involved. when the prevailing civil service culture was averse to change and risk taking.
Sean Bedford’s style of management was dictatorial. His staff and the computer vendors were loathe to question his commands. He also gravitated towards positions of power in his activities outside Revenue. He chaired the Irish Computer Society and the Honeywell Computer Users Association for the UK and Ireland. He was also the supreme knight of the Knights of Columbanus and led the successful campaign to have Oliver Plunkett canonised.
Pat Daly and I arrived in the data processing (DP) division in July 1969 and did the Cobol programming course run by Honeywell in the Civil Service Training Centre, Lansdowne Road. We got some programming experience before successfully completing a systems analysis course at the Polytechnic School of Management Studies in London. This was held in two three week modules, the first in late 1969 and the second in early 1970. Harry Nally and Tony Ford from CDPS also attended. There were written and oral examinations at the end of this course, which led to the NCC certificate in systems analysis.
An AP, Noel Hearne, managed the DP division with the assistance of two HEOs. Tom Mulherin dealt with technology and Donal Shanahan managed operations. Our software reflected established practices with standalone systems for specific taxes. There were about 18 programmers and analysts and a dozen computer operators, plus a big data preparation group with 80-100 staff. Taxpayer data was entered using paper tape which was considered more efficient than punched cards.
Revenue ran two Honeywell H200s around the clock. The night shift was the time for program development. Getting a clean compile could take a couple of attempts to correct errors. Some programs had over 3,000 lines of Cobol code held on punched cards. Hand-punches were used for urgent changes. Where a punched card had just one wrong character, some programmers re-punched that character and filled the wrong hole with a piece of chad. If the program was recompiled a few times, there was a risk that the blocking piece would fall out with unpredictable results.
The computer was also used at night by the Met Office and the Departments of Finance and Social Welfare.
For my first assignment I joined experienced programmers John O’Dwyer and Kay Deasy to maintain and enhance the turnover and wholesale tax applications. I was also assigned a bright new EO who had joined the civil service from school in 1970. Fergus Pollard had a particular aptitude for software development and went on to become a GCOS expert. He was poached by a company in Phoenix and eventually successfully launched his own company.
In late 1971 I was given responsibility for delivering a system for the new value added tax (VAT), which would be introduced when Ireland entered the European Economic Community in 1973. VAT was a modern replacement for a variety of duties collected by Customs and Excise (C&E) for specific articles as varied as mineral waters, rosary beads, windows, fireplaces and even wigs. The collector general’s office was assigned responsibility for implementing VAT because Sean Bedford already had a computer installation, the expertise and a proven track record. This decision had long-term consequences for C&E, where no champion emerged to automate its procedures.
We developed and tested a suite of data entry, validation, sorting, updating, collection and statistics programs for VAT. The postponement of the implementation date to November 1972 allowed us to modify the software to reflect emerging needs and to perform good assurance tests. Legislators did not understand the limitations of information technology at that time. At one stage they proposed using the computer to match every invoice in the country with claims for VAT credit on purchases. When the scale and logistics of data entry, validation and processing were understood, a more practical solution was adopted. Even then, the size of the physical VAT return form had to be carefully matched with the capacity of the NCR accounting machines which were used in the collector general’s cash office.
Careful monitoring by the VAT inspectors disclosed a pattern of suspicious repayment claims after the system went live. A limit had been set for the automatic refund but a large number of claims were just under this limit. We quickly developed scans to identify suspect cases and their addresses for examination by the VAT inspectors. The attempted fraud was nipped in the bud.
Principal officer Séamus Páircéir (1927-2011) was business manager for the VAT project. He was a charismatic leader, who had been an amateur actor and played a leading role in the first performance of “An Ghiall” by Brendan Behan. He went on to serve as one of the commissioners and as Revenue chairman. In these subsequent roles he encouraged and supported the expanding use of IT throughout the organisation. Having such powerful backing was vital when we proposed major changes.
In 1973 I was promoted to AP with responsibility for the programming function. We developed and implemented a standards manual and a development methodology. High quality programs were written, tested and released on schedule. Cobol remained our principal programming language.
Revenue now held over three billion characters of tax-related data on magnetic tape. The rapid advances in technology offered the possibility of instant online access to the latest taxpayer information from any of its offices around the country. This would eliminate the printing, distribution, filing and storage of vast quantities of paper, but it would require a more powerful mainframe, the introduction of magnetic disk storage, the construction of a telecommunications network and a substantial investment in VDUs.
The Department of Finance, which had to approve all expenditure on computer technology and staff, queried the business case for the move to online working, the cost of acquiring the 102 VDUs we envisaged and the procurement process. Each VDU cost about £5,000 which was almost the price of a new three-bedroom house at that time.
