THE NEW ZEALAND TAB PROJECT
THE NEW ZEALAND TAB PROJECT
By Marcel Dayan - 28 April 2008
As these events occurred more than 35 years ago, and as with the RIMFIRE project, I am indebted to a number of colleagues who have helped me on details and to compile this narrative. The paper entitled "Considerations and Requirements of an Offtrack Betting System" written by Stuart Broad and David Rolls (and scanned by Stuart) provided a good basis to jog my memory on the technical considerations of the project. Peter Dulmanis added numerous anecdotes and asides which together with photos and scanned articles from "Between Ourselves" that Ron Bird provided, scanned articles from TABGENT that Bob Jensen provided and reviews and comments from various others, including Stuart Broad, Tom Daniel and Bob Jensen, all contributed to this paper. Subsequently, Bob Jensen was able to find a copy of the Project "Wrap-Up Report" that has been scanned and is linked to this document.
I am also indebted to Geoff Hipwell who scanned some of the slides and uploaded this paper to the web.
With the success of the Control Data supplied RIMFIRE and CARBINE II systems for VicTAB which went live on May 3rd 1971 and the computerisation of the NSWTAB with an IBM supplied system, the NZTAB decided it was time to investigate computerisation of their operation: They contracted PA Management from Sydney to act as consultants who then issued an RFI (Request for Information) in May 1971.
At that time Control Data was represented in New Zealand by
"Flight and Field Services", a company that was run by
"Digger Harding", a very well connected New Zealander who
had been a Battle of Britain pilot in WWII and had established
a successful business selling aircraft and ancillary equipment
to both the defence and commercial airlines. They operated
out of a very striking building called Anvil House in down-town
Wellington (see photo).
Peter Dulmanis and Marcel Dayan travelled to Wellington in August 1971 for meetings with NZTAB management and PA Management that Digger Harding had organised. Also included were meetings with the English computer company ICL (which was particularly well established in New Zealand) and Databank, a company set up to provide an IBM based centralised computer facility for the 4 large banks. Both ICL and Databank (who's CEO Gordon Hogg was a very flamboyant personality), wanted to explore a joint proposal where Control Data would provide the "TAB expertise" and ICL and Databank would provide their NZ incumbency.
PA Management was led by a very fiery Ian McCaulay, ably supported by Gus Richards and the local PA representative Ralph Bolton. The GM of the TAB was Maurie Smythe who had designed the manual system and had been the founder of the TAB in its current form, the AGM was Ernie Collins. Stan Dabrowski, the Administration Manager was put in charge of the computerisation project and his team included Don McKenzie, John Maher, Rangi Pope and Blake Henderson.
We had to get CDA's approval to bid under some duress as the company's first attempted entry into NZ was a costly lack of success. Peter Dulmanis recalls being grilled and having to make ambitious commitments in a one on one meeting with Herb Hughes one Sunday at his home. As a result after giving approval, Herb took a personal interest in our progress and this was most important in the battle to get the required resources and priorities. Control Data responded to the RFI in its own name, extolling the virtues of the CARBINE and RIMFIRE Systems.
RFP - Proposal and Pre-Sales Efforts
The formal RFP was issued in August 1971 (with a closing date of November 16) and work on the proposal began in earnest. The sales team was headed by Peter Dulmanis with technical support from Marcel Dayan, Stuart Broad and Chris Reilly (although other members of the Carbine/Rimfire team were also asked to make specific contributions.) Because NZTAB was significantly smaller than VICTAB, and it was going to be a very cost competitive process, it was decided from the outset that:
* 3100 computers would be used rather than the 3300 used by CARBINE II (called
the Master Control Processor - MCP)
* Two 1700 computers (essentially the same as RIMFIRE), would be used to
control the retail ticket network (called Ticket Issuing Processors - TIP)
* The RIOT terminals would be reconfigured so that one printer and one electronics
unit would be shared by 2 operators (called the Dual Ticket Issuing Machine - D-
TIM). There was requirement for 450 selling windows, a total of 225 D-TIM's
* Similarly the Telebetting terminal would also share electronics between two
operators (called the TELET). There was a requirement for 66 telebetting
positions, a total of 33 TELETS.
The payout fraud problem with RIMFIRE (where tickets were paid out twice or more) had been identified and it was Chris Reilly who came up with the concept of holding a bit-map on the MCP for each ticket sold and marking it off once it had been paid to overcome this problem.
