Six steps to build things that work for everybody.

Most things get built for the "average person." That person does not exist. This program shows you a different way. You start with the people the system fails the hardest. You build for them first. Then something strange and wonderful happens: it works better for everybody. Six steps. That's the whole thing.

Three things I believe

1. 01

   Build around the people who are pushed to the edges, and everyone ends up better off.

2. 02

   New ideas start where identities overlap. Nobody is just one thing. Where several things stack up on one person — that’s where a system fails hardest, and that’s where the best ideas are waiting.

3. 03

   Turn pain into projects. The thing that hurt you is not just a wound. It’s information. It’s a map. It’s the start of something you can build.

Why "The Overlap"?

Picture three circles that sit on top of each other. Not crossing. Overlapping. Each circle is a part of who someone is. Deaf. Grew up poor. Speaks Spanish at home. The little patch in the middle where all three sit on top of each other is a real person — and that person hits problems no one in any single circle hits alone.

All three at once is not three problems. It's a different problem. And nobody designed for it, because everybody designed for one circle at a time. That middle patch is the overlap. It's where the system tells the truth about itself.

The idea has a name and an author: intersectionality, named by Kimberlé Crenshaw in 1989. She was writing about Black women losing discrimination cases — because the court would look at race, then look at gender, and never look at both together. The harm that only happened at the overlap was invisible to the law. Say her name when you teach this.

And why overlap, not "crossing"? A cross is an X, and an X means cancelled. Nothing here gets cancelled. Things stack. A person carrying four things at once isn't diminished — they're carrying more information about your system than anyone else in the building. Overlap is addition. That's the right word.

The story I open with

In 1950, the U.S. Air Force had a problem. Their pilots kept crashing. The planes were fine. The training was fine. So they looked at the cockpit — built to fit the "average pilot" across ten measurements. Then a young lieutenant named Gilbert Daniels measured 4,063 pilots and asked one question: how many are actually average on all ten?The answer was zero.

They had built a cockpit for a man who did not exist, and pilots were dying in it. The fix wasn't a better average. They made the seat move. They made the straps adjust. They made the pedals slide. Your car seat moves because a bunch of pilots didn't fit an average.

Build for the average and you build something that fits nobody. Build for the overlap and you build something that fits everybody. And notice why the average failed: any one pilot might be average on height, or arm reach. The average only broke down when you asked for all ten at once. The average died in the overlap.

The proof you can point at

Look down at any street corner. That little ramp cut into the curb was built in 1945 in Kalamazoo, Michigan, for disabled soldiers coming home in wheelchairs. Now: parents with strollers, kids on bikes, delivery carts, anybody dragging a suitcase, you carrying groceries. Built for a few. Ends up helping everybody.

Same story everywhere. Captions were for Deaf people; now everyone watches TV with them on. Voice control was for people who couldn't type; now it's in your car. Dark mode was for low vision; now you turn it on because you like it. That's not a nice story. That's a method.

Where this comes from

I did not make this up. Disability justice was started in 2005 by Patty Berne, Mia Mingus, Stacey Park Milbern and others at Sins Invalid. In 2015 they wrote down ten principles. Their first principle is intersectionality — they quote Audre Lorde: we do not live single-issue lives. Their second principle runs this whole program: the people most affected should lead. Not be asked. Not be surveyed. Lead.

Being straight with you: disability justice is a movement, not a business tool. One of its ten principles is against the way capitalism works — and here I am, in a bow tie, selling you a method. That's a real tension and I'm not going to pretend it isn't there. So here's my deal. I name where this came from, every time. The people I work with get paid like the experts they are. They get their names on the work. A set share of what I earn goes to disability-led groups. If I ever stop doing that, call me on it.

Find — who is carrying the most at once?

Not who's the loudest. Not who complains. Who's holding four things at the same time, so your system fails them in a way it fails nobody else? That person is not an edge case. That person is your best instrument.

A trick for finding them: look for the workaround. Somewhere, somebody is doing a weird extra thing to make your system work — calling a friend to read something, photographing the screen, coming in an hour early. That workaround is a flashing light. The person with the most workarounds stacked up is the person you're looking for.

Write one sentence: "This system fails ______ the hardest, because they're dealing with ______ and ______ at the same time." Usually the room knows the answer right away. Sometimes there's a long pause and somebody says a name quietly. That pause is the whole job. Go toward it.

Hand over — put them in charge. Pay them.

