Voltaire first mentioned this – he is supposed to have picked it fron an Italian aphorism. Shakespeare said it better, ‘Striving to better, oft we mar what’s well’ (King Lear). And as anyone who is in the business of problem solving outside the classroom would tell you, one of the practices that really distinguishes the experienced ones from beginners is this fundamental idea: a solution to a problem is rarely (if at all), a binary outcome. In fact, the very nature of most problems in a VUCA world is that they are difficult to define and are shifting in form and scope all the time. It is not unusual to be going back and forth on the problem definition itself during the solution process.
The other day, I read this great post by Paul Romer1 (if you have a list of modern day economists/thinkers that you follow, he is a must-have on the list) who has been a very strong advocate for rapid, wide-spread testing for Covid-19 ever since the pandemic blew up on our faces back in March. Right from the early days, governments the world over were on the hunt for accurate tests. The conventional wisdom (as it has always been) advocated that diagnostic tests should focus on minimizing false negatives – the general principle being that the cost of missing a diagnosis of an infected person is extremely high. And that would in normal circumstances, make complete sense: nobody would want a mis-diagnosed, asymptomatic individual spreading the virus around. However, one thing just about everyone missed was that as the world was waiting for an accurate test, the virus was spreading unchecked – forcing governments to take extreme steps of shutting down entire economies, which in turn caused a different set of problems. And all along, Paul Romer was arguing for an alternative strategy: aggressive, wide spread tests with selective isolation of population cohorts. His posts on this topic are worth a read (https://paulromer.net/covid-sim-part3/)
We are very often faced with making choices between trying to squeeze that extra bit of analysis (say, a 1% performance improvement from a model) and delivering the product on time. And as expected, making that choice is more of an art than a science. There is the story of British radio physicist, Robert Watson-Watt2 who was a pioneer in radar technology. In the lead up to World War-II, he was tasked with leading a team for developing radar technologies to counter Luftwaffe (German Airforce) and one of his key design principles was, as he characterized it, “cult of the imperfect,” which he defined as ““Give them the third-best to go on with; the second-best comes too late, [and] the best never comes.”
The ‘perfection fallacy’ and why do we get stuck in this
We all have been in problem solving situations where we assume that a perfect solution exists and then keep trying to achieve that solution. Why do we end up in such situations? There could be several reasons, but three stand out:
- There is no well understood definition of ‘perfect’ in the first place. If you are trying to build a transaction fraud monitoring system, what are the target false positive and false negative rates? There is obviously no straightforward answer – and it is subjective, contextual goals like these that end up with the analysts scurrying around trying to make up the ‘perfect’ solution
- There is no easy way to evaluate trade-offs between the effort and returns. The 80/20 rule is all fine but when you are in the middle of solving a problem, it is very hard to figure out when you the law of diminishing returns kicks in.
- The cost of a decision – this is the hardest and the trickiest one. When the solution requires significant investments (e.g. a system replacement; organization re-design etc.), the ‘loss aversion’ mindset kicks in and the teams end up seeking the ideal, fool-proof solution
What can we do about this?
I think that of all problem-solving challenges, this one would count among the toughest – and while there is no silver bullet, there are a few methods to consider:
- Simulation: I have talked about this earlier as well – consider building simulation models to evaluate how a certain solution recommendation, if implemented, would play out. It is not very difficult to create simulation scenarios – from relatively simple what-if scenario builders to more complex system simulations. And if designed well, they can be used to model the impact of decision choices incrementally and iterate towards an optimal (not necessarily perfect, but a Pareto optimal one in real circumstances). This is particularly useful in cases where the cost of a decision is high – as it would be in case of say, improving the transaction fraud rates.
- Experiments: The B2C e-commerce world has long used experiments to evaluate alternatives – I have often argued that this has been under-utilized in enterprise settings (both B2B customer interactions and even internal processes). I think there is a clear opportunity to design and execute experiments to iteratively evaluate the impact of design choices, especially cases where the cost of an error is not catastrophic. For instance, the Collections department in a bank can experiment with intervention strategies with their customers on an iterative basis to understand how customers respond to different nudges
- Mindset: It all comes down to the mindset. Managers need to push their teams to work in an iterative, agile manner. The most effective agile teams operate with a continuous improvement mindset – there is no reason that this can extend into all kinds of problem situations (not just software development). And to make this real, managers need to work with the analysts to keep measuring incremental progress – and constantly evaluate possible solution strategies at every stage. As Paul Romer has argued, combining Covid-19 tests with isolation strategies that are continuously re-designed/re-implemented may have helped us be in a much better place than the uncertain situation we find ourselves globally 5 months into this pandemic with no end in sight.
- https://paulromer.net/ : Really thought provoking posts, definitely worth following