Whilst putting VALID together with the Risk Professor, and Bath University, I learned many lessons about risk modelling. One lesson learned was that I’m an Arborist. I mainly know about trees. There’s much I didn’t know about risk modelling. Okay, I did know risk matrices are fundamentally flawed and can’t sensibly rank risks. And you can’t apply mathematical rules to ordinal rankings. Though I was yet to find out how much I didn’t know, I was at least smart enough to realise some people knew much more than me.

Another important lesson learned was there’s too much uncertainty in tree risk to claim single figure value accuracy. It’s not credible to measure tree risk to such accuracy as 1/4, 1/300, 1/20 000, or 1/5 000 000. Nor can you realistically measure the difference between a 1/10 000 and a 1/50 000 risk. And you really shouldn’t be compounding the error by adjusting these very accurate risks by double or single figure multipliers like 0.25 or 2, 3, or 4.

With tree risk, we’re looking to measure something with high uncertainty, and our risk ratings should reflect that. With VALID there are only four risk ratings. None of them is a single figure value.