April 5, 2026
On optimization
Friends, thank you for subscribing to the Coaching Letter. You are amazing. The response to the last Coaching Letter, #229, on bottlenecks and opportunity to learn, was a bit overwhelming, so it will take me some time to think through all of that—I already have a list of 5 coaching letters to write as an extension of #229 and your responses to it—but in particular I want to call out Catherine Saldutti, who, unbeknownst to me, wrote an article about equity and time, task and teaching several years ago and was good enough to share it with me.
So this Coaching Letter is about one idea arising from #229—the concept of optimization, a term I used in #229 in terms of optimizing opportunity to learn (OTL).
But first, a bit of advertising. As many of you know, we have been very invested in Building Thinking Classrooms (BTC) as an embodiment—almost an archetype—of the kind of instruction that we think all kids should experience—one that provides students with grade level work in the form of a challenging task that starts with a low floor but ramps up quickly, asks students to think and to keep thinking, responds to that thinking in real time in such a way that they keep figuring out the challenge, and in doing all this creates a holding environment that engenders the belief that all students are powerful problem solvers and thinkers and therefore belong in an academic space.
This investment extends to hosting the 2026 Building Thinking Classrooms Conference (BTCC26) in New Haven, CT on June 29–30, 2026, plus pre-conference sessions on June 28 for leaders and coaches supporting BTC, and for educators new to BTC. Conference sessions include everything BTC-related, including implementation strategies, task design, and assessment practices. The BTC team will be there in force and Peter will be offering multiple sessions; my friend, colleague and co-author Rydell will keynote (I just watched him in San Diego which, as always, was a little bit of magic); and I will be there, although I do understand that—at least in this case—I am not the big draw. The main conference is $585, the pre-conference is an additional $275, both include breakfast and lunch. Visit btcc2026.sched.com to learn more and register.
So, back to the idea of optimization. When I was first learning about systems thinking and lean production in the 1990s, there was a PowerPoint slide that appeared all the time. It was a cartoon of a copy shop with a sign in the window that said Speed | Quality | Price. Pick Two. (I thought it would be easy to get hold of said slide on the Google, but the best I could find is at the top of this post.) I don’t actually remember the point that was being made with this idea, but I know that it wasn’t the word optimization. That was not a term in common currency. But that’s really what the slide was saying. What the sign means, in parlance that I now understand, is that we are always working under constraints and cannot maximize all desirable outcomes simultaneously. If you optimize for speed and quality, the price will go up. If you optimize for speed and low price, quality will usually suffer. Every system has an objective function, whether explicit or hidden. When we use the phrase “every system is perfectly designed to get the results it gets”, that’s what we’re saying. And also, therefore, there are trade-offs. Optimization, then, is deciding what variable to maximize, and to organize everything else in service of that variable.
I feel like the phrase is newly ubiquitous (optimize for sleep, optimize for building muscle, optimize for productivity) with endless advice on social media catering to whatever your area of interest is. The reason I’ve noticed is, as I’ve written about multiple times, my fitness goals have shifted several times in the last 10 or 15 years, and now my Facebook feed is full of advice on creatine, mobility, and lifting heavy weights. And I’ve noticed that the term “optimization” used to be fairly niche, but is suddenly all over the place.
So I asked Claude to do some sleuthing for me, and indeed, “optimize for” was actually a term used by engineers, economists and mathematicians in the 1940s and 1950s. Practical applications abound. For example, during the Second World War in the Atlantic, U-boats were responsible for the destruction of millions of tons of freight between Europe and the USA, so destroying them was a priority. Initially, air patrols intended to find the subs followed a fairly evenly distributed search pattern. Mathematicians brought in to help with this problem (not all of them were code-breakers) created search patterns that covered less territory but concentrated on predicting where the submarines were likely to be. As a result, the number of German submarines found and destroyed increased. (If you want to geek out on this, the US Naval Institute is a good place to start.)
In education, as in every other field, we are choosing variables to maximize all the time, and they are not always the right ones, because they are frequently not actually chosen—they are unintended consequences of other decisions. For example, many schools have requirements that teachers record a certain number of grades per student for a given unit of time. This means that the system has now optimized for gradable assignments, which fundamentally changes the kind of tasks that students are asked to undertake, because now the task has to have certain characteristics—it has to have a set end point (“completion”); it has to be more or less objectively assessable (there’s a right answer, or there’s a rubric or its friendlier cousin, success criteria); it has to enable completion within a certain time frame; grading it has to be manageable for the teacher; and so on. The system is frequently not optimized for thinking tasks.
I am fairly certain that when schools created such rules, they were responding to parents’ requests to know how their kids were doing—because this is all downstream of introducing apps that allow teachers access to kids’ grades in closer-to-real-time than the old-school end-of-semester report card. And let’s not forget that this causal chain started with us wanting parents to be more involved in their children’s education—and I’m not saying that’s a bad thing, although I confess I sometimes think it—but I don’t think anyone thought that this would lead to—frequently but not always—students being asked to think less.