Revenue purchased a dual Honeywell H6060 mainframe computer running GCOS at a cost of £2 million. This had a comprehensive range of peripherals including Datanet communications processors to handle the network traffic. It was installed in a new purpose-built computer centre at St Johns Road. The installation was on the same electricity circuit as the nearby Dr. Steevens’ Hospital which shielded it from the periodic power blackouts. In time, back-up generators were installed.
The Honeywell sales team for the H6060 was led by Pat O’Sullivan. He moved on to Digital Equipment Corporation to handle the Department of Social Welfare account. Michael Cullinane had moved from CDPS to take charge of IT in Social Welfare. Although its business requirements were similar to Revenue, it adopted a strategy based on minicomputers and became a major Digital user.
The Revenue DP division had grown and was now located in Arkle House, Lower Mount Street, where there was a large VDU room with an ample number of terminals. The time to complete projects was considerably shortened and the existing systems were converted to run on the new architecture in eighteen months. This was facilitated by a timesharing system which allowed software developers to compile and test programs as required and was a great improvement on working the night shift. Pilot projects had tested the feasibility of linking the local tax offices to the computer centre. The Department of Post and Telegraphs had to train engineers to cope with the new requirement for data transmission.
Our EOs were recruited by the Civil Service Commission after completing their Leaving Certificate and were assigned to vacancies in the division without any prior assessment. They were generally very bright. Very few decided to transfer to normal civil service positions. Many went on to study for degrees at night. In my experience, there was no correlation between aptitude test scores and subsequent performance. Most conscientious people could learn to write computer programs. Patience, persistence and accuracy were required. Some were excellent but preferred other types of work.
Among the many talented people assigned as novices was Liam Ryan who eventually replaced me. Liam’s successor, John Barron, also entered as an EO. Dan Loughrey who worked in the computer centre became Revenue’s first public relations officer before moving to similar roles in Aer Lingus and the Bank of Ireland. David O’Callaghan became secretary general in the Department of Defence. Paul Haran, who was our operating systems expert, became secretary general of the Department of Industry and Commerce. Clare McGrath became chairman of the Board of Works. Micheal O’Ceallaigh became the director of the Local Government Computer Services Board.
In 1974 I received a BSc degree in computer science from Trinity College. Revenue had sponsored Pat Daly, Mike Sullivan, Johnny Maguire and myself to attend TCD for three hours on three evenings a week for four years. The class proved to be a valuable network of people with a wide range of experience. In 1977 I became a part-time lecturer on Trinity’s diploma course in systems analysis, teaching mature students how to write and test Cobol software. The demand for programming skills grew and I proposed a new three year course in advanced computer programming. This became very popular. When the class grew to 100, we divided it into two streams because of the need to supervise the project work. I persuaded Michael O’Connor from Revenue to take the parallel stream.
The rate at which advances in IT were being used in business and administration created a problem for academic syllabi and examination systems. The courses had to be constantly revised. It took about five years for the academic texts to reflect actual practice.
When Sean Bedford retired in 1978, Noel Hearne replaced him. I was promoted to fill Noel’s vacancy and took responsibility for applications development and support. Tom Mulherin was responsible for operations, technical support and acquisitions.
Each computer upgrade made the hardware more powerful and physically smaller. New releases of GCOS were implemented when they offered worthwhile benefits. Honeywell’s database management system was implemented in 1980 along with COBOL 74. The live applications were continually enhanced and new ones added. Every year, the proposed changes in taxation had to be evaluated and, when included in the new budget, quickly implemented. Where possible, tax rates, bands and other variables were held on files that could be easily amended and not hard coded in the programs. Accurate costs of proposals were obtained by testing them on the real data.
On one occasion, Revenue estimated the cost of proposals by one of the leading political parties to be twice the figure produced by their economist’s model. This was based on assumptions whereas Revenue used the actual data. The party stuck with their figures until the government tried to implement the changes and it had to admit that Revenue had been correct and the costs were wide of the mark.
There was a lot of controversy in 1981 about a government proposal to introduce a weekly payment of £9.60 per week for stay-at-home wives. Many voters did not realise that the amount would be deducted from the husband’s weekly pay by increasing his tax. We wrote and tested the necessary software, bought the additional printers and ordered the stationery to print over a million cheques each month. The project was scrapped when the government failed to get the budget approved and had to call a general election. That proposal was never resurrected. The project team were very disappointed that their work had been in vain.
The number of messages per day on the Revenue computer network increased from 15,000 in 1976 to 95,000 in 1985. Our annual budget for hardware, software and telecommunications was now £4.1 million and we had installed over 300 VDUs.