Not on a survey. Not on a "listening tour." In charge. Four rules — all four, or you're faking it:

• They can say no, and it sticksnot: they give feedback
• Paid like an expertnot: given a gift card
• In the room when it’s decidednot: told about it after
• Name on the finished worknot: called "a user"

This is the hard one because it costs money and it costs control. That's exactly why it works — everybody else skips it. And one more thing: that person has a whole life. Build the work around that or you'll lose them by week three. This isn't being nice. It's how the work survives.

Map — every place it breaks them.

A break is not their fault. A break is a bad fit between a person and a thing — the thing was designed wrong. Say it that way, out loud, every time. It changes the room.

Sort every break three ways — the same break belongs to three groups. One arm: born that way (always), broken for six weeks (for a while), holding a baby (right now). Remember this table; it comes back in Step 6.

Then rank: how often does it happen? Does it slow them down, or stop them cold? What does their workaround cost them — time, money, pride? And the question most people miss: which breaks only show up when two or more stack together?Those go to the top. Always. A break that only exists in the overlap is a break nobody else on earth is working on. That's not a small market. That's an empty field. Circle the top one. You build from there.

Flip — build from the worst break, not from the average.

Take the system apart into ten pieces — the person, the start, the thing, the before, the after, the rules, the doing, the tool, what's around, the point. Ask six questions of any piece: who, what, where, when, why, how. Make one of six moves: keep it, more of it, less of it, flip it, swap it, drop it. Pick how far it reaches: one, some, all.

Piece + question + move + reach = one idea to try.

Do that a hundred times. A computer can do it in a second — that part is not the skill. Here's the skill: throw most of them away. Score each idea 0–3 on three things: does it make sense, is it really different, could we actually try it. Keep it only if "makes sense" AND "really different" are 2 or better. "Could we try it" never kills an idea. Ever. It only decides what you do first. A great new idea is supposed to be hard to build; throw out everything that's hard and you throw out everything new.

When to stop: go three rounds deep, then stop. Each round must be MORE different than the last — if it's not, that branch is done. If it stops making sense, stop. If you already wrote it down, skip it.

Try — no idea gets ranked until somebody has used it.

This isn't a hunch. In 2025, researchers had experts rate a big pile of ideas. Then 43 of those experts actually built the ideas. The ratings changed. A lot. The ideas that looked best on paper were not the ones that worked. Ranking ideas on paper stopped working. So stop doing it.

The 3-day build: three days, that's it. Spend only what you can lose without it hurting. Make one real thing a person can hold, click, or use. Write the pass/fail question BEFORE you build — once you build something you start to love it. The person from Step 2 uses it. Uses it, not watches a demo. Then rank again. They will change order. Every time. That flip is the most useful thing in this whole program. Write it down. It's your proof.

Share — walk the fix up the ladder.

Remember the table from Step 3? Here's where it pays off. Rung one: does it work for the person we built it for? Rung two: who has this problem for six weeks? Rung three: who has it for six minutes, today? Rung four: what does everyone get out of it? One rung at a time — each rung should make the next one cheaper and easier. Each rung is a bigger overlap: you started with the small patch in the middle, and now you're growing the circles outward until they cover the whole page.

You never skip rung 1 to get to rung 4. Rung 4 is where the money is. Rung 1 is where the truth is.

If you build something that fails the person it was made for but sells great to everybody else — that's not a curb cut. That's taking their problem and selling it back to the world without them.

What to say when someone pushes back

Sources — say these out loud, always

Credit is not a footnote. It's part of the method. If you take the ideas and drop the names, you're doing the exact thing this program is against.

• Intersectionality — Kimberlé Crenshaw, 1989.
• Disability justice — Patty Berne, Mia Mingus, Stacey Park Milbern, and Sins Invalid, 2005. Ten principles, 2015. Read them yourself.
• The average pilot — Gilbert Daniels, U.S. Air Force, 1952. Double-check the exact numbers before they go on a slide.
• Curb cuts — Kalamazoo, Michigan, 1945. "Curb-cut effect" popularized by Angela Glover Blackwell.
• "Solve for one, extend to many" — Kat Holmes and the inclusive design team at Microsoft.
• Ideas on paper vs. ideas built — Si, Yang and Hashimoto, 2025, and the 43-expert execution study that followed.

Turn pain into projects

Somewhere in your life is a thing that didn't fit you. A door you couldn't get through. A form you couldn't fill out. A test you couldn't take the normal way. And it probably wasn't one thing at a time — it was the form and no ride and a bad weekand nobody who spoke your language. All stacked up. All at once.

Most people are told to get over it. I'm telling you it's a blueprint. You already built a fix in your head. You've been running Step 1 and Step 4 your whole life without knowing the names. That's not damage. That's a head start.