It was interesting this weekend to listen to Nikki Lauda team boss at the Mercedes Formula 1 team talk about the measures they have put in place to ensure that they never again make a mistake like the one they made at the Monaco Grand Prix two weeks ago. For anyone who doesn’t follow Formula 1, the mistake in question was a decision to call race-leader Lewis Hamilton in for an unnecessary pit-stop with just 13 laps left to go in the race. While he was stopped in the pits, Hamilton was passed by championship rivals Nico Rosberg and Sebastian Vettel, leaving him to finish in third place instead of what looked like a certain win.
From the analysis that has emerged since the race in Monaco, it looks like one of the main reasons that his team called Hamilton into the pits was a reliance on bad data. The built-up nature of the surroundings of the track at Monaco means that the GPS data feed that teams use to follow the position of all cars on the race track doesn’t work as well there as at other places. This meant that the automated strategy analysis system that Mercedes used to make a call on whether to pit Hamilton or not was using unreliable data about his position, and the positions of Rosberg and Vettel. It seems that the system calculated that there would be time for Hamilton to make his pit-stop and still return to the track ahead of the other two drivers. As Toto Wolff the Mercedes Head of Motorsport said in an interview with formula1.com: “Now in hindsight I have to confess that the data was wrong.”
The pit-stop allowed Hamilton to change to a fresh set of tyres which ensured that his car performed optimally for the last section of the race. This optimal performance was of no value, however, as he was stuck behind Rosberg and Vettel. The thing that confused people was that there was no real need for Hamilton to make the pit-stop. He was a significant distance ahead of his rivals and there was very little of the race left – it was extremely unlikely that either Rosberg or Vettel would have been able to pass Hamilton before the end of the race. This led to many to accuse Mercedes of abandoning common sense and overly relying on data-driven decisions. Toto Wolff addressed this directly in the interview with formula1.com: “We were in a situation of waging common sense against data. Common sense is okay, but it doesn’t win races in the long run. You have to rely on data – and now we have to find out why we got it wrong today.”
This incident, and the commentary around it, raises the interesting question of how best to strike the balance between making decisions based on common sense (or gut instinct, intuition or any other similarly vague notion) or based on data. Toto Wolff is obviously coming down strongly on the data side! We explored this a little in our recent Irish Data Analytics Landscape Survey 2014-2015. In one of the questions on the survey we asked participants “What are the major drivers of the use of data analytics for decision making in your organisation?“. The distribution of the responses given to the question is shown in the image below.
One of the motivations for asking this question was to compare the number of people who chose a “Ensuring greater accuracy in decision making” with the number who chose “Removing gut instinct from decision making”. It is interesting to note that more people chose the first, which doesn’t take the hard line of removing gut instinct from decision making. There are occasions when the output from data-driven models is based on bad data or doesn’t take into account important factors that can’t quite be captured in a model. Sometimes a little gut instinct can help to identify these occasions.
Returning to Nikki Lauda, he did not hold back when discussing the mistake made by Mercedes at Monaco with the BBC: “A top team should not make mistakes like this. I am really upset because it was not necessary. It was the wrong decision to bring him in. It was very obvious. There was no risk to keep him out.” In an interview with David Coulthard during the BBC’s coverage of the Canadian Grand Prix this weekend Lauda explained that from now on they will increase the amount of human examination of the output from the automated strategy system so as to try to find that perfect balance between data and domain expertise. This is the same balance that so many of us try to strike.