Las Vegas 2023
The MIT Sailing System
The MIT Sailing System
AT
Adam Traina
Founder, TrainaThought LLC
Transcript
<silence>
The last speaker of the day, cont, uh, continues the nautical theme, uh, we've had today set by Sean North Admiral, uh, Richardson and Captain Bassett. I think one of the toughest, uh, challenges we have as technologists is around the need for ification. Um, you know, uh, in our world we constantly have pressure to spend all the time on shipping new features at the expense of other important things like reliability, testability, security, uh, and so forth. And so a third of wearing the, uh, of the book, uh, is around the need for ification. Uh, and of the 20 plus case studies, one of the most startling and fun to me was a story about the MIT Sloan Sailing team. This is where Adam Trina leads a team of eight fellow sailors to compete against other business schools, most of whom have no sailing experience. So after taking Dr Severe's course, uh, Adam decides to adopt a management system based on those learnings, adopt a management system based on learning. Whenever they run to problems, they pause just like the Andon cord in the Toyota production system, and they do that even in the middle of a race. So he will share the astounding story of, uh, how this led to years of victories even after he graduated. Here's Adam.
Come away. Come away from away. Away. Alright, come away.
Thanks Gene, and thanks Anne and the it revolution team for bringing us all together today. My name's Adam Trina. I'm the CEO and founder of Train of Thought, which is this company that seeks to use AI to develop high performance systems. The first system I worked out was the MIT sailing system. While I was a graduate student there studying ocean engineering and business, I took Professor Steve Spear's course on the high velocity edge, one of the main tenets of which is developing a management system that can outlearn its competition. Through rapid experimentation and learning from real world feedback, he encouraged us to try that system out and I applied it to one of my favorite areas of the life, which is sailing.
For those unfamiliar sailing or, or going around or a sailing race involves starting between the black boat and the black mark, heading up wind to the market, the top lowering your up wind sails and exchanging them for a downwind sail. Then heading to the turning mark on the left where you change your angle to the wind come downwind. Exchange your downwind sails to your upwind sails, and then finish back where you started. It tests your strategic thinking and how to get around the course fast. It tests your tactical thinking and how you compete with the other boats to test your test, your technical skills in getting the boat to perform against all the angles to wind and test your managerial skills and pulling your team together to perform all those maneuvers when they're needed. Sounding is a little bit different from other sports because the course can change while you're on it. If you're at second base in baseball, no one in the world would accept, would expect home plate to suddenly move to third. In a sailing race, that can very well happen. The wind can shift, the marks can drift, and the race committees can just change their mind on where you're supposed to finish.
That's the agile environment that we were entering to compete with business schools from all, all around the world. In the Rolex International Regatta, the race is at the beginning of the school year, there's very little time to practice. The second year, students had received an invitation to the event and invited me along to drive the boat for them. We got on the plane and met each other in Italy for the first time. We, when we introduced ourselves to the other teams, we found that most of 'em were pretty well organized. There were Olympic sailors, collegiate, all Americans, and circum in a heart to heart with a captain from Dartmouth. He leaned over and confessed. We're pretty worried about this race we've never sailed together before, to which we responded. We're also worried about this race. Many of our teammates have never sailed at all. This is the breakdown of the team skills in that first day. Now, when you break it down, sailing is all about pulling on ropes. The trick is pulling on the right rope at the right time without an opportunity to practice. We decided to figure that out on the way to the race course.
We worked on pulling on the ropes and doing basic turns, and after each maneuver we would sit around and give each other feedback on how to do things better and find opportunities for improvement. We had four in four practice turns. We took our maneuvering time from two minutes down to 20 seconds. One of the biggest hard, one of the hardest things to figure out was how far to pull on the rope. Sailing is pretty complex. The governing equations are still unsolved, and there's a million dollar prize for anyone who come up with an answer. You can use a supercomputer to get pretty close, but those aren't allowed on the boat. It's a real no-code environment.
<laugh>.
Instead, people put visual indicators all over the sails. There's tons of things you can look at to figure out if you're doing it right. The problem is when you're pulling on the ropes to adjust them, you can't see any of those things. So what we figured out to do was to bring the indicator right into the hands of the operator and we got the sail where we wanted them and put tape marks all over the lines. So instead of telling someone, Hey, pull on that hoard until the jib head is block to block with the shiv, which nobody understood, we could say something like, pull the blue. It's pretty easy to follow, and the blue mark would come right into their hands and no one to stop. We just got to the part where we were figuring out how to distribute responsibility among the team. When the radio crackled, the race officials gave us very important information like where to start, where to go, and what to expect. The voice was beautiful, musical strong, and speaking entirely in Italian
<laugh>.
We looked around the race course to figure out where to go, and this is what the start looked like through our binoculars <laugh>. We were so far away from the rest of the fleet that we got to the line seven minutes late. When we crossed the starting line, I can't help but think to myself, boy, great start skipper
<laugh>.
But what I actually said to the team was, well, let's use this race to learn how to sail. So in the first leg, we practiced our upwind turns, we ran around the mark and put the downwind sail up for the first time and congratulated ourselves for making it around the course. We started in last and finished there too. <laugh>, the second race went better. We were on the line at the start. We were actually in third. By the time we got to the first mark, or we had gotten pretty good at that upwind sailing. We had a couple rounds of practice. Now when you go and turn to the downwind, this is what you expect to see. You haul up that big, beautiful downwind sail, it fills with wind and spirits you away off to victory. This is what we actually saw
<laugh>.
