Led by science
Under the current pandemic situation, governments are relying on the advice from scientific bodies. This seems to be the right approach, as only they have the knowledge to understand what is going on and how to control the parameters that are deemed important. Similar to supply chain management issues, scientific models can be used to understand the behavior of complex systems and to predict the impact of measures taken. However, in supply chain management, we have learned that there should be a healthy balance between the vision of a company, the common sense of practitioners working there, and the techniques and models proposed by scientists. The same goes for the relation between politicians and scientists on Covid measures.
To centralize or not centralize?
A fictional company was producing products using a complex supply chain, and market demands were forcing the company to review its supply chain planning and control structure. As complexity was making it difficult for the human planners to make the right decisions under constantly changing circumstances, an advanced planning & scheduling system (APS) was implemented to support the human planners. With the introduction of the APS, planning and scheduling decisions were centralized, meaning that decisions on the shop floor, about what to produce next, were not allowed anymore. The production control framework that was used, was in line with the hierarchical planning paradigm (HPP), which has been described in many textbooks. This approach in structuring planning decisions is lectured to thousands of students every year.
A few years after the APS was implemented, the company embarked on a lean manufacturing program. A senior lean expert was hired, with deep knowledge of the lean principles. He rejected the idea of centralized planning, and started an initiative to push the scheduling decisions to the shop floor. He advocated that the APS should not be used anymore – instead, shop floor operators were much better able to see what was happening and to react to disturbances. So the one approach was replaced by the other, and the overall supply chain performance did not change much.
Now, which approach is better? Within the context of the HPP or Lean, it would be clear what to do and what effect to expect. But when to apply HPP and when Lean, and do these approaches exclude each other, or can these be used at the same time? The answer to this, depends on which expert the question is asked. Within a paradigm or a model, the relation between measures and outcomes are clear, because this is the way the model was designed. But designing a quantitative model always comes with assumptions and simplifications, as a tradeoff needs to be made between the complexity and completeness of the model. Elements that are excluded will not heavily influence the outcome of the model in a particular situation known by the expert, but another situation might invalidate the model.
Models are contextual
This means that applying scientific models should always be done by first understanding how the model was constructed, what assumptions were used and where it functioned well. Even more so, a company (country) should ask itself, whether the people that constructed the model, actually implemented something successfully in practice, using that model. Skin in the game, as Nicholas Taleb calls it. A choice of model also depends on what a company wants to commit to – does it want to go for supply chain performance mainly, which would justify centralization, or do they also want to empower people as much as possible, which would justify decentralization.
Apply models with care
As a practitioner, never be daunted by the in-depth knowledge a scientist or expert demonstrates. Such knowledge is (probably) valid within a certain domain, with certain assumptions, but it is the practitioners (politicians) job to assess the applicability of the models into the company (society) that (s)he knows well. In line with the example described above: do not expect a mathematician to take the empowerment aspect into account too much, just like the sociologist will not care too much about mathematical optimization (forgive me for stereotyping). Similarly, epidemiologists will focus on ending pandemics, and not so much on controlling collateral effects of measures taken. Scientists will ask themselves whether models speak the truth, practitioners should ask themselves when models will work.