Unisys Australia Ltd v RACV Insurance Pty Ltd & Anor [2004] VSCA 81

Court of Appeal (Vic)|2004-05-14|Before: ORMISTON, PHILLIPS and BATT, JJ.A.

Trade and commerce - Misleading or deceptive conduct - Contract to provide computer system for the handling of insurance claims - Negotiations prior to contract - Retrieval times specified in pre-contract documents - Whether represented by provider - Whether representations false - Whether relied upon - Whether "reasonable grounds" for making - Trade Practices Act 1974 (Cth) ss.51A, 52, 82.

On this appeal and cross-appeal I have had the considerable advantage of reading the reasons for judgment to be delivered by Phillips, J.A., with whose comprehensive reasons I agree. In consequence the appeal should be dismissed and the cross-appeal as to damages should be allowed in the manner proposed by Phillips, J.A.

The appellant, the firstnamed defendant below, is incorporated in the United States of America and is registered as a foreign corporation in Australia. Named in the pleadings as Unisys Australia Limited ("Unisys"), it is a subsidiary of Unisys Corporation, a supplier of computer systems. The respondents (the plaintiffs below) are RACV Insurance Pty. Ltd. and RACV Group Services Pty. Ltd. The former is a motor vehicle and general insurer and the latter is a related corporation which provides computer and other services to companies in the RACV group. Like the trial judge, I refer to RACV Insurance Pty. Ltd. as "RACVI" and to RACV Group Services Pty. Ltd. as "RACVGS". In some of the materials in the appeal book and on transcript the latter was sometimes referred to as RACV IT or RACV IS. Although a separate company, it was sometimes referred to as "the IT department" or "the IS department", reflecting its role within the RACV group. I shall refer to the two respondents together either as "the respondents" or as "RACV", when it is not necessary to distinguish between them.

When this proceeding was commenced, there were two further defendants, Deloitte Ross Tomahtsu and Deloitte Touche Tomahtsu, which were different incarnations of the accounting firm known generally as "Deloittes". Deloittes settled with RACV before the trial and although the settlement sum was not made public there was debate at trial about the significance of the settlement sum to the assessment of the damages awarded to the respondents.

The proceeding from which this appeal stems concerned a project for a workflow management system (or "WMS") within RACVI. The project was intended to reform the work practices within that company's motor vehicle claims department. It involved, but was not limited to, the computerisation of those processes to create a paperless office. Deloittes were engaged by RACV in 1992 to act as consultants on the project and in mid-1993 expressions of interest were called for from vendors of computerised workflow management systems. Following a competitive tender, a contract was awarded in December 1993 to Unisys which proposed to adapt its InfoImage system. In due course Unisys delivered a system to RACV and it was accepted after testing in March 1995. However, that system failed in operation and it was soon decommissioned: it was called by the judge a "complete performance failure" and it is common ground that it was properly so described.

Unisys then set about rebuilding or repairing the system and tests were conducted on it in early to mid 1996. RACV remained dissatisfied with some aspects of the system revealed in the testing and it remained dissatisfied with the response times for retrieving images. These were the images of documents which were to replace the paper version under the computerised system. Having set a deadline of 24 June 1996 for the system to be ready, RACV terminated the project on 27 June 1996. It remained in possession of the computer hardware and software but it did not seek to prosecute the project to completion, either with Unisys or with another supplier. Instead, it issued the writ in this proceeding in December 1996, claiming damages for misleading or deceptive conduct, breach of contract and negligent misstatement.

The trial occupied more than two months in the Trial Division during 2001, the trial judge announcing his decision on 24 August 2001 and delivering detailed, comprehensive reasons for judgment. Judgment was formally given on 31 August 2001 in favour of RACV and damages were awarded in the order of $4 million, for expenditure wasted on the project. The judge did not award the damages in contract or for negligent misstatement, or indeed in respect of misleading or deceptive conduct based on representations allegedly made by Unisys after the contract was entered into. His Honour saw no need to deal with those claims in view of his findings in relation to three of the nine representations relied upon which (as he found) were made before the contract was entered into and which the judge found had induced RACVI to enter the contract and RACV to incur the expenditure. By reference to those three representations, damages were awarded to both the plaintiffs for misleading or deceptive conduct, Unisys having failed to satisfy his Honour that there were reasonable grounds for its making the representations in question. See Trade Practices Act 1974 (Cth) ss.51A, 52 and 82.

The three representations which were critical to the judgment below related to the proposal made by Unisys to RACV for a computerised system. As pleaded in paragraph 20 of the amended statement of claim, they were these[1]:-

"(d) the Unisys proposal would if implemented provide adequate storage capacity to provide, on a `date forward' basis, for all open claims on-line, all claims near-line for three months following closure, and closed claims older than three months off-line; (e) high speed image access would be available in a WMS system implemented in accordance with the Unisys proposal for on-line claims with an expected response time in the vicinity of 2-4 seconds; (f) Image access would be available in a WMS system implemented in accordance with the Unisys proposal for near-line claims with an expected response time in the vicinity of 20 seconds maximum."

This was all in the context of the representation, next pleaded, that "image access using the WMS system proposed in the Unisys proposal would provide improved customer service for RACVI through faster response and more efficient operations", for that was the purpose of the project: to improve customer service and in particular to improve upon the handling of claims, being those which were still active or current (that is, open claims). Thus, for RACV retrieval times in the new system were all-important. As the pleading indicates, to be "on-line" meant (at least to RACV) a retrieval time in the vicinity of 2 to 4 seconds and to be "near-line" meant a maximum retrieval time in the vicinity of 20 seconds; and the judge found that the contract was let by RACV to Unisys in reliance upon the three representations I have identified.

Put shortly, it was the requirement that all open claims be kept on line which proved the stumbling block, for if kept on-line the retrieval times of 2 to 4 seconds would generally follow. Plainly, the system as delivered in March 1995 failed on all counts: as already noted, it was a "complete performance failure". The system tested during the first half of 1996 was perhaps better, and the appellant stressed the improvements before us. But by 24 June 1996, the revised date for implementation nominated by RACV and agreed in by Unisys[2], it is the fact that the system as delivered by Unisys still had its defects. Counsel for Unisys submitted, on this appeal as below, that by 24 June 1996 the remaining defects were not only far fewer in number than they had been, but were capable of being cured; but, as I see it, that was not by then the issue. The fact is that in June 1996 the system was still not functioning, or not functioning as it should have been if the critical representations were made as pleaded and were to be met in performance - and to my mind that was the end of the matter.

The trial judge put it succinctly in paragraph [456] of the reasons for judgment, where, when identifying the problem as an insufficiency of "cache" (which was high speed memory for the most frequently accessed data[3]), he said:-

"The situation then is that, contrary to the representation relied on, Unisys did not configure cache with sufficient space to hold all current claims. The representation was as to a future matter in that it concerned the system to be implemented for RACVI. The fact is that Unisys never configured the system to accord with the representation. To put it as the plaintiffs do, the system delivered in March 1995 was not so configured and Unisys did not thereafter rectify the configuration to accord with the representation."

To the extent that the representations made in 1993 were as to the future, Unisys attempted at trial to establish that it had had reasonable grounds for making such representations, but the judge held that it failed. Accordingly, Unisys was bound to answer in damages to RACV.

Unisys now appeals against the judgment below, challenging nearly every step in the trial judge's reasoning and on occasion his Honour's primary findings of fact. RACV cross-appeals on a question affecting the quantum of damages awarded. It also filed notice of contention, seeking to justify the judgment below on grounds other than misleading or deceptive conduct based on representations made before the contract was entered into. As will appear, it is not necessary, in my opinion, to consider the notice of contention: I would dismiss the appeal anyway.

Before turning to the details, I make one or two points of a general nature. At trial, there was a vast volume of documentary material and a great deal of oral evidence, much of it "of a factual and technical nature concerning the computer system". As the judge noted[4]:-

"To the close of evidence the transcript ran to 2,731 pages. It extended to 3,178 pages by the close of final addresses in which counsel spoke to written submissions which ran to 378 pages and five supporting files of authorities. Then, six weeks later, on 15 May 2001 Unisys applied for leave to amend its defence to raise a new point; following a hearing the application was withdrawn ... At the outset of the trial the court book consisted of 49 lever arch files ... which contained 18,935 pages. I was also given another lever arch file which contained the December 1993 agreement and which seemed to comprise several hundred pages at least. When finally tendered at the conclusion of evidence the court book was reduced, by the agreed discarding of thousands of pages, to 28 volumes including [that] containing the pleadings. I am, however, left with many thousands of pages of oral and documentary evidence."

The judge's solution, with such an array of material, was to accept counsel's final addresses as "identifying the issues and the relevant materials". He emphasised, however, that in arriving at the reasons for judgment, he had "read and considered, subsequent to their making, all that counsel said in final addresses, both oral and written, and in their opening addresses, and the evidence of the witnesses." It seems to me, if I may say so, that the trial judge did a most thorough job. The litigation, lengthy and detailed as it was, arose out of Unisys' failure to deliver to RACV, by the cut-off date of 24 June 1996, a system that was functioning and would answer to the needs of the customer. Not surprisingly perhaps, the appellant lost comprehensively at trial. What is, I think, surprising, is that the appellant, with a significant worldwide reputation, was prepared to put that reputation on the line, not only during such a long trial but also during the appeal which lasted some ten days. I return to this in conclusion; for, in my opinion, if, as I think it should, this appeal fails and costs are ordered to follow the event, there is at least an argument that they should be ordered on a basis beyond that merely of party and party.

For the purposes of the appeal, it is not necessary to set out all the facts of the case as found below; they are rehearsed in detail in his Honour's reasons for judgment and reference will be made to them only as and when necessary to deal with the arguments raised on appeal. The following, which is intended simply to give context to what follows, is taken from the reasons for judgment and the appellant's own outline of submissions[5].

RACVI conducted business as a motor vehicle and general insurer. In 1993, the claims process was paper based, each claim having its own paper file, a system which was labour intensive and considered inefficient. One problem was the delay involved in having to locate and retrieve files, which produced inefficiency in the handling of claims and telephone inquiries. Although in 1993 the claims work was centralised in Melbourne, there were approximately 90 claims consultants divided into 6 groups, each with a team leader, and in 1992 they had handled about 75,000 claims, involving about a million documents. A possible solution to the problem of handling an increasing number of claims, and doing so in a cost effective and efficient way with an improved service, was to introduce new technology which might include document imaging. And in the early 1990's RACV set up a committee or task force to consider the possibilities.

That committee produced a report in December 1991 and in March 1992 one Duncan MacCallum, an accountant working with Deloittes and with a background in information technology, met with RACV personnel and discussed workflow reform by means of computerisation. Further reports were produced by the committee in May 1992 and following. In August 1992[6] RACVI, after consideration, engaged Deloittes to provide consultancy services and MacCallum was given the carriage of the project for Deloittes. As MacCallum said in evidence, the aim was to increase the efficiency of the processes employed in handling claims, including reducing the time to process a claim, reducing the cost to RACVI of the processes, and providing improved service to those insured with RACVI.[7].

In December 1992, MacCallum produced for RACV a two volume report entitled Motor Vehicle Claims Workflow Management Study. It contained a summary of existing processes and indicated ways to improve customer service and productivity, with increased efficiencies and cost savings. It was important there be a justifiable return on investment if the project were to proceed. There is no need to refer to the report in detail (it was not seen by Unisys), but the judge identified two aspects, one of which concerned the requirements for the image services facility. Under the heading Referral and Access it was required that there be "high speed access to documents through pre-staging and magnetic caching" and that there be "access to active claim documents within 2 to 4 seconds from the magnetic cache or other pre-staging area. If not pre-staged, then access within 30 seconds". It was recommended that a project task force be appointed, including representatives of RACVI and RACVGS. This was the New Claims Process Team. Chesson-Kistas, a claims officer with RACV Motor Claims, was to work with the project full time from March until December 1993.

