One benefit of the elective component of the management training schemes is the chance to see how different parts of the service interact during the same time period.

The thoughts here came together as the result of one of us spending a placement in the Department of Health, and the other in a large metropolitan health authority, both working on winter planning.

The efficacy of information-gathering processes and their influence on rational policy-making were identified as issues. We tracked the stages of the information 'feedback loop', which moves out from the centre and then back.

First, ministers decide to focus on a particular problem area. At this stage, the information required can be summed up with the request, 'I want to know what is going on.' Then comes a process of interpreting this demand at senior levels within the civil service. The question becomes more specific: what type of information will satisfy the ministerial request?

There follows a series of information demands sent to the service. This typically passes through regional offices, then to HAs, and finally to trusts, which attempt to translate their 'real worlds' into some required data set.

Trust data is then aggregated at HA level, aggregated again at regional offices and again at the central point before being 'digested' and used to provide an 'answer' to the original ministerial request, in the form of a briefing paper. The chief concern with this process has to be the potential for 'error amplification' through the information loop.

A 'Chinese whispers' problem can occur if the initial interpretation of a minister's request results in a method which will answer some questions - but perhaps not the question the minister had in mind. Subsequent manifestations of this problem can occur as the request is made to regional offices, and then to HAs and trusts. Even if the criteria are passed on accurately, there are likely to be wide variations in how they are applied.

Another problem arises at the service end during data collection. Either because the data required is novel, or not routinely collected in that form, or because of time pressures, reports back may be based on anecdotal rather than statistical evidence.

These problems produce doubt about the consistency of data as it is aggregated up through the system. It is possible, in the light of these formal problems, that the information finally received by the minister is fundamentally flawed due to 'deadline contraction' from the centre out. Thus, central teams are given a week to produce briefings, regional offices five days, HAs three days and trusts a day or two.

The desire to gather information towards the centre is a function of a number of political imperatives operating at ministerial level. First, the need to be accountable to Parliament leads to a desire fully to 'understand' the situation.

The consequences of this are poor-quality data, leading to inappropriate decision-making; the loss of capacity while NHS staff struggle to generate information for new data sets; the loss of goodwill at the accumulation of unreasonable requests and the local frustration that the information being requested is often the wrong information to diagnose the problem.

There is a danger of a real vicious circle. First, ministers request information not readily to hand on impossible timescales. The resulting poorquality data is used to inform policy and resource allocations. These central interventions are inappropriate and poorly focused on the real-world problems of local health economies. Consequently, the 'solutions' do not work, and unless the flaws of the information cycle itself are acknowledged, the hunt will be on for where the 'real'weakness lies.

One can imagine ministerial fury being vented on the service, or its clinicians and managers, for not, for example, 'matching resources with reform', and demanding to know why the problems persist. One can also imagine this perceived failure stimulating a new round of problem identification, data collection and policy review, again leading to failure.

The alternative is to let go of the idea of some politicised central management of the service. It is possible to put the 'national' back into the NHS by demanding uniform minimum standards, without dictating how this is done.

This 'step backwards' needs to occur in conjunction with a 'cross-cutting' commitment to local democratisation of, and public participation in, the health service and the public sector more generally. This would lay the foundations for more solid and genuinely responsive reform than the rather shaky underpinning that is emerging from the current top-down, centralist style.