We spent most of the downwind struggling to straighten things out. In fact, the the team from Dartmouth sailed by us half a lap ahead and shouted some helpful advice like give up. Now.
We continued to iterate, learn from our mistakes, learn how to redistribute roster possibilities so no one had to do two things at once. We added closed loop feedback by checking the speed of the boat so we could tell if what we were doing was better or worse, and eventually we were able to anticipate what's happening so well that we didn't have to move around the boat anymore, which made it safer to be there because you didn't have to cross in front of dangerous equipment. We managed to win the final race, but it wasn't enough to overcome our deficit. In the beginning, we learned two major things from the first attempt at this. First is that technical skill was not the most important thing. With sufficiently smart team, with rapid iteration, we could learn our way to victory. The second thing we learned is make sure you have somebody who speaks Italian
<laugh>.
In the second year, we reframed our approach. Instead of focusing on technical skills, we focused on social skills. Our mission was to return home with more friends than we left with. We sort of set the baseline and the tone in our DNA for how we were gonna approach this year. We raised the team BA and selected for basic qualities, people who are smart, humble, collaborative, and wanted to win. Being humble made, it made you more willing to acknowledge mistakes and raise your issues. Being collaborative and smart gave you the chance to learn from those mistakes and build a reinforcing system. When you put all this together, we were hopeful that not never making the same mistake twice would lead to high performance. We also challenged our communication system. Most of the other boats out there had a traditional command and control track where the person at back would give instructions, everyone in the front and they give, give feedback back to the person in the back. We tried this in our first day of practice. So we're sitting in the back of the boat with your half your team, 30 feet away with the wind whipping by, and you yell up ahead, haul on the hoard,
And your ship mate is standing there at the front of the boat holding the hired. Say, let's try it again. Let's try rebooting. Haul on the hired. Say, all right, let's stop this. Let's going on. Huddle up. Do a little a after action review. Ask your shipmate, Hey, how come you didn't haul? And he said, you said hold. I said, no. I definitely said Hall. He looks back and say, well, all you Americans sound the same.
<laugh>.
The first languages on the boat were Portuguese, Spanish, Bulgarian, English, and yeah, we had found an Italian. We changed our language because it was more important for the team to be included and have a common understanding than it was to use the right words. We used pull and stop. After a little bit of deliberation, we also changed the flow of communication. We divided the boat into, into two cells with a bow team and a stern team. This allowed everyone to take ownership over their part of the ship. We also moved our cues away from people and into environments so that everyone was looking at the same kickoff points, taking out the latency of waiting for someone to tell them what to do. Our cues became wind shifts and turning marks in other boats instead of someone yelling at you. We brought our system to Italy where the same sort of thing happened. We got there, we introduced the other ship, introduced ourselves to the other boats. We find out the British Columbia had just gotten there. After finishing a circumnavigation, their captain had brought an augmented reality system that let him see the indicators wherever he looked.
We had brought a roll of tape and we're using it to bind our team together. Our improvement system had gotten so ferocious that we were fixing problems during the race. All the time. Here, one of our shipmates is securing some flapping canvas and trying to do so in a way that'll catch more wind and provide more power for us. One of the improvements we tried went a little overboard. We tried pre-deployment, our downwind sail. While the upwind stale was still set, the downwind sail set itself and got tangled in the upwind gear, and the load became too heavy for our team to continue operating the boat while fixing the problem. We had a difficult choice to make and went back to one of our founding principles, which is a sailor's four favorite four letter F word fast. We decided to stop the boat and fix the problem as soon as it emerged. Instead of continuing to sail with it and trying to wrestle it with all the way downwind. That's our competition right on the hip. And as soon as we stopped the boat, they passed us by, we able to reset the sails, get back on the race course and finish strong. Much better than if we had simply gone along with it.
This is our performance over the years, showing a rapid rate of improvement. In the second year, we won almost every race and continue to be further and further ahead. As the, as the, as the races went on in the after action events, our, the other teams are looking at us and wondering why we're so far ahead, because we couldn't really see what we were doing. They thought it was because we were in the blue boat, unable to recognize the internal dynamics of the system. We finished ahead that year and then thought about how we could re was the system re repeatable. We had managed to do it with a different crew, but could we do it with a different captain or with a different boat? We transported the system to new boats. That's a 1973 CNC. And on the right, uh, on the other side is a 1902 wooden harh translated system to new boats. We ca I graduated and handed off the system to new skippers and we trained new crew. We won the Rolex or Gata three years in a row, and our victories at home continued to mount. This is a bar chart of our performance over the years, and we stopped tracking in 2017 because we didn't have enough room on the table for all the extra trophies.
We continue to win to this day and continue to improve on that high velocity system. Many of those principles are translatable to our current work. We divide and conquer. We use simulation to explore the DI design space rapidly and get feedback from KPIs. And we, instead of using verbal cues, we use algorithms to coordinate hundreds of autonomous systems. Gene asked me to prepare a slide on help needed, and my ask is, is this, if you're interested in testing your hypotheses and exploring the world of design and operations, manage it with ai, please reach out to me. I'm available on Slack and you can catch me through email as well.