The new team met on 18 March 1993 and during the meeting MacCallum spoke about the request for information document which he was then developing ("the RFI"). This set out the user requirements which the vendor of a workflow management and imaging system had to meet. In February 1993, Unisys heard that RACV was looking for a document imaging solution for their workflow processes and a meeting was arranged between RACV and one Schmidt, of the German insurance company DBV, which had itself implemented Unisys' InfoImage system, Schmidt happening to be in Australia in March 1993. He met RACV personnel on 25 March and spoke to them about the system implemented for his company by Unisys. According to the findings made below, the purport of the meeting was that Unisys technology had been successfully used in the USA and Europe, Schmidt saying that it had provided efficiencies and good returns and Chesson-Kistas speaking of the large volume of claims and the need to process them on a daily basis and to have speedy access to stored information. Present at the meeting was the marketing manager of Unisys, Brian Josephson.

In March 1993, the RFI was completed. It was a 71 page document entitled "Workflow Management Project - Motor Vehicle Claims - Vendor Symposia Briefing". Dated March 1993, this was a pre-tender document and it was provided to 10 selected potential vendors. As found by the trial judge[8] (and the following was not seriously in dispute on appeal):-

"The purpose of the RFI was to provide relevant information concerning the project to the potential vendors, including the requirements of the demonstration and the criteria by which they would be assessed. It provided a common basis to facilitate a comparative evaluation of the demonstrations [that were to follow during the selection process]. Hence it set out the background to the project and the intention of RACVI to provide a more efficient service for clients by approving the claims process and developing and implementing a computerised workflow management system. It stated time goals to be attained in handling claims. It described the existing system and with diagrams and narrative referred to a new proposed process."

Section 4 of the RFI was concerned with technical requirements and under the heading "Image Service Functionality", the environment which the selected vendor "must provide" included the following by way of minimum functionality:

"On-line storage of all current, active claims, and near-line storage of all inactive claims within a predetermined time frame".

On 4 May 1993 Unisys presented its demonstration to the New Claims Process Team and this included a demonstration of document retrieval. As the judge found[9], at this demonstration document retrieval was "virtually instantaneous between command and appearance"; indeed Josephson himself used the term "sub-second" in relation to the retrieval of documents. Again according to the findings, Josephson spoke favourably of Unisys' ability to meet the requirements of RACVI as set out in the RFI and he said that the Unisys product was very flexible, integration of virtually anything was possible, the technology was reliable, stable and easy to use and, being one of the "biggest, best and most experienced vendors in the field, it could provide the system required". A promotional brochure on the InfoImage system marketed by Unisys was handed out. In that brochure there was a statement that a benefit of the electronic handling of documents was "increased productivity through immediate, simultaneous access to complete and accurate file folders holding critical business information"[10].

There were demonstrations by other potential vendors during May 1993. The process extended over 7 weeks and covered 10 vendors. On 27 May, the project team met and ranked Unisys second behind Kodak. On 28 May MacCallum prepared a recommendation for further action. The project was to proceed to the next stage with a short list of 5 vendors, including Unisys. MacCallum then drafted the request for proposal ("the RFP"). This was discussed at a meeting of the team on 8 June and, according to Lang (a claims services manager with RACVI) whose evidence was accepted in this respect by the judge, it "accurately reflected RACVI's business requirements". For present purposes the most important business requirement, as the judge noted, "was the ability to quickly and reliably retrieve files and documents"[11]. The RFP was sent to Unisys and the other vendors on the short list on 10 June 1993.

The RFP extended to some 66 pages and consisted of five sections and seven Appendices[12]. As with the RFI, the intention was to have responses on a common basis in the sense that each vendor was required to provide the like information in a particular format, which would facilitate evaluation. Thus in Section 3 [13] it was stated that the response "must" take the form specified and that all requirements in Section 4 and Appendix C "should" be addressed, stating whether the response fully, partially or not at all complied or the requirement was not relevant to the vendor's solution. The response was to "adhere to the format and tables provided in the appendices of this proposal". It was further stated:-

"Any need to reference additional information can be done by placing a note in the `Comments' field of the tables, against the appropriate item, and provide the information in the `Product, Support and Services Material' section."

Section 4 of the RFP[14] set out in some detail the requirements which the system had to meet. Under the heading "Image Service Functionality" commenced[15]:

"The Image Services environment must provide the following minimum functionality: 1. Image capture (scanning and committal) of varying documents, including double sided multi page at rates in excess of 30 pages per minute for scanning in the mail room". . .

and, as the list continued, this was plainly stated: -

"14. On-line storage of all current, active claims, and near-line storage of all inactive claims within a pre-determined time frame. ... 18. Image caching at client and server."

(It should be noted that in stating these minimum requirements the word "must" appears and in Section 3 it had been declared that "must" indicated that the requirement was "mandatory".) It was also stated in Section 4, under the heading Configuration[16]:

"Vendors must configure adequate storage capacity, magnetic and optical, to implement on a "date forward" basis, retaining all open claims on-line, all claims near-line for 3 months following closure and closed claims older than 3 months off-line".

In Section 5 [17], a time table was set out which ended with systems being totally operational by the end of June 1994. The definitions in Appendix A[18] included these:

"Near line Image access from local optical storage. Expected response times in the vicinity of 20 seconds maximum. On line High speed image access from local server cache, magnetic storage or on-line optical drive. Expected response times in the vicinity of 2 to 4 seconds."

Having received the RFP, Unisys set about readying itself to respond and plan the prototype demonstration. On 5 July 1993, and after considering both the RFI and the RFP (sometimes in meetings with RACVI personnel and MacCallum), Unisys submitted its response in writing to the RFP ("the July response")[19]. It was a long document, nearly 120 pages long (much of it devoted to Section 4), and with a number of brochures and printed information, including the 1992 Annual Report of Unisys. As required by the RFP, the response directed itself, in sequential sections, to the RFP.[20]

It is convenient now to set out some of those parts of the July response upon which argument for the appellant focussed. First, in the Executive Overview which preceded the body of the document, it was stated[21]:

"As Unisys has had limited access to RACV, this document does not represent a firm recommendation nor final proposal to RACV Insurance. Unisys strongly recommends, however, that RACV and Unisys undertake a joint Requirements Definition Study in order that Unisys may gain a comprehensive understanding of the RACV Insurance requirements, and so deliver a proposal based on a clear financial business case."

There was none the less a quotation of the price to RACV, $3,304,324, which included the supply of hardware and software. The response to Section 4 of the RFP commenced[22]:

"Unisys is responding to the RACV RFP document as it has been presented and in the context of several brief conversations with RACV personnel and consultants. Unisys has not had the opportunity to assess the detailed requirements of RACV. This document is a direct response to the specific points and questions raised in the RFP document, as requested by RACV. The format of this response is that requested by the RACV."

On page 2-10[23], the July response described the role of optical and magnetic storage in this way:

"As a rule, magnetic storage - often called "cache" - is used to store information that is needed on a more immediate or frequent basis than some other information. An example is a current, open claim compared with a closed, archived claim. The Unisys InfoImage system automatically keeps information that is in current use in the cache area. Information that is archived is sent to the permanent optical storage disks. As well, some housekeeping data used by the system is not required for long term storage, and is only kept on a temporary basis on magnetic cache." [Emphasis added]

It was stated under the heading "Case and Folder Management Functions" that the InfoImage Folder product provided "the facilities to assemble groups of related documents into electronic `Folders'"[24], the term "Folders" being further described a little further on[25].

Importantly, under the heading "Image Service Functionality" in Section 4, Unisys dealt with the several points raised in the RFP (including the above-quoted points 14 and 18). In respect of point 1, Unisys declared inter alia[26]:

"1. Unisys recognises that work flow and image management of the scope envisaged by RACV is a matter of effective transaction processing. Unisys has thus configured a system which can comfortably handle the committal and retrieval rates expected from the volumes outlined in the RFP and Vendor Symposia Briefing documents."

As to points 14 and 18, this was said[27]:

"14. Unisys InfoImage supports fully automatic document management including the use of caching on both servers and workstations. The system configuration in this response allows for the on-line and near-line storage requirements outlined in the RFP document. Document management is more fully outlined in response to Question 18, below. 18. Unisys InfoImage utilises caching techniques at server and workstation to optimise user time. Movement of information around the system is fully automated, without the need for conscious user intervention." [Emphasis added]

Under the heading "Cache Services", four types of cache were identified: batch cache, page cache, print cache and user cache[28]. This portion of the RFP then continued:

"Retrieved images requests are ageable so that the oldest (least recently used) objects can be deleted from cache as the system requires space for new entries."

"Batch Services, Document Services, and Print Services utilise Cache Services as a temporary storage area. The configuration of a cache (including its maximum size) is done during system set-up.

"Document Services" were described as "responsible for the migration of images between cache and optical disk and the committal and deletion of documents".

On page 2-43[29], which was the last page in Section 4 of the July response, it was stated under the separate heading "Configuration":

"The configuration included in this response has been provided exclusive of a Requirements Definition study by Unisys. Therefore, Unisys does not propose that this configuration is recommended to meet the needs of RACV. Should the results of the Requirements Definition study concur, however, this configuration may be the most appropriate for RACV."

That the final articulation of the proposal made by Unisys was to await the Requirements Definition Study was confirmed towards the end of the July response. On page 10-1 and under the heading "Product, Support and Services", Unisys stated that its response had been prepared in accordance with its understanding of RACV's requirements and that responsibility for the final determination of the suitability of what Unisys proposed lay with RACV[30], and added[31]:

"Unisys will submit a firm proposal subsequent to the completion of an agreed Requirements Definition study by Unisys and RACV."

The concluding paragraph on page 10-1, about which much was said on the appeal, is dealt with later.

Much of RACV's case of misrepresentation depended upon the July response of Unisys to the RFP (and in particular the passages I have emphasised). At trial there was some debate about whether the RFP was or was not part of the contract; the July response plainly was. In the end it does not matter, for on appeal it was accepted by counsel for the appellant that, whether part of the contract or not, it was necessary to refer to the RFP to make sense of the July response. After all, Section 4 of the July response contained explicit representations that in the InfoImage system "information that is in current use" was kept in cache and (in respect of item 14) that the "system configuration in this response allows for the on-line and near-line storage requirements outlined in the RFP document".

To continue with the narrative: all of the responses to the RFP were assessed by MacCallum on 19 July 1993. Unisys was ranked behind Tower and the other 3 vendors, but there was still to be the prototype demonstration. On 21 July there was a meeting for the purpose of enabling Unisys to obtain further information concerning the claims process, in order to facilitate the pending presentation; and on 4 August 1993, there was a demonstration conducted at the office of Unisys.[32] Again, according to notes made by Chesson-Kistas, Josephson said in introductory remarks that Unisys believed it could meet RACV's requirements and would demonstrate that. He recapped on what was in the July response and said that the InfoImage system was ideal for RACV's requirements and would meet its business needs. Document retrieval, as demonstrated, was "virtually instantaneous" and within 2 to 4 seconds. Josephson said that Unisys could provide the system RACVI required and that response times demonstrated were "indicative" of response times that could be expected for on-line and near-line claims on an installed system. He said, too, that the system would meet the requirement for timely customer inquiries. Olsen, an expert within Unisys, told Chesson-Kistas (she said) that the system was flexible and could meet RACVI's needs.

Importantly, according to the witnesses no Unisys representative made any qualification as to response times or volume capacity. According to the witness Lang, the substance of what Josephson said was that RACV was seeing what it would get and Josephson acknowledged (he said) that Unisys understood RACV's business requirements as set out in the RFP. Nor did Josephson, in giving evidence, deny saying that the response time demonstrated was indicative of response times that the RACV could expect for on-line claims. According to him, however, he added that this could be expected only for on-line retrieval (that is, retrieval from cache) and would not be expected from optical disk. His recollection was that retrieval from PC cache (and I emphasise PC cache) was sub-second, was 2 to 4 seconds from cache on the server, and up to 20 or 30 seconds from optical disk off-line. From optical disk on-line, it would be much quicker, he said, depending on the activity on the system. Whether all this happened in May 1993 or not was a question at trial: on one view it happened in August, but again it does not matter for present purposes. What has been described is obviously not inconsistent with the representations alleged by RACV. That is what is important.

In late August, RACV evaluated the tenders and ranked Unisys second behind Tower, taking account of cost. Negotiations began with Tower but broke down and in late September Unisys was invited to submit a final proposal[33]. Unisys submitted a further response to the RFP in October 1993 ("the October response")[34]. Again a price was provided, this one lower than that earlier provided. As found by his Honour[35]:-

"The reduction in price was consistent with Willis' evidence that she had been told to submit a second response on the basis that Unisys would charge $1.7M, which was `far below our initial costing'. She said, and I accept, that Unisys had decided to forgo its normal gross profit margin on the deal, as it wanted to buy imaging market share and wanted an insurance reference site in Australia."

