*Published July 11, 2026* --- There's a genre of advice built on one promise: learn a few frameworks (an album of your best ideas, Problem-Agitate-Solution, the Pyramid Principle) and you become, in [Dan Koe's phrase](https://www.youtube.com/watch?v=DKT6m_8vCkA), dangerously articulate. They're real, teachable, and they work. Two corrections before that hardens into gospel. First, articulateness isn't a blank slate that practice fills: twin and adoption studies put the heritability of normal verbal ability at [roughly a quarter to a half of the variance](https://www.cambridge.org/core/journals/language/article/abs/heritability-of-language-a-review-and-metaanalysis-of-twin-adoption-and-linkage-studies/B7A132E68D74B21F3E3E6F4A41192343). That caps no individual's ceiling, since heritability is a population statistic rather than a personal verdict, but it does retire the flattering promise that effort alone carries anyone to any level. Second, the practice explains less than advertised. [Macnamara and colleagues' 2014 meta-analysis](https://journals.sagepub.com/doi/abs/10.1177/0956797614535810) put deliberate practice at about 14 percent of performance variance on average and under 1 percent in professions, shrinking as domains get looser, and communication is about as loose as they come. Articulation is trainable from wherever you start, but the frameworks are one modest input among several. Those are quibbles with the mechanism. The real problem is the payoff. The frameworks carry a second promise: that clarity is a signal, a hard-to-fake proxy for clear thinking, since the bullshitter meanders while the one who understands states it plainly. There's something to this, and a bias on top of it. Psychologists call it the [fluency heuristic](https://journals.sagepub.com/doi/10.1177/1088868309341564): information processed easily gets judged truer and more competent, correct or not. So fluency was never a clean signal but a bias vector: it can make an empty claim land as expertise and a hesitant expert read as unsure. Then the ground moved. Large language models now [produce fluent, well-organized prose on demand](https://lcamtuf.substack.com/p/when-it-comes-to-writing-llms-have). The scarce thing that made polished writing informative, that producing it took thought, is gone. Fluent output was already weak because of the bias. Now anyone can generate it, so it's close to dead. So what is left. What's left is the part you can't prepare or hand off: reasoning defended in real time and unassisted, once the polished draft is taken away. Probe it and the fluency drops out. What remains is whether the thought holds, whether the person can take the next inference, handle the objection, say where their own idea breaks. A model can write the essay, and now hold up its side of a chat. A person in the room can have one in their ear, which is what the [live-interview copilots now sell](https://www.cnbc.com/2025/03/09/google-ai-interview-coder-cheat.html). What neither can fake, once you take the aids away, is you answering for your own understanding on the spot. Move the test there. Delivery still sways the room, but it's the reasoning that has to hold up. A second tradition, usually cast as the framework crowd's enemy, points at that same surviving layer. Jordan Peterson's advice is almost anti-technique: align your words with what you actually believe, and find what you genuinely think rather than what you're supposed to say. His reason is psychological, that words out of line with your thinking leave you incoherent and weak. The reason here is colder: output is now free, no longer where the signal is. Two arguments, one instruction: say what you think and follow it under pressure. The frameworks aren't a rival to that instruction; they're scaffolding for the thinking behind it. So keep the frameworks for what they do to your thinking, not the polish they add. Expose your reasoning to scrutiny early, from people known to be brutal about it; the hours you log matter less than whether someone shows you where it fails. And distrust your own fluent drafts, this essay included: the easier a paragraph was to produce, the less it proves you did the thinking. It reads clean, and by its own argument that's not evidence it's right. Push on the reasoning and see whether it holds when I'm not narrating. This isn't abstract for me. I've never been a natural performer. The easy, fluent delivery that reads as competence isn't how I'm wired, and for years that felt like the deficit to fix. The argument here is that the part that is mine, reasoning I can defend once the script is gone, is the part that still counts. None of this makes articulation less worth training. It makes the reason narrower and more honest. You're not learning to sound competent; that's a trick anyone can now get for free. You're learning to think in a way that holds up when someone stops reading and starts asking.