You can see the end and still build the wrong thing. That’s the trap.
I wrote about Skateboard to Space Station a few weeks ago. Gall’s Law: complex systems that work evolved from simple systems that worked. Start with the skateboard. Iterate up.
There’s a corollary I left out.
You need to know what the space station looks like.
The shape of it. The gist.
Without it, you optimize the skateboard. Smooth wheels. New grip tape. LEDs on the bottom. You polish the thing you have instead of evolving it into what it needs to become.
My first Multi customer pulled me toward Descript’s game. More human editing. More manual control. More polish on each video. That feedback made sense for where the product is today. But I know the space station is agent-first. Agents produce the work. Following that pull means fighting a Goliath on their terms. Counter-positioning is asymmetric warfare. Use it.
The space station told me which feedback to ignore.
The space station is a compass.
You won’t build the one you drew. The real version will look different. That’s fine. The sketch gives direction. It shapes what you work on and what you stop optimizing.
Hold two things. The next step and the end state.
The middle is unknowable. Like planning move 30 in chess, except life has more variables than any board. Your plan blows up by step three. You replan. Blows up again.
I see my old self in this. Planning step 15 before finishing step 2. That planning felt productive. It felt like work. It was procrastination wearing a strategy hat.
Build the next step. Know the end game. The middle is unknowable.
🧭
What’s the space station version of what you’re building right now?
A few years back my feed was full of micro private equity.
Everyone was talking about buying small businesses. Roll-ups. SaaS. I read the threads. I ran the numbers. I started thinking that’s what I wanted.
Then solo VC took over. Everyone wanted to be an angel investor or a solo GP. I was positioned to do it. Trends.vc was rising. I had the deal flow. I had the network. I could have done it.
But I have a practice. I sit in a dark room with my eyes closed. White noise on. Just silence and wherever my mind wants to go. And what came back was: I don’t want that. I want to build. I want to experiment. I want to create.
That’s mimesis. You see what others want. You start wanting it too. Because it’s loud. Because smart people you respect are doing it. The desire feels real. It’s borrowed.
That’s what the practice catches. When I remove all input and sit long enough, the borrowed desires fall away. What survives the silence is mine.
Like sleep consolidates memories, this consolidates identity. You come back to your authentic self. You remember what you were building before the feed told you to build something else.
I almost became a micro PE guy. I almost became a solo VC. Both desires were borrowed. This practice is what caught them.
I built an app for car dealerships. Wrote code for three months before going to work each day. Pitched three. None bought.
I was playing a flat game inside a live one. The code worked. The market didn’t care.
That’s the trap for developers. We spend all day in flat games. Code compiles or it doesn’t. Tests pass or they fail. Then we try sales, cold email, marketing and the outcomes aren’t deterministic anymore.
There are four types of games in life.
Some games are flat. The rules are fixed, the outcome is known and the result depends entirely on execution. Writing code. Shipping a feature. Hitting a daily habit. There’s no randomness. You either did it or you didn’t.
Some games are random. The outcomes vary, but the rules are stable. A coin flip. A dice roll. Run enough trials and you start to see the distribution. You figure out what works 30% of the time and you plan around that.
Some games are live. The people in the game learn. A cold email. A product launch. Sales. Poker. What worked last quarter stops working because the other side adapted. There’s no fixed playbook. You adapt or you fall behind.
Some games are dark. Crises show up and you don’t control when. A market crash. A pandemic. A platform disappearing overnight. Platform risk. You can’t predict the frequency. The best you can do is buffer. But dark games have an upside too. Viral moments. Breakthroughs. The partnership that changes your trajectory. The downside finds you whether you’re playing or not. The upside requires you to show up.
DST kicked in. New York sprung forward. My clocks didn’t. The noon call moved to eleven.
Fix: pin the event to a time zone that doesn’t shift. Noon stays noon for me.
But if it doesn’t drift for me, it drifts for them.
Somebody has to absorb the shift.
Uber rides felt cheap for years. Investors absorbed the difference. Billions in losses to build a habit. Once riders were locked in and drivers had no alternatives, the prices climbed. Chris Dixon calls it “attract to extract.” Subsidize until you have power. Then extract.
AI companies trained on millions of books, articles and private data without paying the creators. The subsidy was other people’s work. Now those same companies attack open-source models for distilling against their outputs. Pull the ladder up once you’ve climbed it.
