Examination of Witness (Questions 660-679)
MR MARK
ADDISON
28 JUNE 2006
Q660 Sir Peter Soulsby: Could we
infer from that that what really went wrong was the failure, at
an appropriate stage, to test the system, to make sure it really
would deliver and really would work? Is that fair?
Mr Addison: As I said earlier,
the amount of testing that was done, the amount of whole system
testing that was done and the amount of whole system knowledge
there was, was, as it turned out, not great enough. One of the
reasons for that was the overall timetable. A lot of this can
be traced back to my earlier two points about the number of things
being done and the timetable.
Q661 Sir Peter Soulsby: You have
explained all of the reasons behind this, all of the factors that
were part of it and you have explained the decisive points, which
were things that you were aware of well in advance. It does strike
me that the actual failure was to test the system and make sure
in a timely manner that it was actually going to deliver. It surely
was the responsibility of the Rural Payments Agency and its chief
executive to test and the responsibility of the Department to
make sure that they had tested it.
Mr Addison: Testing can mean a
very wide range of things and you have had evidence from others
who have explained that a good deal of testing was done. The individual
system releases were extensively tested. If you look at the OGC
programme of reviews, all of those reviews looked at the bits
of the RITA system that were being designed and one of the questions
they asked was whether this had been successfully tested and so
on. So there was a great deal of testing. As it turned out, where
the testing did not happen, in the way that it would have been
useful if it had, was the testing of the whole system which would
have been a very major undertaking, would have taken a good deal
of time and would probably have compromised the original target
in any event. There was also lacking, as I have said before, a
depth of understanding of the way in which the whole system would
fit together and work and those are the issues. It is quite interesting;
it is not quite as clear-cut, if I may say so, as you are suggesting.
Take the partial payment system which was designed and used in
early May; the actual partial payment system which was used to
get the £730 million out. We had a choice around the end
of April about how much testing we should do on that programme.
Q662 Chairman: Are you talking about
April 2005?
Mr Addison: No, April 2006. I
am talking about the partial payment system which enabled the
money to be released. There was a question about how much testing
we should do. As you would expect, having come in afresh, having
a naturally sceptical attitude to claims that X or Y would work
unless I had actually seen it working in black and white, my inclination
was to say that we must test as much as we possibly could and
we did extensive testing on that new partial payment system. However,
there was one thing we did not test. We did not test it end to
end. We decided that if we were to let, say, 100 cheques go through
the whole system and be paid to individual farmers, it would delay
the whole process by several weeks. I cannot remember the exact
timetable. So what did we do? We looked at alternatives. We said
"What is our confidence that the payment system works? What
is our confidence that the partial payment system will work? How
confident can we be that, put together, these two things will
work?". The conclusion that we reached was that we had a
good degree of confidence in the linkage between those two, the
actual calculation mechanism and the payment mechanism, on the
basis of what we had seen already. We therefore decided to advise
ministers that the whole arrangement was ready for deployment
and that, although we had not had a chance to test the whole thing
end to end, we were pretty clear it would work and it worked like
a dream. That was the right call; history convinces me it was
the right call. So it is not always obvious that a test is absolutely
required, if you are up against a very tight timetable and you
have a high degree of confidence that the different bits of the
system will work together. Of course we now know, with the benefit
of hindsight, that we should have done more whole-system testing.
We know that; I agree with you. Was it obvious at the time, given
that the timetable had to be met? I am not convinced.
Q663 Chairman: You took on board
Accenture as your IT partner from the start of the change process
for the RPA, the big change of the whole system. They are there
as professionals and they told us that basically the system they
designed would do what it said on the side of the box. In their
remarks to us they were confident that they had delivered what
they were asked to do. I have just been reminding myself about
how much time remained once the policy was agreed and there was
actually quite a lot of time, at least to tackle the problems
at the front end of the system without having to test the whole
thing. Did you get advice from Accenture about the risks that
this approach was putting to you? Accenture were very careful
in their remarks not to make any comment about any other part
of the RPA for which they were not responsible and, to an extent,
one can understand that. But in terms of the mapping system, was
that ever tested with what I might call some dummy information?
When, for example, was it decided to employ the digitised approach,
a different approach to the IACS mapping system which relied on
ordnance survey maps, to employ that technology? Why seemingly
was it not tested with some real live farmer data to flush out
some of the idiosyncrasies to which we have just been exposed
by a presentation from the valuers? They made very clear to us,
from their personal experience and indeed the Committee have had
many representations, the type of instability on the mapping system,
because that is the most fundamental before you even get to the
point of application validation. Why was that bit not tested more
thoroughly?