And because I can’t resist it, there’s another useful term here: the concept of externalities. An unintended consequence is something that happened that we did not initially intend to happen. An externality is something that happens to other people who were not part of the original decision. There are positive externalities (I wasn’t part of the conversation about whether my wife got a promotion, but I benefit from the extra money coming in) and negative externalities (I wasn’t part of the conversation about whether my wife got a promotion, but I suffer from the fact that she has to travel more). And the externality here is that the decision to require more grades in the gradebook was made without students at the table, but the consequences of that decision land most heavily on them.
So, how the concept of optimization connects to Coaching Letter #229 and bottlenecks in education: Here’s another way of framing the argument in CL #229. Frequently, the system sees low performance and concludes that the problem is the student (he has an IEP so he’s going to need more explicit instruction, she’s reading on a 4th grade level so she won’t be able to do 7th grade work). Once that conclusion is reached, the system optimizes around that assumption. It gives the student easier work, less time with challenging ideas, more remediation, more scaffolding, more telling, and fewer opportunities to think. The system, in other words, is assuming that the bottleneck is the student’s ability, and therefore it adjusts downwards, both giving and requiring less. (Thereby creating the outcome it is seeking to avoid).
There’s actually yet another way of looking at it that gets us to the same place, which is related to the grading example earlier. The system is optimizing for immediate student success, defined narrowly as a grade on the upcoming assignment, or perhaps by parents who don’t want their students to “struggle”. And therefore, it gives the student easier work, less time with challenging ideas, more remediation, more scaffolding, more telling, and fewer opportunities to think (thereby creating the outcome it is seeking to avoid). But if the system were optimizing for long-term success (which it says it is doing, in the form of every Vision of a Graduate I’ve ever seen), then it would worry less about completing an assignment and more about provide thinking tasks; less about what students can do independently today and more about what they are becoming capable of doing over time; it would optimize less for avoiding frustration and more for building competence, confidence, and intellectual autonomy; it would see struggle not as evidence that something has gone wrong, but as evidence that students are being asked to do the very kind of thinking that leads to growth.
By the way, the short-term/long-term distinction is fully embraced by wellness influencers. If you are optimizing for short-term happiness, you watch TV when you want, you eat what you like, you have the occasional drink. If you are optimizing for long-term wellness (AKA longevity), you give up alcohol, and you prioritize exercise, nutrition, and sleep. We are always working under constraints and cannot maximize all desirable outcomes simultaneously. This example also makes the point that there is not a necessary trade-off between short-term and long-term success that the copy shop example implies. Systems often create false trade-offs because they optimize for the wrong variable. The problem in schools is not that we must choose between student success in the short term and becoming successful in the long run; the problem is that we have defined success too narrowly (finishing the assignment) and too soon (by the end of the lesson).
Couple of other connections. Dylan Wiliam wrote a very useful article that I’ve referred to before and that everyone should read called The Secret of Effective Feedback, which is, of course, about feedback. But in it, he makes the very useful point that we tend to give feedback that is intended to improve a given piece of work, whether that’s a student essay or a teacher’s lesson. And that seems really obvious. But rather than optimizing for the quality of the work, we should be optimizing for the quality of the worker—because the work has already happened, so there is limited utility in improving it, but the worker is going to produce more work in the future, so there is great utility in improving their skills. Dylan does not use the word optimize, but that’s exactly what he’s talking about; we are always working under constraints and cannot maximize all desirable outcomes simultaneously, therefore the teacher has to make choices about what feedback to give, and that choice should be based on improving the skill of the student rather than the quality of the work the student just produced. In my experience, this way of thinking fundamentally alters the way that people think about feedback.
Second, what BTC optimizes is the main reason we have been so drawn to it. Building Thinking Classrooms is, among other things, a different answer to the question “what are we optimizing for?” Traditional classrooms often optimize for completion, compliance, correct answers, coverage—for perfectly understandable reasons. Teachers are doing what the system requires. BTC optimizes for something else: students becoming increasingly capable thinkers and problem solvers over time. Which means that it is willing to tolerate a little more uncertainty, a little more messiness, and a little more productive struggle in the short term in service of something much larger in the long term. It worries less about whether students can get to a certain point on any particular day, and more about whether they are becoming the kinds of students who can eventually tackle difficult problems independently. This frequently introduces a significant tension. Most of us say that we want students who are resilient, thoughtful, creative, collaborative, and able to solve novel problems. But those qualities are not built by optimizing for immediate success. They are built by creating the conditions in which students have to think.
Which brings us back full circle, I think—you should really come to the BTC conference at the end of June. And in the meantime, if there is anything else I can do for you, please let me know. Best, Isobel
Isobel Stevenson, PhD PCC
Author of The Coaching Letter
