Nobody Has a Reason to Join the Dots But You
Strategy Sandbox. Published 6 September 2026. Status: Original concept, first publication.
At the Kalicube® Summit this year I asked Teodora Petkova where feelings come into AI-assisted optimisation, expecting an answer, and she turned the question round and asked me where AI-assisted optimisation comes into a world of feeling, emotion and being human, which is the better question and has been sitting with me ever since. She’s written the conversation up, and her diagnosis is one I’d sign without argument: the word “content” has been hollowed out until we bolt adjectives onto it, valuable, helpful, people-first, non-commodity, to recover qualities that should never have needed recovering. Her prescription is where we part company. She wants you to reconnect with the human impulse first and let the optimisation follow, and that quietly hands the most important job you have to somebody who has no reason at all to do it for you.
Google’s Knowledge Graph records where people already walked, and it already holds a thin version of you
Teodora’s best image is a park. You design the paths, you place the benches, you decide how the space is meant to work, and then people walk in and cut their own routes across the grass, sit on things in ways nobody anticipated, and turn the design into something living. She uses it to hold two things together: the deterministic structures we build, and the human ecosystem that grows around them.
Run the image the other way and it gets sharper. Google’s Knowledge Graph isn’t a park somebody designed and then opened to the public: it’s a record of where people had already walked, hardened into paths. That’s the Mirror Principle, and it’s what I was really describing when I started naming Brand SERPs in 2012, that what an engine shows about you is its opinion of the world’s opinion of you. The desire lines came first, the benches went in afterwards, and the structure Teodora treats as the deliberate half of the assemblage is yesterday’s mess with a surface poured over it.
Which changes what you’re doing when you sit down to write. There’s already a record of you, assembled out of everything anyone else happened to say, and it is thinner and duller than you are.
Engines join the dots whether you help them or not, and they join them low
Here’s the part people get wrong about machines. An Assistive Engine doesn’t leave gaps open, it fills them, constantly, because filling them is what it’s for. Ask it about you and it takes whatever fragments it holds, a job title here, a conference listing there, a company name, and it reasons its way to the most likely explanation of who that adds up to.
That reasoning is abduction, the leap from scattered facts to the best available account of them, and it’s the single most consequential thing any engine does about your brand. Do it well and you’re the person who defined a discipline. Do it on a thin record and the likeliest explanation of a few conference talks and a modest bio is a competent minor practitioner, so that’s what comes out, hedged, cautious and technically accurate. That hedge is the Doubt Tax, and it isn’t the engine failing to think about you. It’s the engine thinking about you and arriving somewhere ordinary, because ordinary was the safest reading of what it found.
The industry’s been circling this for thirty years without quite landing on it. Content was king when the problem was having anything worth indexing, context took the crown when the problem was making a machine understand what the content was about, and confidence is king now, because the machine understands you perfectly well and still won’t commit. Confidence decides how far the leap goes, it compounds gate by gate so that one weak link drags down everything after it, and it’s the only currency in which an engine can be said to believe anything about you at all.
Nobody else has a reason to do your abductive reasoning
The engine has no incentive to make the generous leap on your behalf, and this is the bit I only worked out properly this week: neither does anybody else. A person who reads your bio doesn’t stop and construct the argument that your work in 2015 plus your work in 2019 adds up to something nobody else in the field has built. They take an impression and move on with their day, because they have their own lives to reason about. Your competitors certainly won’t do it, and your clients haven’t got time.
For me that settles the argument about modesty. The advice to be humble about yourself assumes a reader who can infer what you politely left out, and it works beautifully on people who already know your field. An engine infers nothing kind. It reads a smaller claim as a smaller fact, repeats the smaller fact, and the modesty you intended as good manners arrives at the other end as a ceiling. If you don’t make the leap, nobody has any reason to make it for you, and the gap gets filled anyway by whichever dull reading was cheapest.
So make it. State what the pieces add up to, in public, in the terms you want repeating.
Framing supplies the interpretation an engine cannot reach on its own
I’ve argued this formally, in The Framing Gap, and the structure there is three steps rather than one. Scattered Proof is the ordinary state of a career, evidence lying around in a dozen places with nothing connecting it. Connected Proof is what you get when the claims and their corroboration are explicitly linked, which most good personal branding manages. Framed Proof is the step almost nobody takes, where you supply the logical interpretation yourself, so the machine doesn’t have to guess what the evidence means.
The engines can retrieve your evidence and they can generate an inference from it. What they can’t do is establish the frame that says which inference is the right one, and that gap is yours to close or yours to lose.
Corroboration separates the prophecy from the fraud
There’s an obvious objection here, and it’s a fair one: this is exactly what a chancer does. State the big claim, get it repeated, let the repetition harden into fact.
Somebody has already run that experiment, which saves us arguing about it. Authoritas took eleven fabricated experts, put sixty-six questions to nine AI models, and found the fakes could occasionally be recognised and never recommended ahead of the genuine authorities, because they lived almost entirely in the shallow layer of syndicated PR with no entity home and no corroboration trail behind them. “Recognition is cheap,” as the study puts it, and “recommendation is expensive.”
That tells you precisely where the line sits, because your own repetition is volume, and volume buys recognition. Repetition by independent people who could have said otherwise is proof, and proof buys recommendation. The claim you make today and the fraud look identical the day they’re published, and they separate later, when the engine goes looking for who else says so and finds either a field of corroboration or a field of your own press releases.
So the frame you publish is a promise with a due date. Say what the evidence adds up to, then go and do the work that makes independent people say it back, and the projection stops being a projection somewhere along the way.
Say the interesting thing, then go and be it
Teodora ends her essay with questions, which suits an exploration, and I’ll end mine with an instruction, because I think the practice needs one. Work out what your evidence actually adds up to, state it in public before anyone else has reason to, point at the proof that already exists, and then spend the next few years earning the rest.
Nobody is coming to join the dots for you.
This piece responds to Teodora Petkova’s “Non-commodity content, dialogic communication and goose-bumps on a boat, not counting the knowledge graph”, written after our Kalicube® Summit 2026 session, “Teodora Petkova on Meaning-Making, Non-Commodity Content, and the Living Knowledge Ecosystem”, recorded on 24 June 2026. The park is hers. The reversal is mine.