To that end Unisys proposed smaller servers which reduced hardware costs. On 15 October, Unisys made a further presentation to the team and MacCallum, at its office. Josephson, in his opening remarks, said that Unisys had done this type of project before and was ideally suited to meet RACVI's needs.

On 9 November 1993, the board of RACVI approved a capital budget of some $4.1 million , 75 per cent of which was for, inter alia, a "workflow management system (including imaging)"[36]. As the judge found, this was subject to approval by management and in that regard the final decision was left to Mitchelson[37]. In November, the project team met to determine a final recommendation and they selected Unisys as the preferred vendor. This was based on their evaluation of all the information that Unisys had provided. An executive briefing followed and each person involved made a presentation. In a letter dated 26 November 1993 MacCallum stated that Unisys had exhibited the best solution and great understanding of the business requirements of RACVI - and this was virtually MacCallum's "sign off" on Deloitte's retainer. Mitchelson, however, wished to be satisfied for himself that the system would perform the tasks in question and would meet RACVI's business needs and so Unisys was asked to demonstrate the system to senior management.

That demonstration occurred on 2 December 1993 and it was in effect, said the judge in his findings[38], an enhancement and repeat of the previous demonstrations. According to Mitchelson, Josephson again said that the InfoImage system was suitable and ideal for RACV's purposes. Mitchelson was impressed: the system looked what was needed to streamline RACVI's claims processes and meet its business requirements. Mitchelson said that Unisys personnel expressed no qualification about its ability to produce the system required by RACVI, including response times. That is not to say that there was any statement made regarding response times, but by the same token no qualification was put on what was proposed in that regard. The judge accepted Mitchelson's evidence[39]. Lang's evidence, too, was accepted by the judge: he believed that what he had seen in relation to instantaneous response times was what RACVI would get if the system was installed. He said that both Josephson and another had told him that "he would get what he had seen".

The matter then proceeded towards contract. Unisys prepared a Project Management Plan for the project ("the PMP"), a document of some 42 pages, dated 23 December 1993[40]. It states that it was prepared on behalf of Unisys Information Services based on details as provided in the RFP. It was incorporated as one of the contract documents. The purpose of the PMP was stated as being to present a statement of all deliverables, work allocations and responsibilities, an overview of resources, schedules and costs and so on. The appellant emphasised that under the heading "Major Gaps Identified" there were 15 matters, identified by Unisys as "missing or deficient for the purpose of progressing to detailed design and implementation"[41]. Of the 15 matters, one (point 12) was "Performance objectives" and Unisys relied upon that in its defence.

The contract was signed by RACVI and Unisys on 24 December 1993. Both the July response and the October response to the RFP were included, together with the PMP and a schedule of payment milestones. These responses were to be read subject to the contract terms, but there is no need to refer further to the contract. The trial judge did not decide the case by reference to the contract and whether or not the alleged representations became part of the contract does not now matter. That is because appellant's counsel on appeal expressly disavowed a submission that a representation which became a term of the contract could not thereafter be relied upon as a representation, founding misleading or deceptive conduct. As Mr. Bathurst put it, "section 52 has gone too far to allow that submission". Accordingly, there is no need to consider further the contract terms: this case was resolved by reference to misrepresentation and in my opinion it is on that basis that the appeal can be determined.

Work began on the project in January 1994. In March 1994 RACV produced what it called a Functional Specification but Unisys regarded that as inadequate and prepared, in its place, a document entitled Workflow Narrative to fill the gaps in the information it had[42]. The Workflow Narrative was the result of Unisys and RACVI working together, the one providing input to the other, although, as is apparent from the witness statements filed by the appellant, Unisys was not impressed with the degree of assistance afforded to it by RACVI. Be that as it may, the product of their joint efforts was dated May 1994 and its purpose, as stated, was "to clearly outline the functionality of the workflow management system".[43]

A little more should be said about Workflow Narrative because of the earlier references in documents passing between the parties to a Requirements Definition Study as well as to a Functional Specification. The likely role of such documents was explained by Mr. Wall, one of those within Unisys most closely associated with the RACV project. In his witness statement, Wall said[44]:-

"Normally, when I had worked on projects of this type, a requirements definition study had been conducted to identify the business processes which were to be automated by the system. When developing the automated mortgage processing system for Allied Irish Bank (`AIB') in Dublin, for example, the requirements definition study took me and another experienced consultant, in conjunction with two mortgage processing specialists from AIB six weeks to prepare. The requirements definition study which we prepared contained accurate descriptions of AIB's business processes, so that technology could be applied to those processes. The study was then translated into a functional specification document which detailed the functions which the system was to perform. A detailed design was then produced from the functional specifications, which provided a technical description for the designers to develop the system."

According to Mr. Wall, it was "to bridge the gaps in the business process as described in the RFP" that Olsen and he "took on the task of preparing in conjunction with the RACV team a document which we called the `Workflow Narrative'"[45].

To return to the narrative: the judge accepted evidence that the Workflow Narrative had "very significant value because it formed the foundations for the scope of the functions of the system which Unisys was to deliver".[46] It is obvious from what followed that Unisys, as the judge found, "was prepared to move forward under the contract on that basis", the Workflow Narrative apparently superseding not only the Functional Specification but also the antecedent Requirements Definition Study. Indeed, we were told that, although argument was initially advanced at trial that the failure of RACV to produce a satisfactory Functional Specification was at least a cause of the subsequent failure of the system delivered by Unisys, that argument was in the end abandoned.

Work proceeded, then, on the detailed design of the system and on 27 March 1995, following testing, an acceptance sign-off was signed on behalf of RACVI and on behalf of Unisys. The acceptance constituted an acknowledgment that the system had been delivered in accordance with the Workflow Narrative and the PMP, subject to certain exceptions. It further stated that in lieu of the warranty period stated in the contract, the warranty period would commence on 27 March and terminate on 19 May 1995. The matter being apparently complete, on 30 March Deloittes rendered their account[47].

The first day of production using the new system was 12 April 1995 and functionality and performance problems were encountered from the outset. They became worse in nature and magnitude as the number of users and volume of material on the system increased. Retrieval times were one minute or more and there was a problem with stability in that the system would sometimes stop processing requests for retrieval or lock up completely. Sometimes there was data loss. And the situation got worse. From time to time the system simply crashed. The judge said (and it is not now disputed)[48]:

"The warranty period passed, but that was hardly relevant. Unisys had delivered a system which did not work. Unisys was seeking to stay with the project and, it is clear, sought to avoid RACVI terminating their further involvement. Unisys searched on for reasons for the failure of its work. For that is what it was, a failed contractual performance, as counsel conceded at the trial."

One point of procedure may be noted at this juncture. The appellant below maintained in its pleadings that it had not failed to deliver in March 1995; the concession that it did fail to perform contractually came belatedly, during the trial, albeit at an early stage in the trial. The judge was critical of this, saying[49]:

"It is important to point out here that at an early stage of the trial counsel for Unisys conceded that in fact the system delivered in March 1995 failed, and had to be re-designed before it could function. If the facts had been . ..admitted in the defence the plaintiffs might properly have required particulars of the respect or respects in which, or as to the reason or reasons why, the system failed and had to be re-designed. By not admitting in the defence what was an undeniable fact, as counsel conceded, Unisys avoided committing itself on the issue pre-trial. The consequence of the denial of the issue in the defence, was that the pleading was misleading and obfuscated on an important issue. Unisys was thus enabled to avoid giving particulars of the fact. It was an important issue for the light it might have thrown on what had happened, and the admission and particulars would have assisted the plaintiffs pre-trial. It would also have saved time at the trial. In any litigation, but in particular litigation such as the present in which the costs must be enormous, the course taken by Unisys was unacceptable and is to be recorded with condemnation."

I agree. Without attempting to attribute final responsibility, it does seem that in some respects the course taken by this litigation is to be deplored. Indeed the conduct of Unisys in so pleading as to obfuscate the issues arising, particularly from its non-performance in March-May 1995, is possibly relevant on more than one count.

To conclude the narrative, work then began by Unisys on reconfiguring the system to see if it could be made to work. Parts were replaced, tests were conducted and changes were made, some at the behest of RACV. Some at least were agreed by means of System Change Requests (or "SCRs") and there were many; and, while an attempt was made by Unisys at trial to blame on these SCRs its subsequent failure to get the system working by 24 June 1996 and so to paint RACV as the architect of its own disaster, that attempt too, we were told, was ultimately abandoned. After all, any SCR had to be the subject of agreement between the contracting parties, for each operated as a variation; and had Unisys seen any one or more of them as critically damaging to its ability to complete, no doubt it could and would have refused agreement. In the end, the system was still not up and running by 24 June 1996 (the date fixed between the parties when RACV's patience became exhausted) and this precipitated RACV's terminating any further work on the project and suing for damages.

Unisys contended, however, that such termination of its labours was premature, despite the passage of more than 12 months since delivery was first made under the contract. It put the case (at trial and on appeal) that by 24 June 1996 the system could have been made to work had further time been given. Counsel argued that, because of what might have been achieved, it could not be said that the system, if implemented in June 1996, would not have worked (or worked sufficiently to have met the requirements of RACV); nor could it be said that RACV had suffered the losses claimed, for those losses could have been largely avoided by its taking relatively simple steps, such as adding more cache at a cost of less than $40,000. Further, while maintaining that representations had not been made as pleaded, counsel for Unisys argued that the correctness or not of any representations made in 1993, or indeed the existence or not of reasonable grounds for making any such representations as were made in 1993, could not fairly be judged by reference to events in June 1996, if only because of the numerous design changes made to the system in the intervening period - an argument which made the SCRs relevant once more, albeit not to causation.

The primary issue argued on appeal was that the judge had erred in finding that the representations had been made as pleaded. In my opinion, there can be no doubt but that the representations were made as alleged. What I have set out thus far is more than enough to justify the conclusion reached by the trial judge, after a careful consideration of all of the evidence before him including the documents, that the representations were made, and indeed were relied upon. Rather than canvass the reasons given by his Honour, I deal with the arguments put on appeal as to why those conclusions were wrong. But before doing so I say something more about the nature and configuration of the computerised system at base.

The Unisys InfoImage system incorporated hardware and software from other suppliers as well as Unisys' own proprietary components, one of the latter being an item of software known as "Event Manager" for routing and retrieval of images in the system. The system Unisys proposed for RACV was to be custom built for RACV purposes; the appellant emphasised that it was not an off-the-shelf system that could simply be installed and made to work to RACV's requirements. The basic system, it seems, employed both magnetic cache (in both server and local PCs) and optical disks on which the images were stored. Simplifying it in its outline of submissions, the appellant described the magnetic cache in the server as being "like the short term memory of the system" and the optical disks as "the permanent or archival storage area". The disks were arranged in a juke box which contained a reader and, in order for images to be retrieved from an optical disk, the disk had to be under the reader in the juke box. If, by chance, the disk was not already mounted under the reader, the mounted disk was removed and the disk from which the image was sought to be retrieved was placed instead under the reader. In the system as configured for RACV there were two juke boxes and each had five disks. Now, images that were in cache or which happened to be on an optical disk that was for the time being mounted under a reader were said to be "on-line": retrieval was speedy. Images stored only on an optical disk which was not mounted under a reader were said to be "near-line" and because of the time taken for the mechanical process of mounting a disk under a reader, retrieval of near-line images took considerably longer than retrieval of on-line images. That, it seems to me, is the crux of this case: for RACV was wanting retrieval times of 2 to 4 seconds for claims still current, while allowing a longer retrieval time (up to 20 seconds) for claims which had ceased to be current during the last 3 months. After that, RACV was content that the claims should be "off-line" and off-line meant on optical disk, even to the stage of the optical disk being stored in a cupboard. Optical disks, we were told, could not be re-written: on the other hand what was in cache could be displaced when overwritten by new material arriving.

