Satisfice then Optimize- Making Decisions Under Time Pressure

The sport of soaring is a particularly interesting activity to analyze decision-making due to its implicit time sensitivity. No matter how good things may be going, we are completely dependent on upon the next updraft to keep us in the air. Quite often, we are 20 minutes from landing at any time. On a ridge, this can be as little as two or three minutes. Particularly critical situations may require almost instantaneous decisions to avert disaster. How exactly do we go about this process and what can we learn from others?

There are two models of decision-making which are relevant to this discussion, namely the “comparative evaluation” and “singular evaluation” approaches. The comparative evaluation approach is the way we normally are taught to consider decisions, by evaluating two or more options and considering their respective costs and benefits (Klein, 1998, p.20), followed by then choosing an action. Simply stated, think of it like choosing a meal at a restaurant. You look at the different options and then decide which is on the balance the tastiest, cheapest and/or most socially acceptable choice at the time. The comparative evaluation approach is an approximation of the general “rational choice model”, which stipulates, canvassing all options, surveying objectives, weighing costs-benefits of all options, evaluating and assimilating information to evaluate options and finally considering contingencies for a variety of consequences. (Klein, 1998, p.28) Time and time again we are taught to come up with as many options as possible prior to making a decision, even in time-critical situations. However, in the real-world and particularly in time-sensitive situations, this approach simply does not work. Instead, we generally follow an intuitive, unconscious approach that solves the problem at hand without going through this entire formal process.

An alternative approach is the singular evaluation strategy. At a decision-point, this strategy involves considering a given solution, running it through a mental simulation and finally determining if the solution solves the problem. If the mental simulation shows that the given solution would fail, then the decision-maker starts computing an alternative solution. Once a solution is found, then the decision-maker immediately implements it.  The mental simulations happen automatically, instinctively and quickly. An important point is that this is not just going with the first thing that pops into your head. Rather, it is going with the first solution that your cognitive process deems will succeed. It just so happens that these processes can often be done within several seconds, assuming that the person making the assessment is competent and familiar with the task at hand. This approach is markedly different than the comparative evaluation approach as it simply seeks to satisfice, rather than optimize.

Simply stated, the goal is for a “good-enough” solution that solves the problem, rather than necessarily the absolute best solution.

One of my own recent experiences with this sort of decision-making put into action was during a rather disconcerting near-miss episode I had last year. I was flying with a novice club member on a short ridge-mission, coming back from the ridge in my club’s two-place Grob Twin Astir. Prior to turning final, I noticed that the towplane was getting ready to tow a single-seat high-performance glider. The novice pilot was doing the landing and I asked him to make a radio call that we were turning final. Unfortunately, the call did not go out because unbeknownst to us, the radio had been inadvertently switched off during the turbulent ridge flight. Secondly, through miscommunication and inexplicable error, the ground-crew launched the glider right as we were coming over the threshold of the runway. All the underlying conditions were now set for a really close encounter of the worst kind.

Once we were several hundred yards past the threshold, I saw the towplane taking off in my peripheral vision and said, “My glider.” This is where things get interesting from a decision-making standpoint.

My first reaction was to close the airbrakes, and make sense of what was happening and what to do next. This seems to be a good course of action in most landing situations as the limited amount of energy you have at your disposal is quite critical. First, I decided to go closer to the parallel paved runway to our left to give a bit more separation between all of us. The principal problem was that the towplane was going to climb and cut directly across my flightpath. The reason is that launching on (the rarely used) runway 7 lines up the towplane with a hill at the end of the runway, requiring it to cut across the runway to make a safe departure. Then, I started going through the singular evaluation strategy to decide what to do. Secondly, while I knew that the glider was on tow, I did not know exactly where he was in relation to me as I was ahead of him.

I started going through mental simulations of different courses of action. This is hard to distinctly convey as I was imagining the trajectories of the towplane, my flightpath and the glider on tow. I was computing where the point of intersection would be and where I needed to be at that point in space to avoid hitting anyone. I distinctly remember feeling the pressure increase and looking really hard at the situation. My concentration became absolute and my stomach muscles tensed up. Time slowed down a little bit. I went through the first cycle of this assessment, which probably lasted something around two or three seconds, if even. Much to my consternation, I did not find a solution that resolved the collision trajectory. My concern was that I did not know where the glider on tow was. While I was behind the towplane, I was worried that if I landed short that I would get in his way. The pressure increased again and I again went into the cycle of mental simulations.

Finally, the satisficing solution was to push the convergence point as far down the runway as possible, where the towplane and glider would be off the ground. I realized that this point would be somewhere near the crossing taxiway midway down the runway. As long as I was low to the ground, it would be like flying underneath a powerline… there would be enough vertical separation between the aircraft to avoid a collision. Furthermore, this would guarantee that I would not be in the way of the glider as he would be above me by the time we converged. I pushed the nose down, flew on the 8 o’clock position relative to the towplane and finally got lower than him. At the main taxiway, the glider over-flew me by about 40 feet and now I concerned myself with getting my glider stopped before the end of the runway.

