[SystemSafety] Fwd: Measurement + Control

[Peter Bernard Ladkin]

On 12/16/13 10:09 AM, RICQUE Bertrand (SAGEM DEFENSE SECURITE) wrote:
> Beware that this depends highly on the different industrial sectors and their own culture.

Yes.

Let me bring in some other considerations to the business of "outsourcing".

We were originally concerned with outsourcing in the case in which some other organisation than the
operator of (say) a plant undertakes critical operations and maintenance activity. And we were
concerned with how effective or ineffective this might be in ensuring appropriately safe operations
and activity.

Ronald Coase's transaction-cost theory of the firm (1937) says that people organise themselves into
something called a firm when internal transaction costs are lower than external transaction costs
for key transactions. The existence of such an organisational form as the firm is of course due
ultimately to the polity - the theory only suggests when people begin to make use of that form. BTW,
Coase died in September this year at the age of 102.

Outsourcing is effective, on this model, when certain predicted external transactional costs are lower.

It is not as simple as this, of course, even were there to be some agreed way to reduce standards of
safety to a transactional cost.

But attempting to do so seems to be broadly helpful. I illustrate with an example, contentious
though it may be. There used to be one National Health Service in Britain, which maintained all
health-related actions in-house. When time came to consider out-sourcing, the first thing that
needed to happen was to develop measures of performance, so that the performance of alternative
service providers could be contractually specified. That was hugely hard, for reasons we do not need
to go into but which are well-known to anyone who has thought about how to measure how good health
care is.

But when you have a set of measures, of course you can also assess in-house institutions, that is,
internal NHS providers, according to those measures also. So, for the first time, the performance of
various NHS entities was evaluated. And many of those entities were found to be lacking. Reviewers
were called in; investigations were undertaken; countermeasures were proposed (often involving
shutting the unit). Serious discussion amongst dedicated, involved people ensued. Whatever the
outcome of individual cases, service-provision problems were identified and resolved through
deliberation across many organisational levels and I consider this a Good Thing. Various regional
hospitals have been closed (Stafford and Cumbria come recently to mind). Units have been argued over
(Leeds paediatrics, Bristol hearts). And some alternative providers have lost their contracts
(Serco, general-practice provision in Cornwall). This has all happened since 1990, starting with the
"internal market", which was basically about management but as I understand it without
externally-derived performance measures. Then in 1997 this was renounced but then reintroduced about
a decade ago once the advantages of some performance measures were acknowledged. Whatever one may
think about its mythical failings, the British NHS provides some of the best, some would argue the
best, value for money in health care in the world.

Now, performance measures can be played - consider the UK Research Assessment Exercise for
universities. But having *some* measures seems to be mostly advantageous.

The same surely holds for activities such as outsourcing operations and maintenance of
safety-critical kit. To outsource, there has to be some kind of contract, and that contract must
specify performance. Both sides have to agree that the specified performance levels are appropriate,
and both of them must at least think that they are indemnified in case of accident. You can't any
more have a manager throw a rule book at a junior employee and say "follow the rules" and then blame
himher when an accident occurs. There is now an interface saying just what that service entity
replacing the junior employee has to do and what constitutes success. And the manager cannot ignore
hisher responsibility for specifying it and adhering to it. There is a contract specifying the
interface.

Problems arise here I when much of the activity needed to ensure appropriate safety is implicit, and
I think this is what Colin Howard and Ingo Rolle were emphasising. Much of the knowledge required to
ensure safety is not explicit but implicit in many such operations. So outsourcing is going to be
risky until that implicit knowledge is made explicit and put into contracts. How to do that and how
long it might take is a question concerning which I have little experience. But it is surely one
major topic for IEC 62879 HF&FS to consider.

PBL

Prof. Peter Bernard Ladkin, Faculty of Technology, University of Bielefeld, 33594 Bielefeld, Germany
Tel+msg +49 (0)521 880 7319  www.rvs.uni-bielefeld.de