Revenue computing: The early years
The Office of the Revenue Commissioners, also known simply as Revenue, is responsible for the collection of virtually all taxes, duties and levies applicable in Ireland. It has a long history and some of the duties can be traced back to Norman times.
The state inherited the British system on independence when it combined the revenue and customs functions into a single organisation with a board of three commissioners. One was responsible for inland revenue, another for customs and excise and the third commissioner, who was the chairman, was responsible for personnel, the accountant general’s office, supply branch and the other central services.
The first computer in Revenue was an ICT 1301. Installed in 1963, it ran a pilot application for the collection of income tax. Revenue had to develop its own software and the emergence of Cobol in the early 1960s offered major advantages for training computer programmers.
It soon became clear that greater computer power and capacity were needed. In 1967 Revenue launched a public tendering process for a new computer and selected a Honeywell dual-processor H200 series configuration. The availability of a working Cobol compiler on this system was an important factor in its choice. Both ICT and IBM wrote to the Minister for Finance in unsuccessful attempts to reverse the decision. The existing applications were converted to run on the Honeywell hardware and new software was added to handle specific taxes.
The H200 was installed in a purpose-built suite in Aras Brugha, Cathedral Street, Dublin. Visitors were impressed by the high speed printers churning out revenue forms and the mailing machines that packed them in envelopes. The other showpiece was the storage drums that rotated at the push of a button to retrieve the desired magnetic tape.
On one occasion, a senior Revenue manager wanted to check that a new demand form was ready to roll. There was a software hitch that prevented the use of the master file containing the names and addresses. The developers inserted a sample of names and addresses in a program and used it to produce the required forms at high speed during the visit. The manager was satisfied and did not examine the actual data on the forms which was repeated.
Two independent Honeywell H200 model 2200s replaced the original H200 in 1968.
Like most big organisations, we had concentrated on automating routine clerical transactions. That world began to change when a new range of software tools became available on relatively cheap microcomputers. When I saw VisiCalc for the first time, I realised that it was ideal for evaluating taxation proposals for the annual budget. The electronic spreadsheet allowed the tax expert to easily calculate the financial impact of changes in tax rates and allowances. Alternatives could be evaluated up to the day before the actual budget statement by the Minister for Finance. In 1983, we purchased an Apple II microcomputer for the legislation and statistics branch. The user, Norman Gillanders, said that the old manual method of costing budget options should be documented as future generations would not believe how complex it was.
Word processing software enabled staff to draft and amend legislation more efficiently. This was particularly useful for the staff negotiating double taxation agreements abroad. We purchased a DEC Rainbow microcomputer in 1984 to run a payment software package for the accountant general’s office. Mainframe computers did not offer such options.
We feared, however, that anarchy would prevail if business areas were allowed to acquire and develop their own systems. For example, when I was asked for my view on a cash accounting system that had been selected for the Customs House, I advised that it could not be linked to other parts of Revenue and was to run on almost obsolete equipment. We offered to provide an alternative solution using standard Revenue hardware and software. Despite the initial resentment by the users, the project leader, Noel Doyle, gained their confidence and together they implemented a system that was a complete success.
To assess the potential of personal computers, I recommended the formation of an information technology working group of POs representing the major user areas. After an accelerated education programme, this group identified 50 potential applications for PC software and selected fifteen pilot projects. The ability to communicate with the mainframe was a key requirement. Some vendors offered excellent individual spreadsheet, word processing, record management, graphics and email software, but only IBM had an integrated solution. In early 1985 it supplied fifteen PCs with its Assistant software. We trained the users, making them aware that using a spreadsheet was a form of programming and therefore subject to human error. They would need to check the results. In January 1986 the working group concluded that the benefits justified the investment.
Liam Ryan, who could see where the technology was going, led the PC roll-out, which started with a local area network in Dublin Castle. He persuaded fifteen assistant secretaries and the three commissioners to join this ‘castle network’ and introduced them to various PC applications. This was a great success as top management gained first-hand experience of the benefits of IT. In time, all of the dumb terminals were replaced by intelligent PCs offering the full range of facilities for the 6,000 staff now serving in Revenue.
I was aware that the normal way to advance in the civil service was to acquire a wide range of experience. This meant changing jobs – and even departments – every three or four years. IT was seen as a specialist area. I decided to augment my technical experience with a formal management qualification. I had discovered a masters degree in management practice run by the Irish Management Institute and Trinity College. I had sought approval to undertake this course when I was an AP, but was refused as I was considered to be at too junior a level. I applied again in 1984 and got approval from Séamus Páircéir who was now the Chairman of Revenue.
