This week, I talked with Ricardo Valerdi, Professor in Systems and Industrial Engineering at the University of Arizona (link to the episode here). We talked about Ricardo’s research in cost estimating, how he applied estimating principles to sports analytics, and how storytelling with data is an important skill for the future estimator.
What is Cosysmo?
Cocomo is focused on estimating software development cost. The model has ben used for over 40 years and there is a whole community of users and researchers within this ecosystem now developing systems engineering cost models.
Cosysmo focuses on systems engineering as part of a whole project where the other elements for example are hardware or mechanical engineering.
Cosysmo or Cocomo can’t be used too early as they need a certain amount of information. For example an estimate of the amount of code or function points or object points in that project or an estimate of the amount of requirements or interfaces. You need some basic information or make reasonable assumptions on that basic information.
Typically this happens after the project has gone through systems engineering or architecting (up front design iteration). It takes some effort to get to this point, but the better information you have the more confidence you can have that the model is meaningful.
How did you make the connection between cost estimating and sport?
Reading the book Moneyball by Michael Lewis – the story of a Baseball Team (Oakland Athletics) – who were able to find inefficiencies in the economics of the sport and were able to exploit it to win more games for a lower cost.
So immediately, there is a cost connection between baseball and military defence. Beyond that it is about measuring human performance. Baseball requires the right players in the right positions, you want to be competitive to win. This is the same as an engineering firm which wants to win contracts and to execute projects successfully.
I broadened it to include basketball, NFL and Olympic sports. There have been a lot of connections both physically and analytically.
What is the next revolution in cost estimating and sports analysis?
This is something we have been thinking about in cost for decades. The tension in sports is whether to trust the data or to go with intuition and experience.
As a coach are you going to interpret in real-time what you see on the field or are you going to look at your data and do what the analytics say? If they contradict each other, what do you trust – the data or human intuition?
This is the holy grail – there are no absolutes – knowing when to switch from one mode to the other is the key to outperforming your opponent.
How can estimators or designers account for externalities in the future?
We need to take a step back – what are our models missing – what are they not accounting for?
We have to ask ourselves how we combine expert judgement with historical data. This is less binary, and more about combining the two and accounting for each sides weaknesses. The balance is knowing when to tune each parameter.
You need to have enough awareness of why you are doing this, how you are tuning from one side to the other. Almost like a meta level machine learning. This is navigating from the science to the art. We need to be willing to take that leap.
Telling a story with data – how does this apply with estimating
All of us can get better at this. This is all about persuasion – there are two ways (in a simple way) – either give them rhetoric (facts, data, etc), or tell them a story.
Most of us as engineers would default to rhetoric, but we can learn a lot about the story telling approach.
I’m a big fan of Pixar studios – who produce stories with the same formula. It works. They know these stories trigger an emotional response from the audience that grabs you.
So how do we borrow this formula? The approach does require some important adjustments. Traditional storytelling is about building suspense to draw people into your story and keep them at the edge of their seat.
For estimating it is more important to give a call to action up front rather than keep the audience in suspense.
Start with a call to action end with the call to action, and in the middle is the analysis and methodology. The objective is to persuade. The protagonist in this story is the audience, as opposed to the hero of story in a traditional story.
What is the Science of Sport programme?
It’s really motivated by upstream thinking. If you want to impact young students choosing STEM careers, you have to influence them at a young age.
Think about your younger self, and think about the point where we decided we liked and didn’t like certain subjects. The Science of Sport is trying to inspire 8 to 10 year old kids into science using sport.
Sports are a passion – American football, baseball, basketball. If you can get kids at a young age to think of abstract concepts using sport, you have a bigger opportunity to get them engaged.
You can use Pythagoras to show how you can be a better passer in football.
What we have done as an official charity in the US is develop programmes for professional sports tams – take the official curriculum that the students have to learn at that age group and put into context of sport but also the local team.
That is a win win for local schools, a win for the sports team that disseminates their brand and also does good for the community. It’s also a win for the university , building a larger pool for the STEM field.
Where can we find out a bit more about Science of Sport?
Website sciencesport.org which includes a series of resources for educators, videos and we are working on an app to gamify the curriculum.
I am working on a book on decision making – one of the important themes in making estimates credible, but also knowing where to look and quantify these important factors. This book is looking at a subset of decision making – counter intuitive ideas. For example Brookes Law i.e. adding resource to a project that’s late will make that project even later. Looking to be released next year!