At one stage on appeal the appellant submitted that the trial judge had fallen into error as to the system itself. His Honour thought, mistakenly it was said, that a document which was introduced into the system went first into cache and only when it was displaced in cache on to optical disk. That, the appellant pointed out, was not so. What happened was this: when a document was introduced into the system for the first time, it went into cache and at about the same time it went on to optical disk, so that for a time a document coming into the system was both in cache and on disk. It remained on optical disk forever; it remained in cache only until it was overwritten and so displaced by further incoming material. The argument that the judge misunderstood the position was based upon the following sentence in the reasons for judgment[50]:-

"Olsen agreed that under this system a current claim had to be in server cache or in an already mounted optical disk to be on line."

But this was no mistake: his Honour was simply pointing out that a claim was on-line if it was in cache or on an optical disk under a reader - and that was so. The judge did not say, or mean, that the information was on optical disk only if it was not in cache. No misunderstanding was demonstrated; nor did such a "misunderstanding" lead the judge (as the appellant submitted) into mistakenly characterising the system as "near line or archival" when it was not.

So far as now relevant, the essence of the representations relied upon by RACV was that the system proposed by Unisys for RACV would, if implemented, have all active, current claims "on-line" and would provide a retrieval time for such claims of 2 to 4 seconds. It was common ground, at least on appeal, that the second would generally follow from the first, for to be on-line effectively meant to be in cache on the server and to be in cache allowed for very speedy retrieval. The appellant contended, however, that such representations, if made as alleged in 1993, were made by Unisys not in respect of the system that was proposed for implementation by RACV, but rather as to the system being marketed by Unisys, the system known as the InfoImage system. In other words, any representations made in 1993 were merely as to the prototype being demonstrated at the time. Accordingly (the argument ran) if made at all, the representations were as to present fact, not future outcome.[51]

In my opinion the submission has no substance. If it were correct, the utility of the elaborate procedure that preceded the making of the contract would be pretty well lost altogether. In a different context, suppose that a person seeking a car has a car salesman demonstrate to him a particular make and model in the course of which the salesman represents that that particular make and model performs in a certain way. Suppose, too, that, relying upon the demonstration, the customer purchases a car of that make and model. Could it seriously be contended, in the absence of any warning given or saving made, that when the car purchased did not perform according to the demonstration, the vendor would escape liability by asserting that the demonstration related to the particular car being demonstrated, and not to the car purchased by the customer, even if the car sold required some adaptation to suit the needs of the customer? Surely not - and that seems to me to be the case here. It seems to me plain that in context the representations made by Unisys in the pre-contract stage related to the system to be provided for RACV by Unisys, if Unisys gained the contract. Those representations were intended to be relied upon and were relied upon, as the judge found, when RACV entered into the contract with Unisys. That finding is scarcely surprising, given that RACV had spelt out the relevant requirements in the first place.

It is true that much was said in the July response about the need for a Requirements Definition Study and that the configuration "in the response" might not be the final configuration; but that was not enough to warn RACV that what it was seeing and hearing in the demonstrations, and that what it was reading in the July response as to holding images in cache and retrieval times, was of no relevance to what would be supplied. In the end, Unisys was driven to contending that the contract was not to deliver a certain outcome but only to do the best it could with the system being adapted, but that was rejected by the trial judge and rightly so. His Honour said[52]:-

"The fact is, I find, that a fundamental premise of the engagement, understood by Unisys, was that the system implemented would provide retrieval in a timely manner. So much was obvious as required to meet the business purposes of RACVI. But over and above that was the requirement of an on-line system in which the retrieval times specified in the RFP were inherent. This was understood by Unisys and hence the explanations and representations of Josephson and Unisys' cognisance of the importance of retrieval time signified by the above references. I reject the submission that under the contract Unisys was to deliver, and RACV was to accept, a system with whatever retrieval times it might come to deliver."

It is surely significant that, as found below, Unisys was anxious to obtain the contract with RACV; Unisys was representing that it could deliver a system to meet the needs of RACV which were articulated plainly enough in the RFP; and if Unisys later found it more difficult to honour the representations than it had at first expected, that does not absolve it from liability once the representations were acted upon. Indeed the submission on mitigation, namely that the system could have been fixed to answer the needs of RACV by the relatively simple addition of more cache, seemed to me to cut across the claim by Unisys that it was to do no more than its best to procure a result. Of course the construction of the representations made in 1993 cannot be determined by the result that might have been close by June 1996, but it does ring rather hollow for Unisys to submit that, despite the possibility that emerged (as it now would have it), Unisys was representing in 1993 (and indeed contracting in December) not in respect of any outcome desired by RACV but only about its using best endeavours in that regard. The submission is rejected.

The representations were not qualified or withdrawn

Counsel for Unisys then turned to more particular provisions within the contract documents in an effort to establish that, if any relevant representations were made, they were expressly withdrawn. Again it is perhaps curious to find Unisys denying that it made any representation as to future outcome and then contending that any representation to future outcome was expressly withdrawn or at least qualified in a material respect by the contract; but, if there be inconsistency, it does not matter because I am clearly of the view that the judge was correct in dismissing the argument anyway.

First, it is necessary to say something more about the precise source of the representations relied upon by the plaintiffs below. As described by the appellant in its initial Outline of Submissions[53]:-

"The system proposed by Unisys was its InfoImage Folder system as described in the July response to RFP, the October response to RFP, a Unisys brochure [in evidence] and as demonstrated in prototype demonstrations in May - December 1993".

As pleaded by RACV, the critical representations were made partly in writing (in the documents just mentioned) and partly orally (at the demonstrations that were given), although, as the appellant pointed out in argument, no reliance was placed by counsel in final address upon the December demonstration, which was that required by Mitchelson shortly before the contract was entered into. An effort was then made on appeal to establish that nothing said at any of those demonstrations (including that in December) amounted in itself to a clear representation either that all current, active claims would be on-line or that retrieval times for such claims would be in the order of 2 to 4 seconds - with the result, counsel submitted, that evidence of those demonstrations could not support any finding that the representations had been made, or made as pleaded.

The demonstrations in question were, however, demonstrations of speedy retrieval times, "virtually instantaneous" according to the witness Chesson-Kistas when describing the August demonstration[54]. In short, even if it be correct, as Mr. Bathurst carefully submitted, that nothing was proved at trial to have been said which might amount to the representations now relied upon, what was demonstrated was certainly not inconsistent with the representations as found in the July response to the RFP. Mr. Bathurst then pointed to the fact that more than one witness recalled its being said at the August demonstration that the response times shown were "indicative"[55], indicative (he submitted) of what could be achieved rather than what would be achieved. But the demonstration was to inform the customer and the distinction sought to be drawn seemed to me of no ultimate significance. In the end, as I followed him, Mr. Bathurst accepted that RACV had not relied upon what was said in the course of the demonstrations as proof in itself of the representations relied upon: rather RACV had relied upon the documents and the demonstrations in conjunction to sustain their claims. And the course of the demonstrations and in context what was said, however it be approached, did, I think, sustain those claims, when considered with the documents.

The absence from the demonstrations of anything actually said in repetition of the representations extracted from the documents was then used by Mr. Bathurst to pave the way for his submission that, in the documents themselves, the representations were withdrawn. This submission turned on the last paragraph on page 10-1 of the July response under the heading "Terms and Conditions" and point 12 on page 7 of the PMP. As to the first, the last paragraph on page 10-1 of the July response read as follows:[56]

"Unisys is not at this stage able to commit to any response time or availability levels as set out in this request for response as we will require further information regarding response time criteria including number of users, definition of what is being measured etc. With respect to availability we will need to review availability in the context of the support plan being contracted for and the configuration over which availability is being measured. Any response times or availability levels which are finally negotiated will only be measured on the system we are contracting to supply."

The judge was persuaded that, because of its context, the reference in this passage to "response time" was not to the time required for RACV operators to retrieve computerised images in the new system, but to the time taken by Unisys to respond to service calls and the like. It will, however, be apparent immediately that there is some mistake in the passage quoted from page 10-1. The expression "in this request for response" cannot be correct. The judge was disposed to read it as "in the request for the response", meaning the RFP, a possible construction if only because the July response had to be read in the light of the RFP if sense was to be made of it. (Before us, this use of the RFP was not disputed). The difficulty with this construction lies in the conjunction of "response time" with "availability", because the latter apparently referred to up-time, the reverse of down-time (that is, the time during which the computer system would be available for use and not out of use because of malfunction or otherwise) and that was a topic canvassed in the July response itself rather than in the RFP: it was to be the subject of some firm commitment by the supplier. The same was true of the response times (if meant in the sense preferred by the judge) and accordingly it is not at all difficult to take the expression "in this request for response" as a mistype of "in this response". That is how I would read it.

When so read, the judge's conclusion about the meaning of response times is simply reinforced, I think: it is even more obvious that the response times mentioned related to service calls and the like and not at all to the retrieval of images. In this regard, it is significant that, under the heading "Terms and Conditions" on page 10-16[57] in the same section of the document, the expression "response times" is found again (albeit this time in the plural) and in a context where it plainly refers to the response to calls for aid, not retrieval times. It is supported, too, by the references to response times in Unisys' own brochure relating to the sort of service that could be provided by way of back up.

When argument was being addressed on behalf of the appellant, it seemed to me that the most that could be said for the construction sought to be imposed by the appellant on page 10-1 was this: that in the RFP RACV had constructed its own dictionary in respect of "response times", and the July response was merely using the same dictionary; that is to say, in the RFP, "response times" could relate to the retrieval of images and so it was on page 10-1. The difficulty with the argument is that the proper construction of the passage on page 10-1 is at least very uncertain and in my view uncertainty is not enough to establish that the representations, if made, were withdrawn. That is sufficient in itself to justify the conclusion, reached by the judge, that the passage on page 10-1 was not such as to withdraw the representations as to 2 to 4 seconds.

The second ground on which the appellant argued that any representation made by Unisys in respect of open claims being on-line or as to their retrieval times was withdrawn in the contract documents, turned on point 12 on page 7 of the PMP[58]. Point 12 was simply "Performance objectives" under a heading "Major Gaps Identified". The judge had little difficulty[59] in rejecting this as amounting to an express qualification on or withdrawal of the representations earlier made, and I agree with what his Honour said. The words relied upon are vague and imprecise; they might or might not have been meant to include retrieval times such as those specified in the RFP for open claims. (I simply note in passing that often the argument before us seemed to overlook that retrieval times were specified not only for current, active claims that were to be on line, but also for near line retrieval.) More information from RACV was obviously still wanted by Unisys at the time when the contract was signed, if only because the task of a more precise configuration of the system for RACV lay ahead. The matters outstanding, it will be recalled, included the Definition Requirements Study and the Functional Specification, both of which, as matters turned out, were to be supplanted in the end by the Workplace Narrative. In my view, the conclusion of the judge was correct; for the reasons given by his Honour, point 12 was insufficient to amount to a qualification on or withdrawal of the representations otherwise made.

Thus far I have agreed in the judge's conclusion that in 1993 Unisys was representing to RACV that, if implemented, the system that it was proposing for RACV would have all current, active claims on line and that in respect of such claims retrieval times would be in the order of 2 to 4 seconds. In my opinion the judge correctly rejected the submission that the representations were not made, or were not made as pleaded, being rather about the prototype InfoImage system than about the outcome desired by RACV (and indeed stipulated for by it in the RFP); and correctly rejected, too, the submission that the representations were qualified by what was in the documents or were indeed withdrawn. The representations were made as pleaded and yet the outcome was never as represented. It did not come about in March-April 1995 when the system simply collapsed and it did not come about by 24 June 1996, the new date for implementation agreed between the parties, by which time RACV's patience had run out. To my mind there can be no doubt at all about RACV's reliance upon the representations, for in part because of them it gave the contract to Unisys. But a number of matters were raised by Unisys in argument against that conclusion and I turn to those now.

First, appellant's counsel bravely submitted that there was no direct evidence of reliance in that the two witnesses who spoke about retrieval times (Chesson-Kistas and Lang) were not experts and the person having the final decision on behalf of RACV, Mitchelson, did not himself recall relying upon retrieval times, according to the evidence he gave at trial. But there were some six or seven people in the New Claims Project Team and it was the team that made the recommendation upon which Mitchelson acted. That recommendation was obviously based upon all the information that the team had at the time - information which included the RFP and the July response, as well as what had been demonstrated by Unisys to RACV over many months. Reliance, I should have thought, was plain beyond argument once RACV gave the contract to Unisys.

