Post hoc, trees are dangerous

Last night (11/4/2012) I took part in an interesting Radio 3 discussion programme called Night Waves . The first contributor, Jonathan Haidt, was fascinating and I’ve just ordered his new book The Righteous Mind.

In the discussion he made a point that resonated with a problem that I have been wrestling with: discussing the relationship between reason and emotion he said, “reason is mostly post hoc”. Risk management decisions are taken pre hoc. They are decisions taken with insufficient information. They are decisions about an uncertain future. Post hoc, after the event/accident, there is usually much more information to which one can apply cause and effect reasoning. Before the event the links between causes and effects are either unknown or shrouded in estimates of probability, with large error bands attached. The information gap confronting a risk manager is readily filled by “emotion” – the natural instinctive state of mind deriving from one’s circumstances, mood, or relationships with others.

Event Trees, beloved of engineers and project managers as a method for describing the risks associated with big projects, represent a denial of this reality (see Dangerous Trees?). All of the branching points on each limb of their trees must be assigned probabilities, usually with large error bands. By the time one reaches the tip ends of the branches the compound error bands are too large to fit on the page, so the magnitude of the uncertainty is ignored.

In the dangerous trees example just referred to an actual tree blew down and killed some one. In Britain, averaged over the whole country, the statistical risk of such a thing happening in a given year is about 1 in 10 million. Britain’s Health and Safety Executive describes risks of 1 in 1 million as “adequately controlled”. But after the event different sorts of reasoning are brought to bear. The police, who are predisposed to allocate blame in such circumstances, arrested the head estate warden and the estate manager on suspicion of negligent manslaughter. The coroner declared, a year and a half later, that all mature trees on the estate’s land adjacent to paths should be inspected and logged – a requirement that, if applied to all the National Trust’s properties (whose estate it was) might have bankrupted the institution.

Post hoc Event Trees are often transformed into Fault Trees that, with the aid of clear facts and “reason”, are transformed into trees with only one branch called “culpable negligence”. What pre hoc appeared a risk of 1 in 10 million, post hoc becomes a risk of 1 in 1.

After a year and a half the Crown Prosecution Service, dropped the case for lack of evidence, but a threat of dire consequence still hangs over the National Trust. Given the many millions of mature trees in Britain, the risk assessment and risk management implications of the coroner’s judgment, pursued to their logical conclusion would require the diversion of enormous resources to inspection and/or the execution of countless mature trees that could not be guaranteed 100 percent safe – and who would offer that assurance in our litigious times? This experience is now colouring the (emotional?) judgment of Britain’s tree managers. In deciding what trees must be executed the legal risk now far outweighs the physical risk.


No ping yet

  1. Steven Hope says:

    When the risk of 1 in 10 million has become 1 in 1, probability has become confused with possibility. Organisations are now acting on the basis that there is possibly the possibility that something can happen.

  2. Andy Reynolds says:

    John, I think you have raised a really key point here. Risk perception has been greatly skewed by post-hoc assessments. Coroners have caused the Ministry of Defence enormous trouble with post-hoc assessments of force protection measures, for example. The role of doctors in promoting safety measures such as bicycle helmets has gone way beyond their biomechanical event reconstruction expertise. Finally, we have the insidious exploitation of overt emotional responses – testimony of grieving relatives being harvested in the pursuit of “saving even just one life”. There are actually people out there who believe that the bereaved are the most highly qualified risk advisers! And of course, the bereaved are the coroner’s front-row audience.

  3. Mark Daniels says:

    I think it is fair to say that “a threat of dire consequence” no longer hangs over the National Trust in relation to this 2005 accident at Dunham Massey, in that the HSE finally concluded in 2009 that there was insufficient evidence to bring charges under the Health and Safety at Work Act. And in 2011, the NT successfully defended a civil case in relation to another tree-related fatality that occurred at another property in 2007. Since then, the National Tree Safety Group has published its guidance “Common sense risk management of trees”. Nevertheless, other organisations are now in the firing line, with two separate criminal cases being brought later this month against a private zoo owner and a national woodland charity following falling tree accidents.

  4. Dr. Robert Davis says:


    Trees (the ones with leaves on them that come out of the ground and sometimes fall over) have long been a favourite target for “road safety” practitioners.

    Any Highway Aouthority that has failed to remove road side trees at locations where motorists may drive off the road into them may be at risk of legal action from those representing the driver (and/or their relatives).

    This brings in a crucial political/cultural dimension. We are not just talking about a body (such as zoo owners, the National Trust etc.) who have to be extra careful about a naturally occurring phenomenon such as trees falling over. We are talking about how traditionbal “road safety” ideology demands that the needs of careless/dangerous/criminally negligent motorists be met. Of course, this will inevitably occur at the expense of other road users.

    For more on this see: http://rdrf.org.uk/2012/02/blaming-bollards-and-trees-%e2%80%93-and-why-it%e2%80%99s-important/ .

    It’s all about a more civilised way of addressing this kind of risk.

  5. P Buddery says:

    Very many things are at least slightly dangerous, and we must accept that life contains risk. Walks by the side of my local lake are, like socks or pet animals, very pleasant but slightly dangerous.

    As a design engineer, I used to spend quite a lot of time in safety meetings – often asleep – and there is a natural tendency to explore the risks, and implications of the risks, other risks concequently discovered, and so on. Even given the vivid originality shown by stupidity victims in the manner of their own deaths, there is a tendency for such meetings to diverge from reality. There is a tendency for such meetings to advocate methods that are inconvenient or impractical and that may even carry dangers of their own.

    Bicycle helmets are a good example of this. Theoretically a fine idea, in reality they can cause serious cranial overheating effects, leading in extreme cases to homicidal mania or death through heat-caused brain damage.

    To conclude, people need to be sensible and a bit brave.

  6. Matthew Squair says:

    Joining in late I’m wondering whether this is actually an example of the importance of framing effects in risk perception?

    People significantly prefer to take risks about events that may occur in the future, rather than events that have occurred in the past even though the probability of the events are exactly the same. While there’s lots of debate about ‘why’ there’s no argument that this is a robust experimentally repeatable effect.

    This bias in preference could explain, at least in part, why there is such a difference in the assessment, and perception of acceptability, of a ‘risk’ before and after an accident.

    There’s also been some work done in the legal field on whether rule based or standards based legislation is more effective due to these effects as well, I wrote a piece on their efforts from a safety perspective (at the link) in fact.


Leave a Reply

Your email address will not be published.

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>