Mr Addison: I am afraid you are
taking me into territory I cannot help you with because I cannot
recall whether I myself, when I was on the Defra side, was part
of those discussions or not. You are absolutely right that the
mapping and the land issues are obviously at the very heart of
this and if one looks at most of the tasks which have to be undertaken
in level two validation, many of them are to do with land and
maps. Accenture are making a valid distinction between testing
the system to make sure it works at a technical level and the
question which I think you are asking which is surely that if
you had piloted the RLR and this new approach, it would have become
clear that there were going to be more changes to land and to
parcel size and applications coming in from farmers than the RPA
expected. That is not a technical point; that is about the way
in which the whole system works. I think that Accenture are making
a valid distinction. I am afraid I cannot help you on whether
that was piloted in some way or not.
Q664 Chairman: Let us come to a fundamental
question which I should have thought the board of Defra and indeed
the Ownership Board of the RPA would have asked, a simple question:
does it work? Let us separate out the question of volume and change
from the stability of the system once in operation. The example
that we had shown to us this afternoon was of a farmer who ultimately
had the issue of her maps resolved after 19 separate attempts.
The first part of the process began in May 2004 when this particular
applicant started the process and it was not actually resolved,
finally, until 16 May 2006. Given that you knew you were going
to have to operate a new digitised mapping system, I really do
not understand why somebody did not do some "dummy runs"
on this. I am sure the board must have asked whether this bit
actually worked, because that was the bit that was ahead of the
bit you have described, the level one and level two parts of it,
which had their intrinsic problem. Did somebody on the Defra board,
when you were getting reports back from the Ownership Board not
say "Have we actually made certain this thing is stable and
it can process a number of these"? I am sorry to go on, but
just to give you an illustration of what we heard, when the valuer
in question got along to say "Right, let me try to help you,
oh client, to get this thing right" at one point quite early
on in the process she told us that the system had just about got
it spot on. So an agreement was reached that she would print-out
a new set of maps for final confirmation. What then happened was
that a set of maps arrived which had a completely different order
of land on it, fields were not there, in other words it was not
the same as that which had been agreed. So you ask why the system
had substituted a series of incorrect maps for ones which had
been deemed correct. It is that kind of very practical example
which does not seem to have been bottomed before you went live
and compounded the problem by a whole series of instabilities
in the system. That is why I come back to the very simple fact,
and Peter would agree with me, that we do not understand why that
bit was not tested, as we as lay-people understand it, which means
putting 200 or 300 real live mapping exercises through it and
seeing whether it actually works. Was that never done?
Mr Addison: I do not know the
answer to that because I simply do not know how the testing was
done.
Q665 Chairman: Who would know the
answer to that?
Mr Addison: The RPA would certainly
know the answer to that question.
Q666 Chairman: So we need to put
that on the yet-to-be-asked list.
Mr Addison: Yes. It is a very
specific question and I am sure you could get a very specific
answer to it.
Chairman: Coming back to the questions
which ministers were asking, ministers would have been briefed
on what it was they were buying. They would have had a description.
All those submissions to ministers would have said that ministers
will be aware that we are introducing a new digitised mapping
system and this is how it will work. Did they not have a demonstration
as to what was going to be coming along? Were they exposed to
it? Lord Bach told us he popped in and out of the RPA with monotonous
regularity and Lord Whitty no doubt did the same.
Q667 Lynne Jones: Or even members
of the Ownership Board like you.
Mr Addison: I cannot remember
the exact date, but what I do recall is that as members of the
Ownership Board we did see the rural land register at a point
at which it was struggling, not so much to sort the problems that
you have identified, but actually simply to work at a technical
level. The volumes of activity that the system was being required
to cope with were overwhelming it, so it was suffering from a
lot of problems. We were in Exeter, we saw it.
Q668 Chairman: That is reflected
in the minutes. What I am concerned about is what happened upstream
of that. Coming back to Accenture, they were remarkably correct
in saying that the system they provided was going to deliver.
I presume Accenture were responsible for the mapping system, were
they?
Mr Addison: For the design of
the RLR.
Q669 Chairman: Was that one of their
boxes?
Mr Addison: I would want to be
absolutely sure.
Q670 Chairman: If you are not sure,
I shall not press you because it would be unfair.
Mr Addison: We can find out. I
believe they were, but we can check. Subsequently of course it
was then moved off at the point where the volume of activity on
digitisation exceeded the capacity of the organisation to deal
with it.
Q671 Chairman: They told us that
each component of the system had been, by their use of the word,
tested and that they were clear that it worked. If the IT supplier
says he is producing you a system which in 2005 they had tested
and was stable—that is what they told us—if it was
stable and it worked, why did it not deliver? That is the layman's
question.
Mr Addison: I agree and it is
a good question but I would come back to try to make this distinction.
