GM. This is The Great Unlock, where we take the machinery of school apart on the kitchen table and figure out which parts stopped earning their place a long time ago.

This week we go after the most sacred object in the English classroom: the essay. Every district grades it, scores it, files it, and calls the job done. And almost none of them can tell you what the essay was actually for. Spoiler... the essay was only ever the receipt. The real purchase was the thinking.

A new writing program just joined the HierLearning slate, and it gets this in its bones. So today is less about that program and more about the idea underneath it, the one most schools have been getting backwards for a hundred years without noticing.

Let’s unlock it.

Writing does not record thinking that already happened. Writing is the machine that makes the thinking happen. Grade the essay and miss the point... the essay is exhaust, and the engine is the child's mind learning to run.

Picture Maya, grade 9, the night before her essay on To Kill a Mockingbird is due. She opens a chatbot, types the prompt, and pastes back five clean paragraphs with a thesis, three body sections, and a tidy conclusion. She hands it in. It earns an A-minus. On paper, the assignment worked perfectly. The artifact is flawless.

The artifact is also the only thing that happened. Maya did not wrestle with Atticus. She did not change her mind halfway through paragraph two and have to go back and rebuild her argument. She did not discover, in the friction of writing the sentence, that she did not actually believe her own opening claim. The essay exists. The thinking never did.

This is not a hypothetical. In 2025 a team at MIT's Media Lab wired up 54 people, split them into three groups, and had them write essays over four months. One group wrote with ChatGPT, one with a search engine, one with nothing but their own brain. The EEG data was brutal. The AI group showed the weakest neural connectivity of the three, the thinnest engagement across the networks you use to actually reason. When researchers asked them to quote a line from the essay they had finished minutes earlier, many could not. Their own words felt like a stranger's. The researchers had a name for it. Cognitive debt.

The machine wrote the essay, so the machine did the thinking, so the student was left holding a document about ideas that never entered their head.

And this is the trap districts keep walking into. When the deliverable is the essay, an AI that writes the essay looks like a miracle. Output goes up, marks go up, everyone is busy and productive. But the entire value of the assignment lived in the struggle you just automated away. You did not make writing more efficient. You deleted it and kept the packaging.

Brian Connell spent 18 years teaching high school English at St. Joseph's Prep and Athens Academy before he built a writing program. His central conviction is that all writing is personal writing, that there is no such thing as a generic essay, only a specific mind trying to make itself understood to a specific audience. Strip out the person and you do not have neutral writing. You have nothing.

That conviction has a hard consequence for how you teach. If writing is thinking made visible, then a teacher reading a student's draft is not grading prose. She is watching a mind work, in real time, on the page. She can see exactly where the logic buckled, where the claim outran the evidence, where the student almost had it and flinched. Writing is the only school subject where you get to see the thinking directly instead of guessing at it from a multiple-choice bubble.

And that is precisely what makes it teachable. You cannot scaffold a thought you cannot see. The essay draft is the x-ray. The teaching happens in the conference afterward, when a human sits beside the student and asks the one question that makes the whole argument fall down and forces a rebuild. That question is the lesson. Not the rubric. Not the grade. The question.

You do not teach writing by correcting essays. You teach writing by interrogating the thinking that produced them, one student, one draft, one uncomfortable question at a time.

The master schedule betrays everyone at exactly this point. Teaching thinking is not efficient. It looks like conferencing, drafting, feedback, revision, and more conferencing. It is slow by design, because minds change slowly and under pressure. A single student learning to build a real argument might need six rounds of a teacher's attention. The kid beside her might need two. The kid behind them might need twelve.

But the schedule does not care. The bell rings on a fixed interval for all of them. Coverage is due Friday for all of them. In our system, time is the constant and learning is whatever happens to fit inside it. So the teacher, forced to choose, does the only thing the clock allows. She grades the artifact instead of coaching the thinking, because grading 150 essays is barely possible and conferencing with 150 minds is not. The essay became the endgame because the schedule made the real work impossible, and then we mistook the surrender for the subject.