Some people criticized my judgment and execution as they pointed out that there was a better solution, to sidestep over to the main runway and stop as soon as possible. This would certainly have been a more optimal solution. I certainly spent a lot of time poring over my decision-making and the exact choices I made in doing what I did. However, when considering the singular evaluation approach, I feel that my judgment and execution was totally adequate. At the time, I was undergoing a novel situation with incomplete information which required rapid processing and assessment to determine a workable solution. Once I found it, the time-sensitivity of my situation simply did not allow a consideration of alternatives. I executed and it worked.

What can we learn about how we approach and teach decision-making between these two different approaches? While the cognitive psychologist Gary Klein argues that most experts follow an approximation of this approach to solving problems, even ones that are not imminently time-critical, I believe there is a spectrum of possibilities depending upon the time-sensitivity. On one end of the spectrum, there is simply no room for judgment and it requires an instantaneous reaction. Consider swerving a vehicle due to an unforeseen obstacle on the road. On the other end, we can have months to deliberate what sort of clubhouse we would like and how much we are willing to pay for it.

For critical situations in glider flying, I would suggest that the best course of decision making is to start with a satisficing approach. Only once a solution has been found and the pilot started to execute, to consider alternatives.

Consider field selection. Suppose we are gliding along at 3000ft near mountains like the Catskills, which require gliding out to landable terrain. At a critical juncture, we failed to connect with a thermal and must bail-out into the lower, landable ground. At this point, there are a number of fields to choose from. I argue that the best course of action is to scan them quickly and intuitively pick one. Often times, even with a quick glance we can assess slope, surface, obstacles and wind, all critical things to consider in choosing an appropriate field. Now that we have a plan and a way to execute, let’s reconsider alternatives. Are there better fields? What are other options in case the field in question turns out it is problematic? If you find a better option while you still have reasonable altitude, suppose at least 1500ft above the ground, then you certainly can change your plan. But I wager that most people end up going with their instantaneous assessment most of the time as it is remarkably good at processing all the relevant information and coming up with the best solution.

Consider thermal selection. Glance at the sky, imagine the courseline and make a gut assessment. Point the glider toward a cloud or thermal generator of some sort that is within 30 degrees of courseline. Once that you’re heading toward that cloud and you’re flying the right speed, reassess. Where are other gliders? What is the big picture? Are the conditions getting better or getting worse? Maybe there is a better choice? 

However, one should also consider some down-sides to this approach. The first one is that once a solution has been made, it becomes rather difficult to consider all the information in its broad context. For instance, once established on a path toward a cloud, it is not too difficult to reconsider if perhaps the other cloud is better. But it is quite difficult to realize that the best solution is to deviate 90 degrees and fly a completely different path after making the initial assessment. If the situation is not imminently critical, be careful not to box yourself in with your first workable solution. This requires discipline in recognizing that once the immediate decision has been made, all the variables should now be free to be analyzed further for a more optimal solution.

Secondly, this approach ONLY works if the person is competent at handling a given problem. If a pilot has little experience, his mental simulations will be much less effective and will compute fewer variables. With less experience, a pilot must either revert to a programmed response/heuristic that was taught in training or rationally work out a number of solutions. For instance, consider if a pilot finds himself stuck on top of a cloud deck. Even if the pilot has thousands of hours, he may never had experienced this situation before. As a result, it is much more likely that his immediate solutions may fail to account for all the variables than someone who may have an instrument rating. In this situation, the pilot is much better off to slow down his thinking, consider all of his options and weigh them accordingly prior to coming to a decision. An example of a programmed response or heuristic is, “Don’t turn once you leave the ridge to come back to the airport!” This lesson is given to all of our ridge pilots at Blairstown so that they don’t panic if they hit more sink than they expect on their final glide. The history, math and landing options all show that it is vastly better to continue the approach than attempt a crash landing in the lower reservoir. So instead of relying on an independent assessment on the part of the pilot, a decision is programmed in to help pilots avoid making known errors.

However, as a balance, I find that in most cases it is better to come up with a workable solution and start implementing it rather than coming with several possibilities and choosing the best one. As an axiom, first satisfice, then optimize. Only if time allows, reassess and consider other options. The more of an expert the pilot is, the more quickly and more likely the initial gut assessment will consistently be correct and not require an alternative course of action. However, let’s discard this notion that it is always essential to come up with several alternatives before making a decision. It can be inefficient, unsafe and unrealistic. Satisfice, then optimize.

Klein, G. (1998). Sources of Power. Cambridge, MA: MIT Press (20-28)