During this period, meetings were held with Databank and ICL to see if there was any scope for a joint proposal. Databank's approach was that as the peak TAB betting was on Saturday, its large mainframe computers would not be handling bank transactions and thus could be devoted to TAB operations. However, when asked what would happen during a week-day if there was a clash of priorities between TAB and the Banks, Gordon Hogg said quite clearly that the Banks would have priority!
In the case of ICL, they wanted to use their mainframe computers and have CDA provide the software. CDA countered that it would supply the total system and they could maintain it! It then boiled down to the cost of the system, CDA's price being far lower than ICL. When they asked us how CDA could do it so cheaply, the reply was "that's our expertise". Needless to say, a joint proposal with either party did not proceed.
Peter Dulmanis remembers targeting Ian McCaulay as being a particular challenge and developing a plan to involve him as much as possible such as agreeing to his suggestion to explore using the NSWTAB AWA terminals and he and John O'Neil went on a tour to the AWA Ryde factory. From this contact we enrolled him as a co-developer of the "optional" terminal solutions and John O'Neil was instrumental in driving this. This of course lead eventually to the D-TIM and TELET but by then he had his finger prints on both perhaps symbolically but it did the job of getting his support.
NZ Department of Internal affairs
We also made contact with the Department of Internal Affairs (the government authority responsible for the TAB). By chance Marcel Dayan had been seeing Rosemary Dare in Melbourne whose father Ollie Dare was Deputy Secretary of the Department, This contact led to a meeting with the Secretary of the Department Pat O'Dea (later Sir Patrick in recognition of his Royal Tour organising) with whom we talked about Australian built terminals and NZ developed software forming the significant component of the contract (all classified as local content under the CER agreement then under negotiation). Again the fact we were extending our focus to this extent provided comfort to the TAB. This was also important in trying to neutralise IBM if /when they started to complain to government about faulty tender evaluation/unfair consideration. We also involved the Australian Trade Commission but we were not sure how much help this was - but at least it covered another base.
The proposal was completed on time and personally delivered to an Air NZ pilot who was to hand deliver it to Digger Harding who in turn would formally deliver it to the TAB Secretary Don McKenzie. Something went wrong and it did not arrive on-time. The proposal that was delivered to TAB on time was a one page Telex that identified the hardware components, stated the price as $4 million+ and ended by saying "details to follow". Luckily this Telex was accepted as a compliant bid (although it is doubtful that the competitors were aware of the saga).
Demonstration in NZ - Phantom Race Meeting
Elaborate preparations were made for the presentations which included running a "phantom" race meeting in the TAB board room in Wellington, allowing TAB management and staff and PA Management consultants to buy tickets from a RIOT terminal ( made to look like a D-TIM) that was connected back to the spare RIMFIRE module in Victoria. Phil Stokes had developed a special program to handle this demonstration from the Melbourne end with Roger Taylor and Harold Ford setting up the terminals in Wellington. There were a number of anxious moments getting the equipment going (especially when Marcel Dayan shorted out one of the RIOT terminals) and we had problems dialling Vic TAB because the telephone number sequence on NZ phones was the reverse of Australia. We were extremely well supported by the NZPO (under the direction of Bill Harlock) who provided the modems and ensured that we had a good quality satellite link.
The race meeting was billed as being an Auckland meeting, but once the video (hired from the NZBC) was started, some of the TAB management knew immediately that it was in fact a Hamilton meeting and started to smile. The second leg of the double was won by a horse owned by Maurie Smythe, as. Maurie calmly advised us at the very end of the demo. There was a belief that we had been extremely clever in our race selection however our convincing naivety probably resulted in more credibility. Monopoly money was given to each person and a bottle of whiskey offered to the person who ended up with the most (monopoly) money.
Final Selection of Preferred Supplier
The demonstration was ultimately a roaring success that arguably was the single most significant action that clinched the contract. The bottle of whiskey was won by the local PA Management Consultant (Ralph Bolton) who was not at all impressed when his win was mentioned in the local newspaper!
