Chaos, Compliance and the Infallible Conditional of Education

Consider this formulation, proposed in a variety of forms, especially by reformsters of the past few decades. I give you, the Infallible Conditional of Education:

If every teacher does X, then all students will learn Y.

It is wrong, every single time. Doesn't matter what you plug into X and Y, it's always wrong. Doesn't matter if you're applying scientific research. Doesn't matter if you're scaling up a program that has worked for small groups of students with a particular teacher. 

It is wrong, every time. 

There are predictable results for trying to implement a program requiring all teachers to do X.

Some students will, in fact, learn Y. Some students will not, but since the program is premised on the proposition that teachers doing X must infallibly lead to students learning Y, people in charge will declare that either A) the teacher (either willfully or out of incompetence) is not implementing X with fidelity or B) the student is defective in some way. Thus the Infallible Conditional of Education leads to placing a high value on compliance-- if everyone would just behave the way they are supposed to, declare frustrated leaders, this would totally work!

There are lots of reasons to fall for the fallacious ICE. Someone looks at research and says "This is how masters of this skill do it, so this is how learners must learn it" or maybe "This worked for a group of ten college sophomores, so it should work for a hundred high school freshmen." Sometimes the leaps are huge-- like the leap from "tutoring of small group of students by human tutors produced theoretical and highly dubious results, so clearly giving many students a computer program to sort-of-tutor them will yield great results."

But mostly I blame the continued attractiveness of ICE on the general public's failure to move out of the Newtonian world and into the modern world of chaos theory and quantum mechanics.

(Note: I am now going to grossly over-simplify complicated stuff. Don't @ me over details.)

The Newtonian world loves nice, linear equations. X + Y = Z. Push this lever with this amount of force and the load lifts that amount of weight. Actions have consequences, and given enough data, we can always predict exactly what will happen. Complex systems may require more data, but the idea is the same. We could, for instance (the Newtonian imagines), collect every necessary piece of data about an individual human being and that would allow us to predict exactly what job that particular human is perfectly suited for. Or, as now-failed Knewton once promised, if we had enough data about a student, we could tell her what to eat for breakfast on the day of a test. 

That's the linear premise-- if we could collect information about all the inputs, then we could predict all outputs.

But everything we've learned in the post-Newtonian world of chaos theory says that is not true for any complex system.

Complex systems are subject to a million tiny and unpredictable variables. The popular notion of chaos is the Butterfly Effect-- a butterfly flaps its wings in China and then it rains in California. Minute differences make huge differences further down the line. One of the granddaddies of chaos theory is Edward Lorenz who was actually a meteorologist. His big aha moment came when he needed to rerun a computer model weather simulation, but to save time he just took the numbers from a midpoint and started from there. He entered 0.506 as a condition instead of the full 0.506127, and the result was completely different. A minute effect resulted in a massive difference. 

Tiny differences have huge effects in complex systems, and virtually all systems built around human behavior are complex. You can have godlike knowledge of all the inputs and still not be able to predict exactly what the outputs will be. Likewise, you cannot design the inputs to create the exact outputs that you want. 

Does this mean complex systems are obviously random? Of course not. With chaos theory we get the idea of strange attractors. The results of a complex system will not graph perfectly onto a single line, but they may cluster around some central shape or area. There will be some wild outliers, but results within certain ranges will be more likely than others. One of the Lorenz strange attractor? It looks a lot like a fuzzy butterfly. (If you want to read more, I always recommend James Gleick's Chaos.)

I often think of Russell Barkley and a five minute video of him offering advice to parents. "You are not engineers," he points out. Further:

So, what we have learned in the last twenty years of research in neuroimaging, behavior genetics, developmental psychology, neuropsychology, can be boiled down to this phrase:

Your child is born with more than 400 psychological traits that will emerge as they mature, and they have nothing to do with you.

So the idea that you are going to engineer personalities and IQs and academic achievement skills and all these other things just isn't true.

Your child is not a blank slate on which you get to write.

...The better view is that your child is a genetic mosaic of your extended family. Which means this is a unique combination of the traits that run in your family line.

Be a shepherd, Barkley suggests. "Let them grow, let them prosper, please design appropriate environments around them, but you don't get to design them."

We can't design or engineer children, and the notion that we could, given enough data or enough science-flavored policies or enough making every teacher do X, is just wrong. We can absolutely implement pedagogy that improves the odds and increases the likelihood of outcomes we desire, but we cannot control the myriad of tiny details that have profound effects on this (or any other) complex system. 

We cannot treat students and teachers like machines, engineering outputs by controlling inputs. What we have learned (or should have learned) in the last half-century or so tells us that the world does not work that way. You strive in the direction of certain goals, and as a living human being you make a million judgements and adjustments along the way, never knowing for certain which will turn out to have massive effect somewhere down the road. 

It is human to tell ourselves (and others) that if one does X just as one is Supposed To, one will without fail arrive at Y (unless some evil party thwarts you). We like to make up rules and insist that everyone comply because then the world would work more smoothly.

It works just often enough to seduce us into believing "If X, Then Y" is a true and dependable rule of the universe, and it makes us feel very safe and secure to think it so. It is not. Nature, the world, our fellow humans-- none of them comply with our desires just because we insist we have discovered a formula that must result in our desired outputs. 

And in education, where we are working with a sample of 4 million teachers and basically every single young human, the formula is certain to fail (followed, too often, by all the wrong reactions to that failure). That does not mean we jettison every possible X or ignore any evidence we can collect about X's likely outcomes. But the Infallible Conditional of Education is a lie. 

Teachers should collect the best tools, the best Xs, that they can, and lean on those most likely to produce the most desired results. But they should also be prepared to adapt, improvise, read the room, change course, and just generally try to be the best butterfly shepherds they can be.