None the less, the appellant's contention was that there was no evidence, or no sufficient evidence, that RACV relied upon the misrepresentations, even if made as pleaded. It was submitted that Mitchelson himself relied on the prototype demonstration (that he himself had required) on 2 December 1993, shortly before the contract was made - which was in contrast (it was argued) to the judge's finding that the misleading and deceptive conduct was constituted by representations made, in writing and orally, down to and including the August demonstration but not, it seems, including the December demonstration[60]. Nor (counsel pointed out) was there any finding of misrepresentation by conduct (as distinct from words)[61]. Given, then, that there was no finding of misleading conduct (within the meaning of s.52) at the demonstration on 2 December 1993, and no discussion either of response times[62] on that occasion, it followed (the submission ran) that when Mitchelson relied upon the prototype demonstration on 2 December 1993 he was acting independently of the misrepresentations made earlier.

In my opinion, the argument is altogether artificial. More than once in submission on this appeal, the appellant was apt to stress a particular occurrence or a particular event, isolating it from its surrounding circumstances. This is an example. The December 1993 demonstration was but part and parcel of a continuing process which led to the making of the contract. That process commenced with a request for information in March 1993; it advanced to the RFP and that flowered into the July response and the subsequent response of Unisys in October 1993, in turn supplemented by the demonstrations in April and August 1993 and, I would add, in December 1993. The representations were, as I have said, to be found in the written documents, but nothing in the demonstrations (and I mean all three demonstrations) ran counter to the representations found in the documents. Indeed, what was seen in the demonstrations served to confirm and to reinforce what was in writing, if only because what was seen in the demonstrations was consistent with the writing. As Mr. Justice Ormiston remarked during argument, if the representations about current, active claims being on-line and about retrieval times in the order of 2 to 4 seconds were not an issue, why should one expect any specific reference to be made to such matters during demonstrations which were themselves not inconsistent with what was in writing. That Mitchelson, in relying on what he saw at the demonstration of December 1993, was in some way putting aside altogether all that had preceded it is a submission I reject. It is plain that Mitchelson, in reaching a decision in December 1993 for RACV to go with Unisys, acted upon all that was before him, including the recommendation from the project team.

That team consisted of Lang, Chesson-Kistas and representatives from RACVGS, and MacCallum, the representative of Deloittes, was acting with them all. Counsel sought then to gain comfort from an opinion expressed at some stage by MacCallum that it was not reasonable to require retrieval times of 2 to 4 seconds, but, again, that was only a part of the evidence and, as will be seen, cannot be determinative. Indeed, the appellant's argument on this appeal was by and large a re-statement of the arguments put below which had been unsuccessful, with the result that we were taken to various parts of the voluminous evidence and asked to reverse findings made below on the basis of what we were shown. While such an approach to the evidence may be appropriate at first instance where the trial judge is fully alive to all that has been put before him in the way of oral testimony and documents, it is scarcely satisfactory on appeal, especially when the trial below occupied more than two months and there is such a vast amount of transcript and documentary evidence. No doubt each side attempted before this court to grapple with the points made by the other as best it could but the appellant's arguments could succeed only with difficulty, unless they were especially compelling. In this case, none was compelling. Moreover, by referring only to parts of the evidence, the argument was apt to impart a "spin" on what was said in the transcript or to be found in some particular document - and that was so in respect of MacCallum's expression of opinion about retrieval times.

Although RACV's consultant during the project, MacCallum was called as a witness by Unisys. As it was put in the appellant's initial Outline of Submissions[63]:-

"MacCallum gave clear evidence that he told RACV that it was not reasonable to expect response times of 2-4 seconds for on-line image retrieval and that reasonable response times were 8-12 seconds".

The source of this was said to be the transcript at AB6 2412-3 and AB7 2423, 2428, 2429 and 2491. Much of this was discussion over MacCallum's report of December 1992 (a year before the 1993 demonstration upon which Mitchelson is said to have relied), a report which was in-house to RACV. At AB7 2423, this is recorded:

"What is your best present recollection of what was said? --- What I spoke about with both Roger [Green] and John [Monument] was that reasonable response times in my view were for on line and cache, eight to twelve seconds and for near line, i.e., in the juke box, having to be mechanically transported to the drive, between 20 and 30 seconds. What's your best recollection of what the response was to that statement that you made? --- This is recollection to the best of my knowledge: I believe they referred to the fact that they had done their own research around image retrieval and felt that two to four seconds was still appropriate in terms of response times."

The transcript at AB7 2491 is to like effect. Earlier, at AB6 2413, Mr. MacCallum had answered in like context:

"They [John Monument and Roger Green] insisted on the two to four-second response time after discussion and following concerns aired by myself that that was not a reasonable response time estimate to be putting in anywhere at this point in time."

Thus, when seen in context, Mr. MacCallum's apprehension, that to require response times of 2 to 4 seconds was not reasonable, scarcely mattered: what went into the documents subsequent to this discussion was the response time of 2 to 4 seconds as required by the others. Interestingly, it was Mr. MacCallum himself who wrote the definitions in the RFP, containing the mandated response time of 2 to 4 seconds for on-line claims: AB7 2428.

Then there is Mr. MacCallum's note of 22 October 1993[64] which the appellant relied upon[65]. This, a handwritten document which again was internal to RACV, showed that MacCallum gave a good deal of thought to the capacity of the several systems then under consideration, and among other things considered the size of cache. What was called "the Unisys configuration" was thought to have "the necessary scalability of cache" but the author added:-

"Note: for contractual purposes RACV should address size of cache and number of od [optical disk] juke boxes and drives, with contingency for growth, change or complexity of requirements".

Unisys suggested that the calculations, evidenced by the memorandum, showed how far RACV relied upon its own experts, rather than any representation made by Unisys. To my mind, the memorandum, read as a whole, demonstrates the concern of RACV, through its consultant MacCallum, that the proposed system, from whatever vendor, should meet its requirements. In other words, the concern of MacCallum goes rather to support reliance upon the representations made, than the contrary - thus highlighting the danger of having regard to one or two passages in a document, and one or two documents in a long course of dealing, when there is so much other evidence to which we were not taken.

To sum up: MacCallum's expression of the opinion about the unreasonableness of requiring retrieval times of 2 to 4 seconds was idiosyncratic within the RACV camp. It was a view which he had aired with others and on which he had been overruled. Nor is there reason to think that it was the view of the team when it recommended that Unisys should be the preferred tenderer. It is true that the only members of the team called to give evidence orally were Lang and Chesson-Kistas, but given that their recommendation was unanimous, I see no why all of them had to be called. Mr. Bathurst emphasised that retrieval times might not have been discussed for either of two reasons: either because they were not controversial or because they were not considered of any relevance. He submitted that in order to demonstrate that it was not the latter, it was necessary to call all members of the team, but that is not so; it was Unisys which was seeking to rely upon the fact that they were not discussed and so it was up to Unisys to demonstrate that they were not discussed because they were not considered to be of relevance. The plaintiffs were arguing from a position of strength, because they started with the documents in which the relevant representations were, as I have said, clearly found. No error was shown on this appeal in the trial judge's finding that reliance was shown on the representations that Unisys made in 1993.

That brings me to the correctness or otherwise of the representations that were made in 1993 about the system proposed by Unisys for RACV: the question of falsity, as it was called in argument. There is no doubt but that the system, as delivered to RACVI by Unisys, was a complete failure - or at all events was so when delivered on 27 March 1995[66]. That soon became evident and it is not now in dispute. What more, then, was needed to demonstrate the falsity of the representations, made by Unisys, as to the system proposed, if implemented? Obviously, if the representations were made as pleaded (as the trial judge found), they were representations as to what would happen when the system was implemented. It was delivered, implementation was attempted in March and April 1995 and the representations were not met.

For Unisys, Mr. Bathurst first submitted that the correctness or not of the representations had to be decided at the time the representations were made, and not at some future time, but plainly that would be so only if the representations were, as submitted for the appellant, representations as to the prototype being then and there demonstrated, and not representations as to the system which was to be provided for RACV and implemented thereafter. As found, the representations were of the latter kind, not the former, and accordingly correctness or not fell to be determined at the date, in short, of implementation. That was in March-April 1995 and at that time there can be no doubt that the representations were not borne out. And, as I followed him, Mr. Bathurst finally accepted that the falsity or otherwise of the representations found to have been made was to be determined at the time of implementation of the system, and not in 1993 when the representations were made. (The existence or not of reasonable grounds for the making of the representations is another matter and one to which I return in due course.)

Appellant's counsel then pointed to what followed. He submitted that what occurred in March-April 1995 was but a first attempt, not the final attempt. By arrangement, Unisys was allowed by RACV some further time within which to deliver its proposed solution to the problems being addressed by RACVI; and Unisys set about "fixing the system", as it was put in the appellant's initial Outline[67]. A new project manager, Birthisel, was appointed in May 1995 and a taskforce created to investigate the reasons for the failure of the system. About a month later it was decided, at a project meeting between RACVI and Unisys, "to implement a main frame change to prevent new claims being processed by the new workflow system" and Unisys set about rebuilding it[68]. It is altogether unnecessary to go fully into the sorry saga that followed: much of the judgment below is taken up with describing events during the 12 months period from June 1995, when the rebuilding of the system by Unisys started, to June 1996, when RACV brought the project to an end[69].

As already mentioned, however, it was the appellant's submission (on appeal as below) that whatever difficulties remained in the system by 24 June 1996 were capable of being cured, and cured fairly readily. Thus, according to the appellant's Outline[70]:-

"The rebuilt system was ready for testing in early 1996 and a number of functional and volume and stress tests were conducted in a period February to June 1996. The result of these tests were mixed. A number of functional errors were revealed by the functional tests and these were progressively eliminated. A log of error reports from the functional tests is at AB32/5564-77. It shows that in the period 13 March to 26 June, 272 error reports were raised. By 26 June most of these errors had been resolved. As at 26 June, 54 error reports - representing 27 discrete errors - remained unresolved: AB35/E6408-9. The evidence at the trial ... was that all these matters were capable of being resolved."

In respect of this submission it is worth noting what Birthisel said in evidence about the response times sought by RACV. As narrated in the reasons for judgment[71]:-

"It was not until about March 1996 that he [Birthisel] became aware of the matter of response times. That was when RACV issued a document prepared by Doherty [a manager of RACV working on the WMS] that said he wanted 2-4 seconds. Birthisel said it came from nowhere and he was taken aback. He had been `solely focussed on delivering to RACV the best system the team could put together that would meet the business requirements and basically get through the volume and stress tests'. By `best system' he meant the first priority was to get a stable, functioning system. A little later in his evidence he said that `we always looked at giving RACV the best possible response times that we could get out of the system. We had not up until March looked at timings. Two to four seconds was not even considered'. He had not read the RFP prior to Doherty's notification of response times. He read the RFP in March and then the Unisys responses of July and October 1993. He had not looked at these documents before. He did not recall reading the RFI in March 1996. He had previously read the contract itself and the acceptance test document but neither mentioned response times. He saw the reference to on-line and near line and timing of two to four seconds in the RFP. Having done so he maintained the position that the performance target to meet RACVI's business requirement was to get the best performance he could out of the system when operating as best it could."

This, then, was the position adopted by Unisys: that its job was to get the best it could out of the system for RACVI, and nothing more than that. In itself, that condemns Unisys as accepting that it was unable still, by the middle of 1996, to make the system operate in a way consistent with the representations made before the contract was entered into.[72]

There was much more evidence to like effect. For example there was the letter - "the bleak letter" the judge called it - from Esmonde of Unisys to the Claims Processing Manager of RACVI, dated 14 June 1996, which included this[73]:-

"With the research we have conducted to date, including input from sites in Austria and the US, we do not believe there are opportunities for major improvements to Folder retrievals, due to hardware and software constraints. I understand David Birthisel has discussed the use of pre-fetch processing with you to assist in providing consistent performance regardless of the number of expected users. We will therefore investigate further the use of pre-fetch as a means of sustaining a 20 to 40 second retrieval time. . . ."