This is The Great Unlock, and it has nothing to do with a chatbot that writes essays. It is a tool that gives the teacher her time back so she can do the human part that no machine should ever do. That is the entire design principle behind YAWP!, the writing program that just joined the HierLearning slate. It was built by teachers, out of Brian Connell's school of writing, and its AI tutor has one non-negotiable rule. It never writes for the student. It sits with the writer and asks guiding questions, the same move Connell spent two decades making by hand, so the thinking stays where it belongs. Then a grading assistant reads a draft against the teacher's own rubric and returns specific feedback in about thirty seconds, work that used to eat a teacher's entire weekend.

Read that sequence carefully, because the order is the argument. The AI does not replace the thinking. It refuses to. What it replaces is the clerical crush that was stealing the teacher's time... the grading pile, the feedback backlog, the triage. It hands those hours back to the teacher so she can spend them where only a human can, sitting beside Maya asking the question that makes her argument collapse and rebuild stronger. YAWP! reports 97 percent of students improving their writing and every teacher in the program satisfied. Believe the exact figures or don't. The mechanism is what matters, and the mechanism is sound. Scale the teacher's time, not the student's output.

Stop asking AI to do the child's thinking faster. Start asking it to do the teacher's paperwork faster, so the teacher can finally afford to do the thinking work with the child.

If you lead a district and you are evaluating any of this, four questions separate a real unlock from an expensive way to cheat. Does the tool ever generate submittable text for a student, yes or no? Does it measurably return time to your teachers, and can you count the hours? Does it keep the human conference at the centre of instruction, or does it try to replace the human entirely? Was it built by people who have actually taught writing to a room of thirty teenagers, or by people who have only ever shipped software? Get those four answers before a single purchase order goes out.

A structural problem hides in plain sight on every district data dashboard. We measure writing instruction almost entirely by output. Essays assigned, essays completed, rubric scores logged, proficiency bands filled in. Not one of those numbers tells you whether a single child got better at thinking. They tell you a document was produced and scored. We built an entire accountability apparatus around the artifact and none of it around the cognition the artifact was supposed to prove.

So when an AI tool arrives that inflates every one of those output metrics by writing the essays for the kids, the dashboard lights up green and leadership calls it a win. The measurement system cannot tell the difference between learning and its counterfeit, because it was not watching the learning to begin with. It was counting paper. Until districts start measuring the thinking, growth in argument quality, movement in a student's reasoning across drafts, they will keep rewarding the tools that hollow out the very skill they claim to build.

This week's flagged phrase: "AI writing assistant."

It is the most comfortable phrase in the EdTech catalogue and the most slippery. "Assistant" implies a helper standing beside the writer, and everyone nods along. But the word is doing enormous work to hide a binary. An assistant that hands you finished sentences does not help you write. It writes, and lets you paste. An assistant that refuses to hand you a single usable sentence and instead keeps asking what you actually mean is doing something completely different, and it deserves a different name. The catalogue uses one word for both because the vague version sells to everybody. Your job is to make the vendor pick a side.

Years ago, back when I still had my own classroom, I sat beside a boy named Daniel who had written a technically perfect paragraph about why his town should build a new library. Thesis, evidence, conclusion, all in place. Full marks by the rubric. I asked him one question. "Do you actually believe this, or did you write what you thought I wanted?" He went quiet, then admitted he did not care about libraries at all, he cared about the hockey rink that kept getting cut from the budget. So we tore the paragraph up and started again on the rink. The second version was messier, angrier, half the grammar of the first. It was also the first thing he had ever written that was genuinely his.

That is the whole job, compressed into one afternoon. The perfect paragraph was the artifact. The torn-up second draft about the hockey rink was the thinking. No rubric on earth would have scored the second one higher. Every teacher who has ever done this work knows the second one is the only one that mattered. It takes time we are almost never given, which is exactly why the tools that give that time back are the ones worth fighting for. That is the yawp, by the way... Whitman's barbaric yawp sounded over the roofs of the world. The neat paragraph was only ever a byproduct. The real point is the sound of a real person finally saying the true thing.

The essay was only ever the artifact. The thinking is the assignment, and thinking takes a kind of time our schools were built to deny. The unlock has nothing to do with a machine that thinks for our kids. It is a machine that finally gives the teacher enough room to teach them how.

 ________________________________________________________

That's all the unlock for today. Tune in next week.

Stay awesome, you unlockers!

Want more? Order The Great Unlock here.