Simple solutions for complex problems - do they work?
When curious thinkers are gathered in a room, interesting conversations flow and intelligent ideas follow. This week, an interesting conversation emerged on key performance indicators (KPIs) for strength and conditioning (S&C) coaches.
There are a range of possible KPIs for S&C coaches. Taking the example of an S&C coach working in a team sport like soccer, one KPI could be the number of injuries that the playing squad sustain over the course of the season. Another possible KPI could be the subjective happiness of the players and coaching staff that work on a daily basis with the S&C coach in question. Or a different option might be a specific measure that the S&C coach is responsible for, like max strength in a number of key lifts, or 10/20/30m time over the course of the season showing if the players are getting stronger or faster? It could also be a combination of any number of these simple KPIs that could form the evaluation of a S&C coach's performance over the course of any given year.
On initial reading any of these metrics listed above may sound appealing - simple and easy to evaluate from one time stamp to another. However, these metrics bring to mind Goodhart's law:
Taking each of the KPIs mentioned previously and assessing it through the lens of Goodharts law, we can quickly see how flawed the metrics are.
In the case of injuries - low number of injuries being good and a high number of injuries being bad. If a coach wants to avoid injuries, the best way to achieve this is to do nothing. While doing nothing would be an extreme action to take and possibly unrealistic, this thought process could manifest in other, more moderate ways. An S&C coach may suggest to reduce training time from 60 to 45 minutes, or reduce the amount of high intensity activity that players do throughout the week.
Subjective opinions on a S&C coach can be very misleading. A good S&C coach can be disliked. A bad one can be liked. If a S&C coach wants to get a good evaluation, then they may exhibit behaviours that are not conducive to performance - reducing the "hard yards" players must complete, or becoming very agreeable with coaches instead of challenging them when ideas need to be challenged.
Reducing a S&C coach's performance down to players' physical capacities, will lead the coach to prioritise the physical capacities that they are evaluated against. This can lead to a host of issues, including stakeholders (sport coaches, physio, S&C) all working in silos, and the S&C coach prioritising certain physical capacities when all capacities should be developed.
The easy argument to all of this is that S&C coaches are not stupid or self-centred, and they would not sacrifice overall team performance for their own evaluation. And I agree (for the most part). It is important to note that this may not result in grand decisions, instead more subtle ones that may not make a big difference individually, but collectively will have a major impact of player performance over the longer time frame.
The same thought process can be applied to the impact of individual statistics on player performance. Using GPS metrics, for example high speed running (HSR), as a KPI for a player's performance is far too reductionist in my opinion. It can be argued that a team who worker "harder" than their opposition will create openings, but I would argue a team who works smarter - knowing when to run (timing to beat the offside trap) or where to run (overlap or underlap). There has been research to suggest that losing teams run more, and research to suggest that winning teams run more. This, to me, shows how more context needs to be added to the data collected on players, GPS or otherwise.
Complex Adaptive Systems and the Cynefin framework
I have written previously on how students can be considered complex adaptive systems (see here). The same can be said of athletes, coaches, or indeed humans - we are all complex adaptive systems, which can be defined as systems with many interacting parts (Davids et al, 2013). In a nutshell, this means that the behaviour of complex adaptive systems cannot be controlled, but rather influenced through shaping intentions. As a result, we (humans) demonstrate key features of non-linearity. Chow et al (2011) detailed the key features of non-linearity:
Non-proportionate changes: practicing for a long time may bring about small changes, practicing for a short time may bring about large changes. By assuming learning occurs in a linear fashion, this can bring about huge frustration, for practitioner and player.
Various solutions: there can be multiple ways to accomplish a task goal, as long as the intended outcome is achieved.
Scaling of task constraints: scaling of equipment can lead to the emergence of different behaviours.
Variability in practice: adaptation of movement behaviour over repetition of movement behaviour.
Reductionism - in the form of the KPIs listed above - can be useful to an extent. However, this provides limited insight into how an athlete has to adapt their technique to achieve their aim in chaotic environments like a game. This brings to mind the Cynefin framework (for more information on this framework, read here, or watch here):
This is a framework that helps people (leaders or otherwise) identify that every situation is different, and to help make sense of the problems that they face. Different kinds of solutions exist for different kinds of problems. Anyone who claims they have created a simple solution for a chaotic or a complex problem. is lying, and likely selling something. Almost all problems that arise in sport are complex or chaotic. Working in sport, with neurobiological systems (people), requires practitioners and pedagogues to embrace this complexity and chaos. Simple solutions in sport do not exist. Context is key at all times.
Davids, Keith & Hristovski, Robert & Araujo, Duarte & Balagué, Natàlia & Button, Chris & Passos, Pedro & Memmert, Daniel & McGarry, Tim & Travassos, Bruno & Vilar, Luís & Duarte, Ricardo & Kelso, Scott & Fernandes, Orlando. (2013). Complex Systems in Sports.
Chow, Jia Yi & Davids, Keith & Hristovski, Robert & Araujo, Duarte & Passos, Pedro. (2011). Nonlinear pedagogy: Learning design for self-organizing neurobiological systems. New Ideas in Psychology. 29. 189–200.