The child who tries to open a door has to manipulate the handle (the input) so as to produce the desired movement at the latch (the output); and he has to learn how to control the one by the other without being able to see the internal mechanism that links them. In our daily lives we are confronted at every turn with systems whose internal mechanisms are not fully open to inspection, and which must be treated by the methods appropriate to the Black Box.
Ross Ashby, 1956
Black box is a term used in systems theory to describe a system whereby only the inputs and outputs are observable, not its inner workings.
The term was used in education by Paul Black and Dylan Wiliam to describe the inner workings of the classroom in their seminal paper, Inside the Black Box (2001). Strictly speaking, the classroom is not a black box as its walls are not opaque, and we can walk in at any time and see what’s going on! However, Black and Wiliam’s point was that educational policy seems to treat the classroom as if it is a black box, with teachers left to work out how to deliver the desired outputs when demands are made by management and policy makers.
Perhaps the most problematic true black box in schools is the human mind. The history of psychology is divided by those who believe it is possible to peer inside the mind to speculate on its inner workings and those who study only the outputs that result from certain stimuli and environmental conditions.
But I want to talk about a different kind of black box in schools, namely the ‘closed door’ decision making that only a select few – those making these decisions – witness.
We can be fairly transparent about decision making in most areas of school life. If we re-tender the contract for school transport, we open up the process so that trustees and auditors can assure themselves that best-value decisions have been made. If we ask staff to adopt a new behaviour management approach, we can set out the intent and assumptions for how this approach will be more effective than the last. If we redesign our curriculum, there is no secrecy about what content we choose to include, and how this is sequenced to best effect.
However, some decision making is more opaque, either by necessity or design. Such black box decision making includes processes that determine the suspension of pupils, staff conduct, safeguarding, or critical incident response. The opaqueness of these systems is often due to the need for confidentiality. This means that it is difficult to share the specifics of decision making in these domains; we can only talk hypothetically and in terms of processes and principles, without the richness that examples bring.
Another reason for opaqueness in decision making is that things move too quickly and there is no opportunity for transparency. It is almost four years to the day that the pandemic caused schools to close. This was a period of absolute opaqueness as those in the room as critical decisions were made barely had time to process the implications of their decisions themselves, let alone lay out their rationale for others.
Black box decisions can lead to organisational problems. Those on the outside can only speculate about how and why decisions have been made on the basis of the inputs and outputs they observe. In my experience, that speculation is often inaccurate.
Let’s take the example of the decision making process that goes on when a headteacher decides to suspend pupils from school. From the outside, we can observe inputs and outputs. Child A swears at a teacher and is suspended, but Child B commits the same offence and is not. Child C is suspended because they graffiti a toilet, whereas Child D is not suspended for what appears to be a more serious offence. The conclusion we might reach from outside of the black box is that the decision making process is inconsistent. We may assume that inside the black box there is a decision making algorithm that seeks to match offences to sanction, and that there is a linear, sliding, relationship between one and the other. However, what we observe may lead us to believe that the black box is faulty.
Now let’s peer inside the black box.
Black box decision making systems like the process that decides suspensions have a number of features. First, there are rules. Rules exist to guide decision makers and are often in the form of policy or official guidance (sometimes statutory). There will be a behaviour policy signed off by governors, statutory guidance around safeguarding, and an emergency plan to guide anything from snow days to lockdown.
Second, there is precedent. Many black box decisions are referenced against past decisions to achieve consistency over time. Precedent is rarely recorded so exists in the memories of decision makers. This means its influence on decision making varies according to whether there is a change in who is making the decision.
Third, there are the mental models of those making the decisions. These are themselves black boxes. If the decision maker has expertise in the domain, they will tend to rely on their mental model more than on the rules and precedent (although their mental model will have been shaped by repeated application of the rules as a novice). Mental models include decision making heuristics. These are ‘handrails’ which act as shortcuts and ‘red lines’ which the decision maker will note if crossed. For example, a headteacher may have a handrail that leads them directly into suspension territory if there is abuse of a member of staff and a red line like physical threat to a member of staff that removes much need for nuanced thinking. Heuristics are valuable in fast-paced environments but can lead to problems when overly relied upon. Mental models will also incorporate values held by the decision maker. For example, physical assault may be valued as more serious than verbal assault.
Finally, there are the inputs: in the suspension example, this is information about what has happened, and why, and with what effect? These inputs will also include disinformation and blind spots.
Inside the black box is a process, but not an algorithm. The process is sense-making. Sense-making is the mind’s attempt to give meaning to the information at hand. Black box decision making systems will often intentionally involve more than one person so that a shared situational awareness can be achieved. Collaborative decision making inside the black box mitigates against myopic or prejudiced behaviour.
Black box decision making can be rationally consistent when viewed from within the box but appear irrational to those outside of the box. This can lead to loss of confidence in decision makers. This is a pressure of headteacher’s jobs that doesn’t often get talked about. Often, the headteacher is unable to breach confidences by setting out the decision making logic and can only address criticisms by talking in the abstract and through hypotheticals.
In my experience, the frequency and difficulty of black box decisions is increasing in schools. At the same time, the demands for transparency are also rising. On the one hand, there are more parental complaints, allegations against staff, complex behavioural issues, and expectations for what schools can and should do. On the other, freedom of information and subject access rights are being used to do a post-hoc critique of decisions being made in pressured circumstances. The heat is on.
So, what can we do to ensure that black box decision making stands up to increasing levels of scrutiny?
We should begin by ensuring our rule-books are fit for purpose. It sounds dull, but policy review processes must take the time to reflect on whether the policy was actually helpful and led to better outcomes.
Next, we can run simulations to provide a low-stakes environment for decision makers to test out their mental models. These may range from desktop emergency response role-play to lockdown drills.
Last, we should carry out regular case reviews where the thinking inside the black box is exposed to scrutiny.
We might also develop a language around black box decision making in our institutions so that we all know what it is and why it exists. If those who sit outside of the black box observing the inputs and outputs can gain some insight into the internal mechanics they may be less likely to judge harshly when the logic of a decision is not obvious.
Matthew Evans 