The course required attendance on two or three days a month for two years and involved a practical work-based project. Mine was the development of a comprehensive information systems plan for Revenue. I reviewed the characteristics which differentiated the Revenue mission and practice from those of business, which was the focus of most management literature. Then, using John Rockart’s critical success factor method, Tom Mulherin and myself interviewed the fifteen assistant secretaries and identified their potential requirements for information systems. I augmented this research with a survey to identify areas of user concern. Senior Revenue management approved the adoption of this plan in 1986, when I completed the MSc.
In 1988 a competition was held to fill the combined post of accountant general and information and communications technology director. This was open to POs in all government departments, but I was appointed. Later that year Tom Mulherin succeeded Noel Hearne as collector general.
By the end of 1990, the Revenue network had 1,774 VDUs plus 691 PCs in use for word processing, local record management, spreadsheets and graphics. An automated entry processing system based on EDI started to collect Customs and Excise data in April 1991.
We also arranged for business continuity in the event of a major disruption. During the 1970s, when the Revenue computer centre was a possible target for a terrorist attack, we stored copies of all computer files and programs in a remote secure site. As the systems were based on batch processing, it was possible to do the work on other similar Honeywell configurations. We made agreements with five government installations in other European countries for mutual back-up in the event of a major disruption, running periodic test exercises at weekends. This cooperation developed into a useful forum for exchanging both IT and revenue experience.
The shift to online systems made these arrangements obsolete. We had to implement a backup computer centre to ensure continuity. It was difficult to get the funding for this, but I persuaded the Revenue chairman to write a formal letter to the Department of Finance which made it responsible in the event of a disaster. We got approval and the new backup centre opened in Dublin Castle in 1991.
We had gradually automated our clerical processes, but Revenue still operated as a series of largely independent silos joined at the top. The major applications often needed the same data. Many of the collection processes were similar. Senior managers therefore agreed that a consolidated approach would be more effective.
In 1990, we invited leading management consultancy companies to tender for a “data study” to recommend the future direction and awarded the contact to Andersen Consulting. This study led Revenue to adopt a customer focused strategy and to provide its staff with an overall picture each client’s liabilities and entitlements. We implemented this strategy in stages: a case management system in 1994, a corporate information facility in 1995, a common registration system in 1996, links to external debt collection agencies in 1997 and automated tax clearance in the same year. Integrated taxation processing went live in 1999. By then, the Revenue network linked over 5,500 workstations which generated 800,000 transactions per day. Over ten million items were posted each year.
A complete overhaul of the Revenue organisation followed, enabling us to deliver a top quality, easily accessible service to compliant taxpayers.
The unions agreed to combine specialist tax grades with the general service grades. We divided the country into four regions, augmented by a division that dealt with the largest businesses and wealthiest individuals and another that investigated non compliance. We began the implementation of this totally different approach to tax collection in March 2003 and completed the transition within twelve months. This was the logical conclusion of the integration of our IT systems.
The really extraordinary thing was how quickly we reformed structures that had not changed since 1922. I was glad to have made my contribution and that we had our share of good fortune to accompany the effort.
Last edit: May 2020
© Seán Connolly 2020
Skills and salaries
‘The common difficulties of recruiting and retaining suitable computer staff are aggravated in the case of the Civil Service by the fact that computer personnel are integrated into the general service salary and grading structure, thus limiting the degree to which the special and changing requirements of these scarce personnel can be met. …..The grading and salary structure for computer personnel would however need to be directly related to those outside the Civil Service and kept sufficiently flexible to adjust to changes in the supply and demand situation.’
– Professor FG Foster: Computers in Ireland (page 45), Economic and Social Research Institute 1971.
This problem was never properly addressed. The demand for experienced staff and the gap between the civil service salaries and those on offer in the private sector led to the continuous loss of valuable people. Some of Revenue’s GCOS specialists were head-hunted by companies in the United States.
The Public Service Advisory Council recommended in October 1986 that “immediate attention be given to the question of retention in the civil service of an essential minimum of expertise in the new technology. The problem might be addressed by the provision of a scheme of merit pay or special premiums for the good computer specialists.”
We did succeed in implementing a gratuity scheme for EOs in July 1989 years, but this was withdrawn in October 1995 when a general service grade claim was settled with the unions. The personnel branches in all departments did not support special arrangements for IT staff as they feared further claims for parity.
In 1997, a human resources policy group recommended EO as the basic recruitment grade with a progression to the maximum of the HEO after ten years. Nothing happened.
The failure to deal properly with the problems identified in 1971 also led to a gradual outsourcing of work. When the internet emerged, we had to contract out interface programming to companies willing to pay the going rate. The cost to Revenue and the public service of outsourcing far exceeded what it would have cost to adjust its own remuneration levels to retain skilled and experienced people.