The pre-fetch process would call into cache ahead of time a document likely to be needed, thereby reducing retrieval time when it was needed, but of course pre-fetching depended upon advance planning, if not knowledge; nor was a retrieval time of 20 to 40 seconds acceptable for current, active claims. The letter then proposed a technical feasibility study to commence "next Wednesday" and continued:-

"The objectives of the study will be to assess various options, to determine the likely impact on existing functionality and performance and to estimate the effort required to specify, develop, test and implement the solution. Unisys is prepared to conduct this exercise at nil cost. However, before progressing to the development stage, we will need to confirm funding arrangements."

To my mind, this is a far cry from the existence of a satisfactory conclusion that was all but in place by 24 June 1996; Unisys had again simply failed to deliver a system which was working. True it is that some of the errors earlier seen in the system had probably been fixed, but certainly not all.

When narrating events earlier in the judgment his Honour had said[74]:-

"I note the evidence of Chesson-Kistas that on 26 June 1996 Birthisel told her that Unisys was awaiting a response from the United States about the problem of the system producing multiple copies of documents."

Appellant's counsel sought to argue that such was the nature of the beast: any system newly installed was bound to have teething problems which needed fixing[75], but all, he submitted, could have been cured in time. The same submission was put to the judge, who was not impressed. His finding in this regard was as follows[76]:-

"In summary, at 24 June [1996] there were unresolved issues of functionality. It seems that the issues could have been resolved, but as to when and with what effort and cost it is not possible to make a finding. I prefer and accept the evidence of the plaintiffs' witnesses on the issue of functionality and stability of the system. The overall performance of Unisys in seeking to provide a computer system inspires no confidence in its ability to have expeditiously resolved all issues. I do not accept that the situation was that acting reasonably RACVI ought have accepted the system and gone live with it, with the unresolved problems. I do not accept the evidence of Johnson (or other witnesses) as to going live with the problems. Johnson's evidence as to that was quite unrealistic. People had to come from the United States to fix the problems: as that did not happen he could not say if the problems could have been fixed. In his evidence Birthisel confirmed reliance on the United States to fix the outstanding functional errors, assuming they could be fixed. By contrast, Olsen's view was that the system would have been ready to go live in four to eight weeks. I find that the system was not ready to go live on 24 June and that an indeterminate time would have passed before it could have achieved that position. I repeat that it is not possible to find when that might have been. I reject as entirely lacking in credibility, tested against its performance during the contract, evidence of Unisys in that regard. The matter is speculation."

The problems may have been capable of being fixed, but when and how could not be determined. There was no reason why RACV, already more than patient with the inability of Unisys to deliver according to its representations, should have had to wait any longer.

On 27 June 1996, RACVI brought matters to an end by purporting to terminate the contract for non-performance and, although RACVI's right to do so was challenged in terms of the contract, the appellant can have no ground at all for complaining that too little time was allowed it for rectifying a system first delivered more than 12 months previously: more than enough time had been given to it to complete the task. Mindful no doubt that the falsity of the representations could be held to turn on the time for implementation of the system proposed by Unisys for RACV, appellant's counsel submitted that despite the passage of more than 12 months since first delivery the time for implementation had still not arrived, but that cannot be correct. The warranty period had come and gone but, as the judge said, that was scarcely relevant[77]. Unisys was in a period of extension, which the parties themselves agreed was to culminate on 24 June 1996, which was itself not unreasonable. Unisys' failure to perform by that date left it vulnerable to suit, and that is all that need be said. Despite counsel's efforts to introduce other considerations, we are not concerned with the precise position arising under the contract, for this was not resolved as an action for breach of contract. Instead, we are concerned with the time for implementation, in order to determine the correctness or not of the representations that were made, and that time arrived very clearly on 24 June 1996. If on that date Unisys was close to completing according to the contract, that might bear on RACV's obligation to mitigate any loss suffered (as indeed was argued) but it does not falsify the conclusion that the representations made in 1993 were not borne out by events and, subject only to the next consideration I mention, Unisys is liable accordingly for misleading or deceptive conduct.

There remains the appellant's argument based on s.51A of the Trade Practices Act that, if the representations were made in 1993 and made as pleaded (as has been concluded) so that they related to the outcome of the system if implemented and were thus as to the future, Unisys had "reasonable grounds" for making the representations. I do not know that it was seriously in dispute that the existence of those reasonable grounds had to be determined as at the date of the making of the representations and not, for example, in June 1996, the "cut-off" date. But as to reasonable grounds, the onus of proof lay squarely on Unisys and the trial judge rejected its submission that reasonable grounds were shown.

His Honour said that Unisys did not configure the system appropriately to meet the representations made[78] and he found, too, that Unisys never intended to configure the system to meet those representations[79]. Before us, counsel for Unisys submitted at first that the reference to intention disclosed error in that the judge was addressing a representation (constituting misleading or deceptive conduct) as to the intention of Unisys, a representation that was never pleaded. But a representation, made in 1993, that it was the present intention of Unisys to configure a system to meet the needs of RACV was never in issue: the judge found, and I have agreed, that the representations were as pleaded, namely, that the system proposed by Unisys, if implemented, would (in short) answer to the needs of RACV. Those representations were as to a future event, an outcome under the contract of supply, and I do not read the judgment as referring to intention as relevant otherwise than in respect of the plea made by Unisys that, if it made the representations as to outcome, it had reasonable grounds at the time for doing so.

Nor is it misplaced to refer to intention in that context. The question whether Unisys had reasonable grounds for making the representations needs, perhaps, to be extrapolated. First, when the cause of action is for misleading or deceptive conduct based upon a representation as to a future outcome, that seems to pre-suppose that in the result the underlying representation has proved incorrect or, as it was said in argument, false. Be that as it may, the existence of reasonable grounds for making the representation cannot be supposed to depend upon evidence at large: it depends upon evidence that the maker of the representation had such grounds. In other words, once the making of the representation has been established, it depends ordinarily upon evidence that the maker of the representation believed the representation to be correct at the time it was made (that is, was acting honestly in making the representation) and had reasonable grounds for that belief.[80] Obviously, whether the grounds are "reasonable" is to be determined objectively; but whether that belief was held and held on those grounds must be determined subjectively. In that context it was relevant to consider that Unisys did not configure the system to accord with the representations, and never intended (if that was the fact) to configure the system to accord with the representations.

Now while denial of the making of a representation does not per se preclude reliance on reasonable grounds for making it, it must surely make it more difficult for the representor to succeed; the question becomes one of reasonable grounds upon the basis of a finding that the representation was made without evidence of such from the maker[81]. In this case, the appellant relied specifically upon the evidence of Mr. Olsen, the technical expert who was involved for Unisys in the configuring of the system to meet the relevant requirements. At trial Mr. Olsen said he read the RFP and made reference to retrieval times of 2 to 4 seconds for claims on-line. He said, too, that he believed it to be achievable[82]. This was evidence that Mr. Olsen was aware of the requirement for retrieval times of 2 to 4 seconds for active, current claims and evidence that he believed that such times were achievable by use of cache. But it does not follow, as the appellant then submitted, that the inference should be drawn that Olsen had the belief that he would so configure the system for RACV as to achieve those retrieval times. Obviously, to believe that retrieval times of 2 to 4 seconds were achievable with the use of cache is one thing: to believe that the system being configured for RACV would deliver such retrieval times in the case of all current, active claims which were to be "on-line" is another matter and one which the evidence referred to did not establish.

Then there was other evidence referred to, by one side or the other, in relation to the argument over reasonable grounds. There was the experience of Telecom which had an InfoImage system installed and which was demonstrated to RACV in October 1993[83]. According to her witness statement, Chesson-Kistas took a note of the visit and in consequence the judge found that "Telecom was happy with the way the project had been implemented and the way the system worked". It was, however, an accounts payable system and so the requirements of Telecom were not necessarily those of RACV. There was a note, too, about slow response times and at trial Unisys had submitted that, after seeing the Telecom demonstration, RACV could not have relied upon the earlier representations concerning retrieval times. The appellant pointed out to us that the judge had dismissed its submission, finding instead that Telecom had demonstrated "a positive experience".[84] None the less the differences between the requirements of RACV and of Telecom remained significant.

There was also "the McFarlane e-mail" dated 27 October 1994, an internal document within Unisys[85]. In it Mr. McFarlane spoke of the increase in the number of claims to be processed by the new system and queried whether the system had the capacity to handle the necessary number, though in the end it was the suggestion of RACV's own expert, Dr. Thorne, that increased cache could have handled any increase in the number of documents being processed and therefore it would not been an enduring problem. And finally in this regard, there was the Esmonde letter of 14 June 1996, "the bleak letter" already mentioned.[86]

This letter, bleak though it was as to the future of the project, was far distant from the making of the representations and so, said the appellant, it was altogether irrelevant to the question of reasonable grounds in 1993, as indeed was also the McFarlane e-mail of October 1994.

Of course, neither the McFarlane e-mail nor the Esmonde letter would have given much comfort to Unisys at trial. Be that as it may, they were of no direct relevance to the existence or not of reasonable grounds, coming as they did well after the event (the making of the representations). The appellant therefore seized upon a sentence in his Honour's reasons for judgment and claimed that it bespoke error. The sentence was the last in a paragraph[87] in which his Honour summed up his conclusions to date, thus:-

"I have found that the representations as to on-line and response times were made and were not withdrawn. I find the plaintiffs relied on them and that they were a material inducement to RACVI entering into the contract and RACVGS expending effort and money. I find that the plaintiffs acted on the basis that Unisys would observe the requirements of the representations when it designed and supplied the system. That was a reasonable expectation and was deliberately induced by Unisys. But Unisys acted in a contrary way, without explanation or approval of RACVI. It designed and supplied a system in which all current claims were not held on-line as defined in the RFP or even just in cache. It provided an archival near line system using unmounted optical disks in a jukebox. The consequence, which was foreseeable by Unisys, was a system that did not conform with the representations and in which slower response times were inherent. (By way of reminder, the retrieval time from an unmounted optical disk could be 20-40 seconds up to minutes.) The representations were broken. I have found that Unisys did not have reasonable grounds for the on-line representation and that conclusion applies equally to the response time representations. On the latter point, there is evidence in the Unisys case which supports the conclusion that Unisys did not have reasonable grounds for the response time representations; there is the experience at Telstra in 1993, McFarlane's query in October 1994 whether there was enough cache, which does not seem to be dealt with, and the admissions in Esmonde's letter dated 14 June 1996 of hardware and software limitations."

The appellant joined issue on appeal with the words "which does not seem to be dealt with", pointing that some steps were taken in that regard, though that is not presently material. The appellant submitted that the judge had fallen into error in having regard to both the McFarlane e-mail and the Esmonde letter when considering reasonable grounds. I would reject that submission. In context his Honour was doing no more than pointing to both the e-mail and the letter as further evidence of a continuing concern over the capability of the system in a relevant respect, a concern which at least sat ill with there being a belief since before the end of 1993 that the system for RACV would deliver according to the representations made before contract. It was no more than that and as such it was not error.

Although the question of reasonable grounds was debated on appeal, in the end, as I followed him, Mr. Bathurst accepted - very properly, if I may say so - that there was no direct evidence that Mr. Olsen, or indeed any officer of Unisys, had the belief that he could configure the system to meet the requirements in the RFP about current, active claims being on-line and retrieval times of 2 to 4 seconds for such claims; and in this regard it is sufficient to refer to the transcript at AB4 1563-9 and 1651. If it was not shown that Unisys had the necessary belief in 1993 on which to base a finding that it had reasonable grounds for making the representations that it did about the outcome of any contract let to it for the delivery of a new computerised system to RACVI, no error is shown in his Honour's concluding that Unisys had failed to show reasonable grounds.

It follows that the cause of action was made out and that Unisys was liable in damages for such loss as was caused by its misleading or deceptive conduct. The question of causation was addressed in the various outlines of submissions with which we were supplied, but it occupied little time in oral argument. On the face of it, RACV demonstrated that it relied upon the representations made by Unisys in 1993 simply by letting the contract to Unisys and awaiting implementation under it and the loss that followed was caused by the sorry saga of non-performance. I have already mentioned in passing[88]

that at trial Unisys made some attempts to portray RACV as the architect of its own disaster, by pointing to its failure to provide a satisfactory Functional Specification, as also to the great number of SCRs that were agreed between the parties during the course of the project[89]. According to his Honour[90], those attempts were abandoned during the trial. But there was another point that was not abandoned: it was another change that was made and it was relied upon by Unisys, both at trial and on appeal, as at least a cause of the difficulties encountered in getting the system up and running. I refer to the so-called change from single image (or document)[91] retrieval to retrieval by folders.