As usual for IBM at the time when they sensed the deal may be slipping, they lectured more forcibly and treated questioners with arrogance. Basil Logan national sales manager drove this process to a loss result. Stan Dabrowski later told Peter Dulmanis that a big difference between CDA and IBM was that IBM pitched and lectured (they knew best). Control Data on the other hand, listened, questioned, understood, went away and then responded comprehensively to whatever issues were raised. Control Data put a lot of effort into the extensive confirmation documentation that was passed back as a result, much to the cynicism of some or our colleagues. Peter says that this lesson stayed with him for the rest of his professional career.
As questions and answers were being exchanged, it was becoming clear that Control Data was heading the field, and it was in January 1972 that it dawned on Control Data that winning this contract would mean relocating the development team to Wellington, opening up an office and setting up a maintenance organisation. This commitment was insisted by and given to TAB.
On February 23rd 1972 TAB advised that CDA had been selected to enter into a contract for the supply of the system, provided that such a contract could be negotiated within 60 days - and with PA Management overseeing the process!
Contract Negotiations and Project Start-Up
With the formal announcement that Control Data were the preferred supplier in February 1972, work immediately began on the logistics of setting up an office in New Zealand and recruiting a team to undertake the system development. Peter Dulmanis was appointed as the Branch Manager of Control Data New Zealand and Marcel Dayan the Project Manager for the NZTAB project.
Offices were leased in Featherston St. (walking distance to the TAB Head Office in Lambton Quay) and contract negotiations began in earnest during March and April 1972 - especially the compilation of the "Functional Specification" that would be appended to the contract.
In parallel with these negotiations, the selected team members started preparation for their relocation. There were no established procedures or regulations for relocating CDA personnel to a "foreign" country, and so the quite generous US regulations were used and modified accordingly. Apart from Peter Dulmanis and Marcel Dayan, 7 Australian programmers relocated to Wellington. These were Stuart Broad, Bob Jensen, Phil Stokes, Linda Stevens. Glynn Hooper, David Rolls and Colin Gardyne. Peter and Glynn were married, David Rolls and Colin Gardyne both married American ladies during in 1971 their USA relocation but the rest of the team were single at the outset. However, by the end of 1972, Marcel, Stuart, Bob, Phil and Linda had also married - obviously no one wanted to be single in New Zealand!
Later Geoff Ellis and Pat Vallance (Customer Engineering) also relocated from Australia and a few more local CE's (Laurie Lee) were hired in New Zealand to build up the CE team.
ACROSS THE TASMAN
Pictured right in the reception area of the Control Data NZ office in Wellington:
Seated: Carol Clark - private secretary and Linda Stevens - Leader, TAB Off-line Programming Team: Standing, left to right: Bob Jensen - Leader, Master Collating Processor Programming Team (3100's); Phil Stokes Leader, Ticket Issuing Processor, Programming Team (1700's);Glyn Hooper - MCP Programmer; David Rolls - MCP Programmer; Laurie Lee - Customer Engineer (NZMOD) Colin Gardyne - Consultant to Control Data Programming Team and Stuart Broad - TAB System Programming Project Leader
Treble Bet Option
One major change to the specification was requested by the TAB during the negotiations, namely to add a Treble bet type to the specified Win, Place, Each-Way and Double bet types. The problem was that CARBINE had only 2 bits to specify the bet type and adding a 5th bet type would be a relatively major change. Jim Walters, the project manager of the first CARBINE system, had returned to work for CDC in Minneapolis and had floated an idea of specifying a bet in a "generalised" way to make it easier to add new bet types. Marcel Dayan picked up on this idea and suggested to TAB that they fund a study to see if this approach was better than just a straight expansion of the 2 bits.
TAB agreed and Jim Walters came to New Zealand and wrote a report entitled "Generalised Wagering System- GWS" which did away totally with the MCP/TIP concept and instead had a bunch of interconnected 1700 computers called TICs and TOCs. And - guess what - his estimate of effort to implement this completely new system from scratch was exactly the same - 149 man-months - as to modify the existing Carbine/Rimfire system! Marcel was quite shocked at what was recommended and offered Jim two alternatives, either change his conclusions that although GWS was a great concept, there was no way to implement it in the time frame (especially with liquidated damages that had been included in the contract for late delivery) or he could leave and Marcel would change his conclusions and present his report! Jim agreed to the former, changed the conclusions and then went onto Melbourne where with Herb Hughes blessing, he and Ray Sharp sold the GWS concept to VicTAB - and the rest is history.
