With almost every endeavour participants have many more ideas on what might be done that than there are resources to get them done. Prioritisation is thus essential, but, my goodness, it’s hard. Many groups will slide back to making almost everything a priority, which means, of course, that nothing is a priority. I reflected on all this as I participated in a WHO project to prioritise research into the non-communicable disease with particular reference to low and middle income countries. I thought too of Britain’s politicians having to prioritise budget cuts.

The first step in a prioritisation exercise is to develop a list of what might be done. A logical way to go about this is to divide the territory and ask groups or individuals, let’s call them “experts,” to generate a list for that territory. With non-communicable disease it was natural to divide the territory into the four diseases (cardiovascular disease, diabetes, chronic obstructive pulmonary disease, and common cancers), the three risk factors (tobacco, poor diet, and physical inactivity), and some important cross cutting issues (primary care, social determinants, and getting research into practice) plus something that seemed important when the process began (genetics) but less important as it reached its conclusion.

Inevitably the “experts” come up with formidable lists for their territory, about which they are bound to be enthusiastic and from which they earn their living. Creating lots of territories has to produce lots of priorities – in our case around 150, far too many. But what number of priorities is right? WHO is producing its priorities for a diverse audience including ministers of health in low and middle income countries, international donors, researchers, health professionals, and many others. Maybe with such a large and diverse audience it doesn’t matter if there are more than a 100 priorities, but most felt that action depended on drastically reducing priorities, perhaps to just five.

So how to reduce 150 strategies to five? One way to start is to force those within the territories, let’s call them silos, to cut their priorities down to five and rank them in order of priority. This was tried, but the easiest way to cut perhaps 15 priorities to five is to combine them. This generally was what happened, producing unwieldy single priorities that were major research programmes in themselves.

Maybe the experts do this because they fear if they are too energetic in cutting their priorities then other silos will benefit. This is a version of the “prisoners’ dilemma.” Or maybe it feels to the experts like prioritising their children, all the priorities are equally important.

Another way to cut 150 priorities to five is to take a few people from different silos, lock them in a room, and tell them they can’t come out until they have five focused and achievable priorities. We tried this, and other people’s priorities, just like other people’s prose and budgets, are much easier to cut than your own.

But cutting across territories is an “apples and oranges” problem, meaning that direct comparisons are impossible. And what criteria should be used? We thought about impact (but on what?), achievability, evidence base, cost effectiveness, affordability, and time frame. But no matter how many numbers can be produced under each heading these are still synthetic and subjective judgments, the stuff of politics. It means that the group will be swayed by the most powerful, articulate, and well connected.

As we grappled with our priority exercise – which might not anyway have much impact in the real world as ultimately funders and researchers will make their own decisions – I thought of the politicians in Britain’s star chamber prioritising cuts to government departments, cuts that will have a very real impact on people’s lives. Priority setting is a messy and hard business, which is why we avoid it if we possibly can – even to the extent of simply pretending to do it. But this choice wasn’t open to the politicians, and as all shades of opinion accept that some cuts were necessary we should perhaps congratulate the politicians for having done it, even if we disagree violently with some of their judgments.

Richard Smith is a former editor of the BMJ.