I am sorry I cannot be more helpful about exactly what happened
and when. There is a distinction between a system which is technically
stable and a whole work process which is not performing satisfactorily.
Just by way of an example, take the authorisation checks. The
authorisation checks were not technical difficulties. The RITA
system was working, the individual operator would identify a problem,
the claim would be identified and the batch would be stopped.
The issue was the way in which the whole work process functioned
and in particular, in the case of land, the way the relationship
between the customer and the RPA worked or did not work in relation
to the flow of information between them. That is not necessarily
the same, in fact it is not the same, as saying the system was
technically unstable. The system may have been technically quite
stable, but the whole work process was not functioning properly,
which clearly it was not.
Q672 David Taylor: What, in one line,
would be the responsibility of the Ownership Board? Just in one
sentence. How do you recall the brief of that board? Just one
sentence.
Mr Addison: The Ownership Board's
responsibility was to do two things and this is general in Government:
one was to agree the plans and targets and resources for the agency
and advise ministers accordingly; linked to that, to take responsibility
for the Agency's performance and advising ministers again on how
it had performed against the targets which had been agreed. That
was one set of performance measures. The other is to be the representative
body in the Department for that agency or organisation.
Q673 David Taylor: If I were Johnston
McNeill on gardening leave, sitting in my potting shed this sunny
afternoon with the radio on listening to the evidence that has
been given, I should be thinking that Mark Addison was being given
credit for sorting out problems for which he, as part of the Ownership
Board, was substantially culpable. The Ownership Board collectively
signed off the specification, did you not? You must have done;
you must have signed off a transaction-based system, an antiquated,
historic approach to major system design equivalent to a single-task
production line in an engineering factory decades ago when people
had moved on to team approaches where sub-component parts of the
car or whatever it might be were assembled at that point. The
Ownership Board were responsible for that, were they not?
Mr Addison: Maybe I can deal with
two of the points you make. On the Ownership Board responsibilities,
I have already said as clearly as I can that the Ownership Board
back in 2003, on the basis of advice from the RPA, endorsed the
view, as Sir Brian Bender said, that the RPA change programme
could be run in parallel with the CAP reform initiative. History
tells us that was not the right decision. Everybody around that
table, including me, bears some responsibility for that decision.
Q674 David Taylor: I am not holding
you individually responsible; I am saying you are corporately
culpable.
Mr Addison: That is absolutely
clear and I have tried to be as upfront about that as I can. As
to your second point about claiming, me personally and my successors—
Q675 David Taylor: It was your admirers
who claimed it.
Mr Addison: No, but the point
you made was about successes at the RPA. I am also very keen to
be completely clear with you that the progress that was made at
the RPA during my tenure was about the successful delivery of
substantial sums of money—
Q676 David Taylor: It was the little
boy saying the King has no clothes, was it not? It was stating
the obvious, was it not?
Mr Addison: If I may finish the
sentence. It was about getting money out, it was not about solving
the fundamental problems which will take some time to resolve
and the RPA will be doing well to improve in 2006 on performance
with the 2005 payments and we all hope that it can create some
more significant improvements in 2007 and maybe 2008 is the first
major opportunity.
Q677 David Taylor: Moving on, you
expressed the opinion that it would have been a good idea to have
had more of an independent perspective, people who knew something
about ICT rather than mandarins and outsourced staff who did not
have an invested interest in it. Do you regret now being substantially
responsible for making redundant 1,600 to 3,500 people who might
have given some input to improve the way in which this system
was delivered? Almost a half of the 3,500 staff were made redundant,
that is right, is it not? You were part of that.
Mr Addison: Certainly that was
the plan. Again, with the benefit of hindsight—
Q678 David Taylor: It is not with
the benefit of hindsight. With great respect, you, Defra senior
management outsourced a great deal of your ICT expertise. Those
people with the knowledge of the area concerned, who had the experience
and could give advice, could have pointed to the significant and
serious flaws that were always there in that original system design,
when the people charged with signing it off, that is the board
of which you were part, clearly were either hoodwinked or negligent.
That is a fair assessment is it not?
Mr Addison: Absolutely not. Maybe
I could just explain why. First of all, by "the decision
to outsource" I take it you mean the Accenture involvement.
Q679 David Taylor: It is a general
Civil Service trend to outsource as much as possible of that which
is crucial to the future of the organisation and retain that which
is not. Yes, I suppose I do, in relation to the Accenture contract.
Mr Addison: That is not right.
The planned reduction in staff numbers, which of course has not
happened in total, but the planned reduction in staff numbers
of the 1,500 that you mentioned was not around the outsourcing
of IT, it was to flow from redesigned business processes which
Accenture were there to help with.
| 
