Towards a driverless supply chain: the GPS paradox

Article

The less a GPS is followed, the more it needs to recalculate. This is what I decided to call the GPS paradox. Lots of decision systems, including supply chain departments are victim to it.

People who have my age remember the time that the first GPS navigation systems were launched. I used to have a friend who had one in his car but who followed his old road maps. It was funny to see how the navigation system was constantly recalculating routes that my friend was not following. This is what I decided to call the GPS paradox: the less a GPS is followed the more it needs to recalculate. Sometimes the navigation system proposed the road my friend wanted of course and then he was satisfied: "See I am right?"

These days very few people still use their navigation system like that. Most of us follow the directions with few deviations. We even use advanced systems that guide us through country roads we didn't know existed (this is true at least for the 65 million of Waze users, like me). And we get close to the point where we let algorithms not only decide the route to take but even to steer the car itself. The world has changed a lot.

The less a plan is followed, the more it needs to be recalculated.

Navigation systems are not the only decision systems that are subject to the GPS paradox. Entire supply chain departments are victim to it. Who does not know a company that has invested a lot of money in planning software but where the planner is constantly overruling the plan made by the system? Or a company where planners update the plan constantly because actual production does not follow their plan? Here the GPS paradox is still visible. An algorithm calculates and optimizes the plan, the human overrules it, the algorithm calculates again and again. Or supply chain calculates an optimized plan, production overrules it, supply chain calculates again and again. Maybe the number of times that a plan needs to be recalculated could be a good indicator of how poorly it is followed in operations?

What makes us inclined to overrule plans ?

What is the use of recalculating a plan if it is only to overrule it? When a plan is overruled are we really taking better decisions than if we would simply follow it? What is it that makes us so inclined to overrule plans made by other people or algorithms?

The knowledge that algorithms make mistakes. There is no doubt that planning algorithms will not always take the right decision. But nor will a supply planner. Or a production manager in a plant. The question is not “Does the algorithm always take the right decision?” but “Does the algorithm take better decisions than humans in the same situation?”
The difference in sensitivity for wrong decisions made by others and wrong decisions we made ourselves. Colleagues that proofread this blog mentioned the famous cases where people crashed because their navigation system told them to take a road that did not exist.
The need to feel that we have a choice. Another colleague told me that he always looks at the different routes Waze proposes and picks the one he thinks is best. The feeling that he picked the route makes it far easier afterwards to follow-up it.

Towards a driverless supply chain ?

The question is not “Does the algorithm always take the right decision?” but “Does the algorithm take better decisions than humans?”

With navigation systems, most of us have learnt that the algorithm might not take the road we had in mind but that it brings us to our destination efficiently and with far less effort. Wouldn't that be true about supply chain plans too?

Shouldn't we promote the same approach to supply chain and build processes where the plan is not to be overruled? Wouldn't that lead to more stability and predictability? And be a first step towards a driverless supply chain?