Why cricket and trees help you work with evidence

In this first of two planned blogs, I will touch on some thoughts on working with evidence during accident investigations. Much like my first blog ‘How to knit fog and herd cats at an aircraft accident site’ this blog will use some terrible analogies to inelegantly make my points. It is most definitely not a scholarly article, this is a blog to express my thoughts and observations based on personal experience of working with delegates on our safety and accident investigation courses here at Cranfield. Opinions will differ, but then they always do, and a healthy debate is no bad thing in itself.

So, in no particular batting order* (we’ll come to that later), the first five are:

When a Negative is a Positive

Sometimes the lack of, or absence of evidence is as powerful as the presence of evidence. We are seeking evidence to establish what happened, how it happened and why it happened. So, having evidence, and hopefully multiple mutually supporting sources of evidence, is a really good thing. Sometimes though, a lack of evidence is also as informative. A good example would be rotating helicopter components such as main, intermediate and tail rotor gearboxes. These are ‘single points of failure’ with very serious consequences for continued safe flight if they fail. As a routine, oil (or grease) samples would be harvested for laboratory testing to establish the health of the transmission system. If the testing of the oil samples fails to find any evidence of abnormalities within the gearbox, then this can be eliminated as a causal factor, contributory factor or safety issue. So, whilst we might still be trying to establish ‘what happened’ we now know something that ‘definitely did not happen’.

Think Before You Leap!

When evidence is analysed, it is a process of logical thinking about the evidence to reach a conclusion on its meaning and its importance to the investigation. This logical thinking, or reasoning, can be deductive or inductive. There is a huge volume of work describing deductive and inductive reasoning which I will not bore you with here but for some reason seems to involve a lot of black furry cats with sharp teeth. What is vitally important as an investigator is to understand the difference and recognise when you are taking deductive or inductive ‘leaps’ when working your evidence. An analogy is crossing a stream using stepping stones. When you cross a stream using deductive steppingstones, they are all of the same size and distance apart and you can clearly see the other bank. You can then safely ‘leap’ merrily from one steppingstone to the other across the stream and reach a confident evidence-based conclusion on the other side. Inductive steppingstones however are tricky little things. They can vary in both their size and distance apart, they can be a bit wobbly, and you cannot quite see the opposite bank as you set off. To complicate matters further, there may even be more than one steppingstone path to choose from as you move from stone to stone. Consequently, each inductive ‘leap’ you take is more unsafe and unsteady than the last and when you finally reach the other side you may not necessarily have reached the right conclusion. Working evidence inductively can often be a necessity but a wise investigator considers carefully any inductive ‘leaps’ they take toward their conclusions.

Fail to Plan – Plan to Fail

Whilst the adage ‘no plan survives first contact’ is a truism, it is without a doubt always better to have an evidence plan than to not; ‘fail to plan, plan to fail’ springs to mind and this so very true for safety investigations. The plan can be as simple or as complex as the circumstances dictate i.e. you would need a far more robust and comprehensive evidence plan when investigating a cruise ship event (very large organisation, complex systems, multiple actors) as opposed to a single manned/owned fishing vessel (very small organisation, simple (relative) system, single actor). At the very least it will ensure you are thorough in your evidence collection but also efficient as time is rarely on your side! Experienced investigators have typically worked out what evidence planning technique or methodology works best for them. Some like flexibility and may chose mind mapping techniques whilst others might favour a structured approach by using a model such as SHELL** to ensure they ‘cover the bases’. We strongly encourage our delegates to develop an evidence plan ahead of their accident simulation. It is readily obvious at the simulated accident site who has planned, and who has not and, who is following the plan, and who is not!

Sort Out Your Batting Order

As part of the evidence planning process above the investigator must also make a subjective assessment of the priority order in which they need to collect the evidence identified in their plan. In simple terms, they need to consider two factors. Firstly, ‘what evidence is likely to be critical to my understanding of what happened and why?” For example, recorded cockpit, bridge or locomotive voice data can be pivotal in determining how the operators were responding to unfolding events and is therefore critical by default. Secondly, ‘what evidence may be perishable and if I do not take immediate steps to capture that evidence it will disappear never to return’ e.g. delicate ground marks on soft muddy soil. Quite often key evidence will be both critical and perishable placing it at the very top of the evidence plan priority list e.g. the FDR and CVR from Air France flight 447.

See the Wood from the Trees***

One of the perennial problems we witness during our accident simulations is the difficulty inexperienced investigators find in determining what is important evidence and what is just ‘stuff’ hanging around at the simulated accident site. The solution to this, unfortunately, is experience and there is no easy route other than to learn as you go. However, here are some observations to help sort the ‘wood from the trees’ or, in other words, how to think about the problem logically and break the problem down into simple systems thinking. One of our accident simulations involves investigating the sinking of a fishing vessel. The fishing vessel has been salvaged and the investigators are given the opportunity to inspect the vessel and gather evidence. The salvaged vessel and its surrounding areas present an extremely chaotic and cluttered scene, and it can be very challenging for the investigators to work out what is critical/important evidence and what is not i.e. ‘see the wood from the trees’. In order to help plan what evidence they need to determine ‘what happened’ they need to ask themselves some basic question. One of these basic questions, amongst others, is:

- What was the vessel (system) doing when the accident occurred?

In this case, the vessel could have been in one of three phrases. Either sailing out to the intended fishing area, in the act of fishing or transiting back to harbour once fishing operations are complete. As part of their pre-site visit evidence planning we give the investigators some preliminary evidence indicating that the vessel was most likely engaged in the act of fishing, so they can immediately narrow their focus to the fishing phase. This now breaks down to a Human Machine Interface (HMI) question, ‘what were the humans and the vessel, as a fishing system, likely doing when things went wrong?’ To ‘fish’ the crew need the fishing gear system and all its ancillaries (net, trawl wires etc); a winch system to ‘shoot’, ‘trawl’ and ‘haul; the heavy fishing gear system and vessel controls in the wheelhouse to control the speed and direction over the vessel over the water. In order to determine what evidence they should include in their planning they need to ask themselves the next question:

- What phase of fishing (process) was the vessel (system) undertaking when the accident occurred?

If the vessel was fishing it would either be ‘shooting’ the fishing gear to start fishing, in the ‘trawl’ i.e. towing the fishing gear to catch fish or ‘hauling’ the fishing gear back on board to land the catch. The vessel’s systems need to be in distinct configurations for each phase e.g. winch brakes on or off, winch clutches in or out etc. So, by asking themselves the question:

- In what configuration are the systems of interest?

The investigators can then focus some of their activities on site, amongst others, on determining the configuration of the fishing gear, fishing gear winch and wheelhouse controls. Gathering this evidence will then allow the investigators to deductively conclude the phase of fishing the vessel and crew were engaged in at the moment of the accident i.e. ‘what they were doing’. Once they understand what the system was doing, they can then start to put the pieces of the jigsaw puzzle together to determine ‘what went wrong and why?'

Well, that’s all for this blog. In my next blog we will look at some other aspects of working with evidence including some exciting topics such as ‘release the inner five year old – don’t take things at face value’ and ‘signposts – listen to the evidence’.

*https://en.wikipedia.org/wiki/Batting_order_(cricket)

**https://en.wikipedia.org/wiki/SHELL_model

***https://www.collinsdictionary.com/dictionary/english/cant-see-the-wood-for-the-trees