Four in
Twelve

Robert McLaughlin built carriages in Oshawa from 1876. His son Sam sold the company to General Motors in 1918, and the town spent the next century assembling vehicles that somebody else had designed somewhere else. Families here absorbed a story so settled that nobody needed to say it aloud: another person draws the plan, and you build it well.

General Motors announced the end of Oshawa assembly in November 2018. The final vehicle left the line in December 2019. Two years later the plant reopened after a retooling investment of C$1.3 billion, and on 10 November 2021 the first Canadian-made Chevrolet Silverado came off the restarted line.¹⁴ Jobs returned to the floor. The story underneath the jobs never moved.

Malcolm Gladwell has a name for that kind of story. He calls it the overstory.

Gladwell took the word from forestry. An overstory is the upper canopy of a forest, the layer of leaves so high that nothing on the ground ever looks up at it, and yet its density decides how much light reaches every seedling below. In Revenge of the Tipping Point he argues that communities grow canopies of exactly that kind: narratives so settled that people treat them as weather rather than as choices.¹ Three levers decide whether an idea spreads under such a canopy. Gladwell names them the overstory, the superspreaders, and group proportions.

Marketers spent twenty-five years mining his first book for tactics. I want to use his second one as a launch plan, in the open, with the numbers visible.

QwaiAI launches this year. QwaiAI sits at the centre of the Qualped platform as an author-mentor, life-designer and community-architect, and it walks a member through twelve months to produce three things: a Life Plan, a smartbook of personal intellectual property, and a core community. Here are its three levers, named and costed.

The AI Age arrived with a canopy already grown over it, and that canopy is automation. AI writes your emails. AI summarises your meetings. AI drafts your reports. Every one of those claims is true, and the canopy is low and dense, and almost nothing new grows underneath it. I made the argument at length in my white paper AI Is World-Building: automation asks how to do the same work faster, and world-building asks what reality we can now create.¹³

Evidence suggests the canopy has already begun to thin. Anthropic's Economic Index, which reads millions of real conversations rather than survey answers, found that by November 2025 augmentation had risen to 52% of Claude.ai traffic while automation fell to 45%.¹⁰ Augmentation counts the conversations where a person learns, iterates or asks for feedback. People increasingly reach for these systems to think, not merely to finish.

Adoption tells a stranger story. The Reuters Institute found that 59% of people aged 18 to 24 across six countries had used a generative AI tool in the previous week. Among people aged 55 and above, the figure fell to 20%.¹² Its 2026 Digital News Report, drawn from close to 100,000 respondents across 48 markets, put weekly use of AI chatbots for news at 10% worldwide, up from 7% a year earlier, with 1% naming a chatbot as their main source and trust in AI news sitting at 20%.¹¹

Read those figures together and a gap opens. Usage runs wide. Trust runs thin. Authorship barely registers. Millions consult these systems every day, and almost nobody uses one to write something they intend to keep.

That gap is the seedbed.

Qualped plants one sentence in the canopy: a life is a manuscript, not a product specification. You are not the assembly stage of somebody else's design. You hold the pen.

Overstories stay local before they travel. Gladwell's cases are places, not demographics: Miami, Poplar Grove, one stretch of Los Angeles. Qualped's canopy therefore grows first over Durham Region, Ontario, population near 753,090,¹⁵ a region whose founding story is assembly. Changing that story here carries weight. Changing it on the open internet carries none, because the internet has no canopy, only weather.

Storytellers change overstories. Thirty years in newsrooms taught me how.

Aerosol science holds a finding every founder should read. Sima Asadi and colleagues at the University of California, Davis measured particle output during ordinary speech and published the results in Scientific Reports in 2019. Emission ranged from roughly 1 to 50 particles per second and climbed with loudness, and a small fraction of participants behaved as "speech superemitters", releasing an order of magnitude more particles than their peers. Loudness alone failed to explain the difference. Something in the speaker did.⁷

Epidemiology shows the same lopsided shape at population scale. Dillon Adam and colleagues traced 1,038 confirmed cases in Hong Kong and estimated that 19% of them seeded 80% of local transmission.⁵ Jacob Lemieux and fifty co-authors sequenced the Boston outbreak for Science, tracing roughly 100 direct infections from a 175-person hotel conference in February 2020 into an estimated 245,000 American cases by 1 November, within a range of 205,000 to 300,000.⁶ One room. One weekend. Lemieux warned afterwards that the conditions producing such events had not disappeared.

Marketing drew the wrong lesson from all of this and built an industry on reach. Reach counts how many people can see you. Emission counts how much of you enters another person. Those are different physics, and the second one moves ideas.

QwaiAI needs emitters, and I know where they stand.