After many late night sessions, particularly with Ian McCaulay and Gus Richards from PA and CDA management including John O'Neil and Dale Rostamo, the contract was executed May 10th 1972 (although all timetables used a May 1st start date), specifying a 23 months project schedule. This time frame did not include Treble betting, but did include the "hooks' to allow for its introduction soon after.
Note: The paper entitled "Considerations of an Offtrack Betting System" written by Stuart Broad and David Rolls also provides a more detailed account of the project implementation.
The Project 'Wrap-Up' report also provides a more detailed account of the project implementaion
During May and June 1972, most of the team assembled in Wellington and with design work getting underway. TAB hired 2 experienced programmers (John Taylor and Mike Thornbury) and put a number of its staff (Rangi Pope, Brent Longhurst, Mike O'Kane, Fred McIntyre and Graham. Iggo) through a programming course. As part of the contract, these TAB programmers were added to the Control Data team - on the understanding that they would be trained by Control Data such that they would be able to continue the development and maintenance of the system once Control Data personnel had implemented the system. The following organisation chart is taken from a copy of TABGENT (see below). In addition, Chris Reilly flitted from Sydney to Wellington, to Minneapolis developing software for the 3316 multiplexor that front ended the MCP design engineers Harold Ford, worked on the design of the TELET in Minneapolis and Roger Taylor and Frank Holland worked on the design of the D-TIM in Melbourne. The manufacturing plant in Cheltenham manufactured the D-TIM and TELET.
The paper entitled "Considerations and Requirements of an Offtrack Betting System" written by Stuart Broad and David Rolls provides an excellent technical description of the design and implementation of the system.
Communication Speed Requirements
There were a number of technical unknowns that needed to be resolved before the success of the project could be assured. The main one was that due to the high cost of communication lines, it was imperative to be able to get as much data as possible down a line. Whereas in Victoria, where more than 70% of the population lives in Melbourne, this was not the case in New Zealand, where only about 40% of the population live in Auckland, Wellington and Christchurch. To be viable, transmission speeds of 9600 bps between Auckland and Wellington and Christchurch and Wellington was required together with "state of the art" Time Division Multiplexors (TDMs) and this had never been attempted before.
Using 9600 BPS modems from Racal Electronics, an elaborate month long test was carried out to see what error rate would occur on these lines. Peter Rechnitzer wrote a test program for a Nova mini computer to log bit errors and we were able to show that this speed was in fact viable on the Wellington-Christchurch link (which used micro-wave technology) but not on the Wellington-Aukland link (which used regular transmission lines) This impasse was broken when the NZPO guaranteed operation on the Wellington-Aukland lines (as they planned to introduce micro-wave technology on this link in 1973.
Temporary Computer Room
TAB decided that they would build a new computer room on the roof of their Lambton Quay building and so Glynn Staggard and Gary Corcoran were asked to prepare appropriate plans. However, this would take over 12 months to build and so a temporary computer room was set up on the 4th floor of the existing building. Between June 1972 and the delivery of the computers in early November 1972 the team could only undertake design and documentation and were encouraged to take annual leave during this waiting period - in particular a very pleasant 2 weeks skiing at Coronet Peak for a number of the couples and the marriage of Phil and Ann Stokes.
However, at the regular monthly project meeting, Ian McCaulay was not happy to see the team disintegrate for 2 weeks at the start of the project and made the statement "the project is two months late already and there is no documentation". Marcel strongly disagreed and challenged PA to conduct an audit on 2 randomly chosen modules. Poor Ralph Bolton was directed to undertake this audit and happened to choose 2 of the best documented modules - after that we developed quite a good working relationship with Ian.
Ian was subsequently replaced as the PA Consultant for the project by Bob Scott who moved to Wellington from Sydney. Bob was a much easier character to work with. In the later stages of the project Bob was replaced by Nick Theodore who saw the project through to completion.
Relations with Stan Dabrowski and his TAB team were excellent from the start - his philosophy was to put away the contract and just get on with the job. If we had a problem, he treated it as his own and always looked for ways to help. Every Friday night there would be drinks in the TAB canteen and the Control Data team (and partners) were readily accepted as part of the group. The Control Data team also worked and socialised very well together (as expatriates tend to do), with many memorable parties that helped bond the group. Ostensibly the programmers were part of PSD (Professional Services Division) whose line manager was Peter Sumner and who would fly to Wellington every few months to keep in contact. Similarly there were regular visits by John O'Neil, Ron Bird and other CDA management - usually a good excuse for a party for all Control Data New Zealand personnel.
