Death of a Thousand Paperclips

When we talk to clients about the almost universal problems they have gathering basic data (for any purpose), we sometimes refer to this as the “death of a thousand paperclips” effect. Put succinctly, it refers to a condition where an enterprise spends its time generating enormous reams of paper (whether on files on a machine or printed) that cannot be associated. Very often they have all the necessary data, but just have none of it consolidated in any form that can be accessed, so the majority of their effort can be spent accessing data of questionable quality, which is also difficult to relate contextually.

Part of the problem is that right now traditional safety isn’t participating in management; it is counting activities and generating paper. This is true even to the point where training generates certificates without ever having a way to indicate the value of those papers. Into this mix we have seen numerous additional paper efforts, all of them predicated on some belief that all the past bad paper somehow was itself bad, rather than that it became bad because it was unmanaged. So, we see client companies with massive competency documents describing what a welder should know, who still cannot reliably tell us what Fred Smith actually does. This disconnect is because nothing is related in any way that is reliable; it shows what one used to call a lack of bottom. In plain terms, without a foundation process no amount of tweaking ever produces related value, only larger stacks of irrelevancies.

Form-filling software is state of the art in the traditional safety world. (And had we been less interested in actual management, we would have taken that path to glory.) The problem is form-filling generates forms, not value; and value is the measure by which all processes are actually judged. When you have a stack of forms ten miles high, you can neither find what you need quickly, nor discover relationships easily.

This is another point where the tooling behind a solution-set becomes part of the value proposition, and this is where the payoff comes in the models we developed, where risk factors and controls suddenly link across a broad array of profiling, active and reactive systems to produce opportunities to link and analyse data, address issues of data quality, and generate reams of paper that communicate rather than frustrate.

When we ask a company which has suffered a workplace fatality to be objective, given the current state of traditional safety they will begin to count things. They have to, since they have no other choice. They will fill out forms, counting the fields they can complete, producing an inscrutable but perfect form. They will then attempt to subjectively qualify the fatality, because really they have nothing to draw upon to test any hypothesis. They cannot, for example, really analyse this fatality in context of fifty similar near misses to identify what control failed and how it failed. Nor can they analyse the employees training instantly, relating the risk factors that they encountered that killed them against the ones the training they had protected against. Nor can they analyse what training the occupation they had should have had versus what they did have. All that can happen in traditional forms is during an emotionally distressful instant checkboxes can be checked, counted, and filed.

Real value is lost when outcomes are believed to be systems, because when the actual underlying systems fail the outcomes is shattered. In essence, the focus on outcome obscures the cause in favour of a focus on the effect. No one can be objective when the outcome becomes the exclusive focus.

Our systemic approach though isn’t about safety at all, but about risk-management. Even in the worst scenarios it is easy to be objective (generating value) when your entire process is able to answer questions that matter. If someone is dead, that won’t change; but the next person who is killed could be saved if the system can answer exactly what controls failed to allow the death, exposing faulty controls we can universally rectify, or an absence of control, or unknown risks. Even in the purely reactive conditions imposed by an accident, a risk-management model is focused not on the fatality, but on the mechanics of how to prevent the next probable fatality. It is less interested in blame than focused on avoiding repetitive incidents. Our system can tell you what the person was doing, whether they should have been doing it, what risk they encountered, what the expected severity of that risk was, what specific controls attached to that risk failed, and even go so far as to tell you whether the dead worker should have been doing the job at all given their training controls.

Death of a thousand paperclips kills the ability of traditional safety to impart a defence against disaster. No matter how you treat an accident under the traditional approaches, you have no objective weight to any outcomes, because the circumstances of the outcome provide too much focus potential. Subjective analysis is, simply, not real analysis.

Risk-management tooling is the critical difference between effective risk-management and is where our research and development effort shows real value. It is never good enough to pay lip service to an idea that can be qualified, quantified, and cyclically improved.