Call them the Hinge Hundred. A hinge person is somebody others already consult at the moment a chapter turns. Nobody appoints them. Nobody counts their followers. They hold the position anyway:

• Pastors and youth pastors
• High school guidance counsellors
• Barbers and hairdressers
• Settlement workers and diaspora radio hosts
• Funeral directors
• Bookkeepers to small businesses
• Language teachers and adult educators
• Coaches of teenage sport
• Nurses in family practice
• Editors of community newspapers

Three questions select them:

1. Do strangers ask this person what to do next?
2. Does this person talk with people rather than at them?
3. Do this person's circles overlap at least three others?

Qualped recruits one hundred. Not a million, not a mailing list, not an audience. One hundred named human beings inside one region, each standing where a life bends.

Situation matters as much as person. Biogen's conference filled a crowded indoor room for two days, and the room did much of the work. Superspreading needs somewhere to happen. Qualped therefore builds the room, and the room carries a number.

Rosabeth Moss Kanter wrote the founding paper in 1977. Studying a Fortune 500 sales force, she showed that relative numbers, not personalities, govern how a group behaves. Groups she called skewed, where one type overwhelms another, punish the rare few: tokens absorb a disproportionate share of attention, differences get exaggerated, and individuals shrink into role entrapment.² Add members of the minority and the same people behave differently. The number changes the behaviour.

Damon Centola put a threshold on it. His team ran ten online groups of twenty people who had settled on a naming convention, then introduced committed minorities of varying size. Minorities below the threshold failed every time. The largest failed minority reached 21%. The smallest successful one reached 25%, and at that point the majority flipped quickly.³

Mariateresa Torchia, Andrea Calabrò and Morten Huse tested the idea inside 317 Norwegian firms and found the threshold expressed as a headcount rather than a percentage. Moving from one or two women on a board to at least three raised the level of firm innovation.⁴ Three, not thirty per cent.

Erica Chenoweth's figure travels furthest and suffers most from misuse. Analysing 323 campaigns between 1900 and 2006, she and Maria Stephan found nonviolent campaigns succeeded 53% of the time against 26% for violent ones, and Chenoweth later observed that no campaign drawing active participation from 3.5% of a population had failed.⁸ She has since published cautionary updates on that rule, noting that only 17 of 169 nonviolent campaigns in her expanded dataset ever crossed the line.⁹ Treat 3.5% as a destination. Never treat it as a launch condition.

Founders keep calculating their proportion against the market, where the number crushes them before they start. Proportion is a property of a room. Calculate it against the room, and the number becomes four.

Here is the rule that launches QwaiAI: four authors in every twelve.

Twelve, because twelve is the largest circle in which every member can still speak every other member's name in one sitting, and because a hinge person can actually gather twelve people on a Tuesday evening.

Four, because four of twelve is 33.3%. That clears Centola's 25% flip with margin, sits inside Kanter's tilted range rather than her skewed one, and holds one more than Torchia's three, so a single departure does not collapse the minority back into tokens.

Growth follows from the rule rather than from a budget. One hundred hinge people convene one pod each. Every pod that reaches four authors sends those four out to convene pods of their own in the following season. Qualped already runs on seasons, because a member re-authors the blueprint four times across the twelve-month journey.

Honesty demands the caveats. Pods will fail. Members will move away. Some hinge people will say yes and never gather anybody. Halve every figure in that table and the shape survives, and the timeline stretches to eight seasons instead of four. Centola ran his experiment in an artificial system with a financial incentive to coordinate, and a naming convention is not a life plan. Chenoweth built her dataset from campaigns to remove governments, not to launch platforms. Models describe a shape. Only the work produces the numbers.

Gladwell subtitled his book The Rise of Social Engineering for a reason. His closing chapter shows the same three levers assembling the deadliest phase of the opioid epidemic: an overstory that treated pain as undertreated, a small number of prescribers doing most of the damage, and a reformulation that changed group proportions and pushed users toward the illegal market. Nothing in these levers carries a conscience. Anyone who lifts them gets the same mechanical advantage.

Two rules govern their use at Qualped.

First, break the monoculture. Gladwell puts it plainly: The best solution to a monoculture epidemic is to break up the monoculture.¹ A pod of twelve people drawn from one workplace, one congregation and one decade of birth will spread nothing new, because everybody inside it already believes the same things. Qualped therefore builds pods deliberately mixed by trade, age, faith and country of origin. Mixing is not a virtue statement here. Mixing is an epidemiological requirement, and a monoculture pod fails the design.

Second, the author owns the pen. QwaiAI drafts, questions, structures and remembers. QwaiAI never decides what a person should want. Purdue Pharma changed an overstory in order to sell a product into a population. Qualped changes an overstory so that a population can write its own. The difference sits entirely in who holds the pen at the end, and any system that quietly takes the pen has stopped serving the human being and started farming one.

Oshawa spent a century proving that people will build magnificent things to somebody else's drawing. The drawings were never the problem. The absence of a pen was.

One hundred people. Twelve to a room. Four who pick up the pen.

Watch what grows once the canopy opens.