The computers were delivered in late October and had to be lowered by crane and in through the windows of the building (see photos). It is worth remembering that the car park of the TAB building was on the roof and it was accessed by driving down to it from the "Terrace". Wellington is earthquake prone and it was always a worry that the TAB building was actually built on the fault line which runs between the Terrace and Lambton Quay. Marcel vividly remembers seeing the computers in the temporary site "dance" across the floor during a particular strong quake.
THE COMPUTER ARRIVES ON THE ROOF AT HEAD OFFICE
Click to enlarge photos
Security of the TAB building consisted of a live-in care taker with an Alsatian dog called Karl and one Sunday he heard noises from the 4th floor and investigated what he thought was a break-in by letting Karl loose to intercept the intruder(s) - very much scaring one of our programmers and his wife who had come in to do some work over the weekend - an unheard of occurrence for TAB staff in those days. Subsequently, the caretaker got used to the idea of programmers coming in to work at all hours of the day and night.
NEW HEAD OFFICE AND COMPUTER ROOM
Click to enlarge photos
Project Manager Marcel Dayan with the prototype D-TIM terminal
By the end of 1973 the system was ready for module testing, together with the first batch of prototype and production D-TIM's and TELET terminals. Because of the threat of liquidated damages, there was a great incentive to get the system on-line by the contractual date of April 1, 1974, and Control Data management agreed that if this was achieved, Marcel could take the team and their partners for a one week "Wrap-Up Conference" to Fiji.
The project proceeded on schedule and towards the end of 1973 and early 1974, the system went through a series of "Informal Acceptance testing". "Formal Acceptance testing" and then a period of "Parallel Running". Marcel asked VicTAB if they would be able to send Tom Daniel to help NZTAB set up these testing procedures and run acceptance tests and they agreed to have him come to NZ for about 4 weeks. Bob Jensen had a bet with Rangi Pope that Tom would be able to crash the system on the first day, and duly won his bet much to the newly appointed Computer Operations Manager John Maher's chagrin. John was quite upset and took Tom to task, until Operations Manager Colin Mc Conniche intervened and smoothed out relationships!
There were many tense situations during the acceptance testing phase, with a daily 10 am meeting in the foyer of the computer room around a large white board in which new bugs were added and fixed bugs rubbed out. This way everybody could see the progress towards completion and it was a great team effort by everyone to make it happen on time. Telephone betting did in fact go live on April Fool's Day 1974 (right on schedule) with cash betting going live some 7 weeks later on May 24th 1974. Below are scans of two D-TIM tickets sold at that time (neither of them winners). The system exhibited very few problems and both TAB and Control Data management were extremely happy that the project had achieved all of its objectives
With the project end in sight, CDA hired John Watson to support the TIP after the Australian based staff had left New Zealand. The system was then progressively expanded to encompass the 3 major NZ cities of Auckland, Wellington and Christchurch and (as a second phase) was subsequently expanded to include a number of other NZ cities.
Peter Dulmanis's comment in retrospect is that the system had a project life span and cost benefit analysis over 10 years. Not only did the team get the project in on time and budget, but it ended up far exceeding the projected returns to NZTAB when it was later superseded by the integrated On-Course/Off-Course/Telebetting Jetbet system in 1982/3. This was especially so as one of the unanticipated benefits of automation was that it allowed betting from race to race which increased the turnover quite substantially and hence the project returns.
Towards the end of April 1974 the team together with their partners spent a week in Fiji, 3 days on Castaway Island and 4 days at the Fijian Resort. A good time was had by all and surprisingly a comprehensive report was actually produced. Two full days were set asside for the actual wrap-up conference to discuss the individual reports by each team member and to compile a "Wrap-Up Report" of the project. This report makes for very interesting reading after all these years and gives a more detailed account of the project.
From the left:
Eva Dayan, Lynn Rolls, David Rolls, Ann Stokes, Stewart Broad, Deidre Broad, Alan Cracknell, Linda Cracknell, Marilyn Hooper
FIJI Wrap-Up Conference
From the left:
Marcel Dayan, David Rolls, Phil Stokes, Linda Cracknell, Chris Reilly, Bob Jensen, Glynn Hooper, Stuart Broad