In essence, Unisys' submission was that initially RACV had in contemplation the retrieval of single document images which, if managed by way of an index to electronically stored data, could be achieved a good deal more speedily than if folders were to be retrieved and the contents then scrolled through. Mr. Wall, himself skilled in this area, gave evidence about how he approached the design of the system to provide timely image access, making extensive use of indexes and based predominantly on single image retrieval from the index. But, the judge found (as he was entitled to do), that was an assumption made by him, not an assumption provided by RACV. Wall, it appears, developed some prototypes for RACV to choose between and it opted for folder retrieval. Wall accordingly changed the design to provide for retrieving the entire claim folder. This change was incorporated in the Workflow Narrative and the detailed design document prepared in August 1994.

Importantly, as the judge noted, there was no suggestion that either the exercise just mentioned or the RACV decision to go to folder retrieval was outside, or changed, the contract. No System Change Request was raised and Wall explained the absence of such a request on the basis that the feature was to be built into the initial system. Again as the judge observed, counsel for RACV pointed out that no witness linked the index issue to the failure to meet the requirement for all current claims to be stored on line or to retrieve folders or documents within 2 to 4 seconds. Nor could that link be forged, because slower retrieval times were inherent in a system using unmounted optical disks for current claims. As disclosed in the reasons for judgment, there was ample justification for the judge's rejecting the submission by which Unisys sought to blame the change to folder retrieval for the difficulties, or some of the difficulties, experienced in implementing the system it was providing for RACV.[92] No error was shown in this regard.

Having mentioned the matter of single image retrieval, I should perhaps add that, somewhat surprisingly, in its initial Outline of Submissions on the appeal[93], the appellant repeated its submission that Unisys first proposed to use an index system for the retrieval of single document images and that any discussion of response times was based on that, not on the retrieval of folders with multiple images. The respondents' answer was contained in their Outline of Submissions, dated 15 August 2003, paras. 4.22 to 4.25, which, in the main, followed the judge's reasoning on this aspect. In oral address, Mr Bathurst mentioned what he called "the August departure", but accepted that Mr. Wall made a presentation to the project team in order to show the retrieval of folders and single documents. Expressly, counsel disavowed any submission that the representation earlier made had been withdrawn by reason of the departure from single image retrieval: rather he characterised the change from single page to folder retrieval as scarcely relevant. He referred to it only for the purpose of emphasising that any comparison between what was represented in 1993 and what was delivered thereafter broke down because, bearing in mind the changes that were made to the requirements of RACV, comparison was not really possible. That is a matter to which I return. Importantly, Mr. Bathurst did not embrace the submission found in the appellants' Outline of Submissions in Reply[94], that the representations relied on by RACV "unambiguously concerned only single images or documents". Anyway, I should have thought the July response, tied as it was to the RFP, was too plain to admit of that argument: all current, active claims were to be on line with a retrieval time of 2 to 4 seconds. The question of "single image" retrieval therefore lost significance.

Nothing more need be said now about causation; after all, RACV was seeking only to recover the expenditure that it had wasted on the failed project. But another matter needs mention, affecting damages generally, before I turn to the more particular points raised in argument. That is the question of mitigation. It was Mr. Pearce who undertook on behalf of the appellant the task of persuading us that the judge had erred in making his award of damages by failing to take into account the obligation cast upon the plaintiffs to mitigate their damage. As he began his address, it became apparent that this was not separately addressed in the appellant's Outline of Submissions (the two volumes initially delivered) and so we required a further volume to be produced - and fairly quickly it must be acknowledged - on the question of mitigation. The production of that further volume served to confirm that the propositions in support of the appeal on mitigation were indeed culled from various places in the initial outline, something which, bearing in mind the preparation that must have gone into this appeal, was perhaps unfortunate.

Put shortly, however, the argument on mitigation went like this. As at 24 June 1996 there were only a small number of outstanding functional errors, all of which were capable of resolution. By that date the system was stable, having passed stress and volume tests in February, March, May and June, notwithstanding that the system had been set up to operate inefficiently during those tests in order to subject it to maximum stress. It was conceded by Unisys that the size of cache still needed to be increased, but, it was argued, it was premature to suggest such a solution while the system was being stress-tested [95]. Had the system been put into production, improvements would have been made to response times by increasing cache to hold all images for current, active claims in cache at a cost of about $37,000, and certainly for less than $40,000.

In my opinion, the judge's answer to this argument is already contained in what has been set out above.[96]

It was not the case that the system could be implemented, save only as to the size of cache: by 24 June 1996 there were still problems of functionality which, although it was said that they were capable of resolution, depended upon advice from America and, as the judge noted, whether that could be done and within what time frame were matters of uncertainty. It was Dr. Thorne**,** the plaintiffs' own expert, who, at trial, had suggested that increasing the size of cache could have solved the problems of retrieval times for active, current claims; for adding sufficient cache should have ensured that they were all on-line. It was important that they be in cache because the only speedy alternative, of finding them on an optical disk which was mounted under a reader at the relevant time, was attended by uncertainty.

In all of the circumstances, it cannot be said that the plaintiffs were not entitled to their wasted expenditure because of the possibility - the mere possibility - that within an uncertain time active, current claims could have been put on-line and retrieval times brought within the requirements found, so long previously, in the RFP. In particular, Mr. McArthur, who argued mitigation on behalf of the respondents, pointed to the lack of evidence below about the effect of increasing cache. He said, too, that there were "software constraints" still existing and he pointed to the evidence in support. In reply, Mr. Pearce accepted that if there were "underlying software problems" the addition of cache would not be an answer in itself. As he put it, the appellant met the point by saying that there was no finding of such a defect and no evidence, or no sufficient evidence, to support such a finding. Indeed he submitted that the findings of the judge were inconsistent with any underlying software defect precluding an increase in the response times by the addition of cache. He pointed, in this regard, to paragraphs [410] and [412] of the reasons for judgment.

I found the argument unpersuasive. In paragraph [410] of the reasons for judgment his Honour did discuss "the option of increasing cache to convert the system to one in which all current claim images were held on-line in magnetic cache", although, according to the findings, Unisys never gave consideration to this option at the time.[97]

"Hence the solution, accepting it to be such, was not tried and we have this litigation. If it had been tried, the answer as to its effectiveness, and the cost and the effort involved in implementation, would be known. But it was not and it is difficult, having regard to all that happened in attempted implementation of the Unisys system, to be as confident of success as counsel for Unisys contend would have been the case. More particularly, in all the circumstances I am not persuaded on the balance of probabilities as to any actual or approximate period of time, or the effort, that would have been involved in adding sufficient cache to enable all current claim images to be held in magnetic cache. At one stage it was submitted I could find that the necessary work would have taken no longer than 60 days to perform. I am left with little confidence in Unisys' capacity to so act. In my view, on the evidence, it would be speculation to make that finding. I reject the submission."

With respect, none of this bespeaks error. The solution of adding cache was certainly untried and the cost and effort involved in implementation were not known.

I have set out the foregoing only because it was in that context that the judge proceeded to say, in paragraph [412]:-

"I am prepared to accept that if Unisys and RACVI, with inexhaustible patience and a continuing application of resources, had kept working away at the project at some time Unisys could have delivered a system which met the requirements of RACVI. I reject the case that this would have occurred within a matter of weeks or even within 60 days. I find that it would have taken longer on the balance of probabilities."

This was the passage relied upon by Mr. Pearce to support his contention that the trial judge's findings were inconsistent with the existence of any underlying software defect that precluded the extension of cache. But that is not how I read the passage just quoted. The judge was "prepared to accept" that the system might have been made to meet the requirements of RACVI, but that was not a finding to that effect. The judge was "prepared to accept" it for the sake of argument: for he was dealing, at that point, with whether the representations had been met or not. His Honour was clearly of the view that they had not been met and the mere possibility that within perhaps a year the system might have been made to work to the requirements of RACV was rejected as irrelevant. As I say, that the judge was "prepared to accept" the possibility was not a finding that it could have been done. I would accept Mr. McArthur's submission and reject Mr. Pearce's.

As already indicated, in the main the judge awarded the damages claimed by RACV, which consisted of wasted expenditure over the not inconsiderable period taken by Unisys in attempting to deliver according to its obligations. (Like the trial judge I am not speaking of the contract position, for there is no need to.) But one particular issue affecting quantum was raised by the appellant: namely, the effect in the computation of damages, not simply of the receipt by RACV of the settlement sum from Deloittes, but of the allowance, if any, to be made for the costs incurred by RACV in proceeding against Deloittes by action.

Below, both sides made submissions about the settlement sum. It will be recalled that Deloittes (in two guises) were initially named as defendants to the proceeding but the claims against them were settled before the trial began. The amount of the settlement sum was not made public (or so we were told): nor was it made public whether the settlement sum was paid for claim and costs or claim only or costs only. On behalf of Unisys it was contended that any sum awarded by the trial judge for damages should be reduced by the amount of the settlement sum, it being a sum which had been obtained from co-defendants in respect of claims that were concurrent with those made against Unisys. For RACV it was accepted that due allowance should be made for the settlement sum, but only after reduction, it was claimed, for the costs incurred by RACV in pursuing Deloittes. No doubt because of the settlement no order for the payment of costs by Deloittes was sought or obtained and no attempt had been made by RACV before trial to quantify in any way the costs which it had incurred in respect of the action against Deloittes.

At trial, Unisys submitted that because the costs were not quantified the full amount of the settlement sum (whatever it was) had to be allowed by way of reduction in any damages awarded. For RACV it was submitted that the judge should make an estimate of the costs incurred in pursuing Deloittes[98] and reduce the settlement sum accordingly before, in effect, setting it off against any damages awarded. In his reasons for judgment, the judge rejected both these approaches, choosing instead to refer the plaintiffs' costs to the Taxing Master for assessment and ordering, in effect, that once ascertained the amount of those costs should be deducted from the settlement sum before it went in reduction of the damages. This was the subject of the complaint in the notice of appeal, it being contended that the judge ought to have allowed the full amount of the settlement sum by way of reduction of any damages otherwise ordered, ignoring any amount for costs simply because those costs had not been quantified by the time of the trial. It was submitted to us that, in ordering as he did, the trial judge in effect allowed RACV to re-open its case to call further evidence.

Before proceeding any further, I say something more about the orders actually made below. So far as presently relevant, they were these:

"1. The Firstnamed Defendant [that is, Unisys] pay the Plaintiffs the sum of $4,032,232.00 including $1,488,076.00 by way of interest. 2. The Plaintiffs' costs of the proceeding against the Second and Thirdnamed Defendants [Deloittes] be taxed as between party and party. 3. The parties take such steps as may reasonably be necessary to have such taxation concluded as soon as practicable. 4. When so taxed (or agreed) the Plaintiffs pay the Firstnamed Defendant the difference between the amount of such costs of the proceeding against the second and Thirdnamed Defendants and [a sum specified in the order of some hundreds of thousands of dollars] with interest from 29 August 2000."

Thus, the parties were not required to return to his Honour. Assuming that the sum specified in the orders was indeed the settlement sum received form Deloittes, it would appear that the judge made no reduction in his award of damages for the settlement sum, resting content instead to order that that sum be repaid by RACV to Unisys after itself being reduced by the amount assessed by the Taxing Master for the plaintiffs' costs of the proceeding against Deloittes.

Of course, what was being ascertained by these orders was the costs, not of Deloittes, but of the plaintiffs in pursuing Deloittes and it was those costs that were ordered to be assessed on a party/party basis. We were told on inquiry that the order for costs was in the form sought by RACV's counsel at trial and so it is unnecessary for us to consider whether it might have been open to order the assessment of the plaintiffs' costs on the basis of solicitor and own client, the better to reflect the costs incurred by RACV in pursuing Deloittes, as distinct from the costs that RACV might have recovered against Deloittes had the matter proceeded to trial and damages been awarded. That is the first matter I mention at the outset.

The second is that the order made by the judge for assessment of the plaintiff's costs by the Taxing Master was not an order sought below by either the appellant or the respondents. Unisys had contended that the full settlement sum should be allowed; RACV, that the judge should make his own estimate, as best he could, of the costs involved. The latter had been the course followed in Boncristiano v. Lohmann[99], a like situation in which the costs had not yet been ascertained by the time of the trial. On the appeal, RACV contended vigorously that the course taken by the judge had been open to him and that his order was perfectly proper. In the course of his argument, the Court inquired of Mr. McArthur whether Unisys had in fact taken objection to the order made below, that having some bearing, perhaps, on whether it should now be entitled to be heard on appeal against the order that was made. The Court was assured by counsel for the respondents that, so far as recollections went, no objection had been taken below to the making of the order in a form which, it was accepted, had been sought by none.

Somewhat surprisingly, therefore, on the following morning and immediately after senior counsel for Unisys had addressed in reply, junior counsel rose to follow his leader on the question of damages and promptly produced a copy of the transcript below, which he acknowledged had been available to him the previous afternoon. It showed that objection had been taken by Unisys and the reasons given by the judge for overruling the objection and making the order that followed. Despite the terms of the notice of appeal and despite more than one directions hearing, the transcript had not been included in the voluminous material placed before the Court, notwithstanding its obvious relevance. RACV's counsel generously took no objection to its being placed before us informally and so Unisys was able to demonstrate that objection had indeed been taken.

As to the point taken on appeal, I think that it has no merit whatever. After some discussion, Mr. Pearce conceded that, by one means or another, RACV and its solicitors could have had their costs taxed, and taxed on a party/party basis if so minded. He conceded, too, that there was no question about the judge's power to make the order which he did. The submission was a limited one: that, in the proper exercise of discretion, the judge ought not to have made the order he did - in part because, as the appellant would have it, the judge was thereby allowing the plaintiffs to re-open their case and call further evidence. In my opinion, the submission should be rejected. As Mr. McArthur put it in answer, the judge was doing little more than referring the matter out for inquiry and report. The judge was plainly satisfied that significant costs had been incurred by RACV in pursuing Deloittes and thus in recovering the settlement sum. It was only fair, if the settlement sum was to go in reduction of the damages awarded, that the settlement sum itself should first be reduced by the costs incurred (and whether on a party/party basis or on the basis of solicitor and own client does not now matter). The reduction had to be assessed somehow and the most obvious way was to allow the Taxing Master to do the assessment. To that end, the judge could have adjourned the hearing to allow for assessment before pronouncing final orders: instead, he chose to make the orders without reduction for the settlement sum and then to order repayment to the extent of the settlement sum less the amount for costs allowed on taxation. The order simply anticipated the Taxing Master's assessment. That was the exercise by the judge of his discretion in a matter of practice and procedure and nothing has been shown as to why his Honour ought not to have made the orders which he did.

At one stage Mr. Pearce suggested that a principle was involved, but I do not see it. It being common ground that the settlement sum (or at least some part of it) should go in reduction of the damages, and it being his Honours' view (which was open to him) that the settlement sum was properly to be reduced by the costs incurred by RACV in obtaining the settlement sum, it was perfectly appropriate for the judge to put in train the necessary steps to bring about that result. That the orders were fashioned to avoid the need for the parties to return to the judge after the Taxing Master had performed his task should be matter for satisfaction, not complaint. Certainly, I reject any submission that the judge was required to refuse any allowance for the costs incurred by RACV in pursuing Deloittes simply because the step of quantification remained to be taken. For a time it seemed to me arguable that, as the appellant was itself the one seeking a reduction in the damages by reference to the settlement sum, the onus was on Unisys to satisfy the Court what part of the settlement sum should properly go in reduction - an onus which Unisys had not discharged. A like argument may have been rejected, however, in Boncristiano[100], although in that case, the trial judge had been prepared to make his own estimate of the costs involved. It is unnecessary to pursue the possibility here because, as I say, the course taken by the judge was open to him and the appellant has not established any ground for appellate intervention.

Finally, I simply record that although argument commenced in relation to ground 11 in the notice of appeal, Mr. Pearce did not pursue it when we refused him leave to amend.

We were told that, if the appeal by Unisys against the judge's acceptance of RACV's claim fails, so too does its appeal against the judge's dismissal of its counterclaim. Thus, the foregoing leaves for consideration only one other matter and it, too, affects damages. It is that raised by RACV's cross-appeal. It will be readily understood that, given the time taken by the failed project, RACV spent a good deal on staff labour which, in the end, was altogether wasted. What it did, because of the complexity of the system, was to direct some of its more experienced staff to spend time on the project and, in order to cover their absence from their normal duties, engage additional staff. At trial, RACV sought the costs of employing those additional staff and RACV sought also the costs spent on those members of its permanent staff who worked on the failed project. Of course it could not recover twice for the same loss, and so much was accepted. But the claim was put on both bases and evidence was led as to both.

The judge rejected the claim for the cost of additional staff because he expressed himself as far from satisfied about the costs involved. That is not now in issue. What is in issue is the next step taken by the judge: his refusal to allow the claim for the money spent on permanent staff who were engaged, for such time, on the project that failed. His Honour accepted, as did the parties, that the proper measure of damages required that the plaintiffs be put in the position that they would have been in, had there been no misconduct (that is, no misleading or deceptive conduct by way of a misrepresentation which led to the contract in the first place). That, said the judge, was the cost of the additional staff, and not the cost of the permanent staff engaged on the project; for the permanent staff would have been paid in any event. If they had not worked on the failed project, they would have been working on their regular duties. It is that rejection of RACV's claim which is now the subject of the cross-appeal.

In my opinion, the cross-appeal should be allowed. This is not, as counsel for the appellant sought to say, a claim for loss of profits or savings foregone. It is a claim for the cost incurred in taking steps which were required because of the misconduct basing the cause of action. Of course that cost might be measured by looking at the additional staff employed to cover the absence from their regular duties of the permanent staff; but I see no reason why, on an alternative basis, that loss could not be measured as a cost of employing the permanent staff so long as they were working on the failed project. What one is seeking to do is to quantify the loss suffered by reason of the misconduct; RACV suffered it in the labour costs incurred and it seems to me not to matter that there were two ways, not one, in which those labour costs might be measured. Just as RACV lost, perhaps, the cost of employing additional staff to cover the absence from their regular duties of the permanent staff (because had there been no misconduct those additional staff would not have been needed), RACV also lost the cost of employing the regular staff while they were employed on the failed project, for they certainly gained no benefit from those labours. As I say, it was always accepted that RACV could not recover twice over for the same loss: but that does not mean that the loss could not be measured in more than one way.

In support of its claim RACV referred the court to the discussion at first instance of the like claim in NRMA Ltd. v. Morgan[101], contending however that the decision in that case should be distinguished[102], a point disputed by Unisys[103]. In NRMA Giles, J. held[104] that staff costs were not recoverable because the plaintiff had not proved payment of additional or increased salaries and wages which would not have been paid in any event. His Honour said[105] that so far as the evidence went NRMA would have paid by way of salaries and wages what it in fact it paid over the relevant years, whether the employees were working on the proposal or not. It is very different here, where RACV demonstrated that it did hire additional staff to take the place of regular staff working on the project. The appellant then sought to call in aid Westsub Discount Pty. Ltd. v. Idaps Australia Ltd.[106] but that too is distinguishable, for there "none of the work wasted by staff [was] sufficiently identifiable"[107]. Here the trial judge quantified that which was identifiable. This case is closer to Durham v. Aduke Pty. Ltd.[108], which appellant's counsel, conceding that on its face it supported the claim made by RACV here[109], submitted should be distinguished or not followed, but I disagree.

Accordingly, for the reasons I have given I would dismiss the appeal by Unisys, but allow the cross-appeal by RACV, ordering that the latter's claim for wasted labour costs be allowed and damages awarded accordingly. We were told that the parties were agreed that, if that claim succeeded, the further sum of $350,000 should be allowed plus interest from the filing of the writ. Because our order takes effect, albeit by way of substitution, on the date on which it is made, interest should be allowed from the filing of the writ until the date of our order: Nicol v. Allyacht Spars Pty. Ltd.[110].

In concluding that the appeal should be dismissed I have not dealt with all that the trial judge discussed in his long and detailed reasons for judgment, nor of course have I canvassed all of the evidence and documents that were before him: that would simply not be possible on an appeal. I have confined attention to those matters raised by the parties in their lengthy but helpful written submissions, supplemented as these were by oral argument. I am confident that both sides said everything that they wished to say, and if I have not dealt with every point taken I have dealt with those that appeared to me the most significant. Indeed it seems to me that this case, which has occupied so much time at trial and on appeal, turned upon a set of facts which was largely not in dispute, but that the parties, or perhaps more accurately one of the parties, has been at pains to drag from those facts every conceivable legal point that might be made, no doubt in the hope that somewhere a favourable case would emerge. That is not what I understand the law to be about.

Let me say immediately, lest I be misunderstood, that I am not critical of counsel on appeal; in particular Mr. Bathurst, senior counsel for the losing side, was both able in argument and fair in presentation. I would be concerned, however, if it were to transpire that, after a long, drawn-out commercial undertaking by Unisys in a field in which it claimed expertise, but an undertaking which caused its customer nothing but problems and enormous expense without useful result, the solicitors for Unisys had allowed their client too much latitude in dictating the course of this extended litigation, even to the point of the client's requiring, after a comprehensive loss at trial, an appeal which seems to me to have had little to commend it. Certainly the denials made by Unisys and the stand taken by it on some points at trial did little to confine the argument sensibly and much to extend it without, I think, sufficient warrant. Perhaps this is to be wise after the event, but I am concerned, as I said

during argument, about the amount of court time occupied by this litigation and the vast expense so obviously incurred. For my part, I would be prepared to entertain a submission that, given all the circumstances including the voluminous but in some respects disorganised state of the material on appeal[111], Unisys should suffer an order for costs on a basis over and above the usual, that of party and party. I would anyway hear counsel on the precise form of the orders to be made.

I agree with Phillips, J.A., whose reasons for judgment I have had the considerable benefit of reading.

I certify that the preceding -1 pages are a true copy of the reasons for judgment of Ormiston, Phillips and Batt, JJ.A. respectively of the Court of Appeal of the Supreme Court of Victoria delivered on 14 May 2004.

[1] AB1 A235. In these footnotes, "AB" refers to the 45 volume Appeal Book and "CD" to the three volume set of "critical documents", prepared by the parties, at the direction of the Court, for use on this appeal.

[2] Judgment [189], [193], [194], [200]. See Hurford's witness statement paras.13-16 (AB10 D598-9 and AB32 E5545, E5554 and E5595). See, too, the letter from Walsh to Hurford dated 25 March 1996 (AB32 E5629). The chronology is set out in the Appellant's initial Outline of Submissions Vol.1 section 2, pp.68-71. Reference may also be made to the letter of Walsh to Hurford dated 21 June 1996 in which the writer addresses RACV's "expectations for system acceptance by the 24 June deadline" and expresses the opinion that "we [Unisys] can meet the commitments made in this letter and our obligations under our Agreement by 24 June 1996" (AB35 E6363, E6366).

[3] As is quoted in the judgment at [80], the July response of Unisys described "cache" thus, on p.2-10: "As a rule magnetic storage - often called `cache' - is used to store information that is needed on a more immediate or frequent basis than some other information".

[5] There is a useful set of abbreviations and list of witnesses in the trial judge's reasons for judgment at [9], [10] and [11].

[7] Appellant's Outline Vol 1 pp.11 ff., para 2.1.1; judgment [56]-[60].

[12] Judgment [76]-[80]. CD1 206ff.

[33] Appellant's Outline Vol 1 page 14 para 13. See also judgment [113]-[123].

[36] Board minutes AB 22 E2229, Appellant's Outline vol.1 p.15, para.17.

[37] Judgment [131]. Mitchelson was general manager of RACVI: judgment [10].

[42] Appellant's Outline Vol.1 p.16 para.21. Compare the PMP p.6 (AB22 E2300).

[51] Appellant's Outline vol. 2 page 222 para. 124. The appellant submitted that Westsub Discounts Pty Ltd v. Idaps Australia Ltd (1990) 17 IPR 185 at 224, 228, 231 was just such a case, but of course the facts were different and counsel had to concede that s.51A had not been in force at the relevant time.

At a glance

Court

Decision date

Before

Source

Judgment (230 paragraphs)

Parties

Cases Cited (2)