The United States’ economic rise was subsidized by seized land and enslaved labor. A country built by forced labor on stolen land now runs the largest mass deportation in its history. The growth was real. The subsidy was human.
Last week I had five style directions for a new video. I liked two. The other three were wrong.
But “wrong” was the most useful output.
One direction tried to look futuristic. Glowing lines, dark backgrounds, sci-fi energy. It looked like every AI-generated image on the internet. Futuristic ages fast. Anything that tries to look current looks dated in six months.
One had a beautiful color palette. Deep violet, pink, gold. But too much going on. Too many layers. Bold color works, but only when it stays clean. The palette was right. The style was trying too hard.
The last was just fine. Nothing wrong with it. Nothing I’d remember in a week. “Fine” doesn’t teach you anything. It’s the worst result because you can’t use it.
Those three rejections became guardrails. I turned them into constraints: avoid futuristic aesthetics, prefer timeless editorial traditions, warmth and texture over polish and glow. The next round will be better because of what I rejected this round.
A failed A/B test tells you what your audience doesn’t respond to.
A product launch that flops tells you which assumption was wrong.
A habit experiment that falls apart tells you which constraint you underestimated.
The wins confirm. The losses steer.
🚫
What did your last “failed” experiment teach you that success never could?
I heard Greg Isenberg say something on a recent podcast that I hear several times a week. Start with the niche.
He’s right. I’ve been on Greg’s podcast. I learn a lot from him. The advice is correct. Start narrow. Dominate a small market. Expand from there.
The problem is nobody talks about how.
“Find your niche” is exploit advice. It assumes you’ve already explored. It assumes the niche is sitting there, obvious, waiting for you to pick it. Most founders, creators and marketplace builders don’t have that luxury. You don’t know which niche until you’ve tried several. You have to fuck around to find out.
The advice skips the messy part. The part where you go wide, try a bunch of things at once, talk to different groups, build landing pages for audiences you’re not sure about, run outreach campaigns to segments that might not work. The part where most of it doesn’t land. Then one pocket lights up. People respond, they pay, they come back, they tell others. THAT is the niche.
The pattern isn’t “pick a niche.” The pattern is:
Go wide. Go narrow. Go wide. Go narrow.
Over and over. At each level.
Amazon started with books. Not because Bezos sat in a room and deduced that books were the optimal niche. Because books were the experiment that worked. He went wide first. While still at D.E. Shaw he evaluated 20 product categories, narrowed to five finalists and chose books because the catalog was massive and no physical store could stock them all. Then he went wide again into every category. Then narrow within each. Prime for frequent buyers, Marketplace for third-party sellers, AWS for developers. Wide, narrow, wide, narrow.
The same pattern shows up everywhere.
Marketplaces. “Start with the high-value niche” is the standard playbook. But which high-value niche? Airbnb started by renting air mattresses during conferences. They tried the DNC, SXSW, other events. Demand spiked during events and died between them. The bigger niche, entire-home short-term rentals, didn’t come from a strategy deck. It came from watching what the market was pulling toward and an investor pointing out the obvious. They had to go wide with event types before the real niche became visible.
Content. “Pick a topic and own it.” I published on habits, experiments, chess, jiu-jitsu, nomad life, pricing strategy, productivity frameworks, AI tools and startup strategy. All in the same year. The topics that resonated weren’t the ones I predicted. The audience told me by saving, sharing and responding. The niche found me. After I went wide.
The “just pick a niche” advice is an exploit strategy presented without the explore phase that precedes it.
This is the pattern I’m running right now.
Go wide. With Multi, I’m shipping to multiple segments at once. Different landing pages, different outreach, different messaging. Real effort in each. I make it easy to measure which segments respond.
Read the signal. Which segment converts without convincing? Which retains without begging? Which refers without incentivizing? That’s the pull.
Go narrow. I pour everything into the pocket that lit up. Customize the product, the messaging, the pricing for that group. Become their thing.
Go wide again. Once I dominate the niche, I expand to the adjacent market. Use the niche as a beachhead. Then narrow again within the new territory.
The cycle never stops. Each “go narrow” is an exploit phase. Each “go wide” is an explore phase. The mistake is thinking you do it once.
The failures narrow the search space. Each segment that doesn’t respond is a segment you can stop spending energy on. The failures are the finding-out part.
When someone says “start with the niche,” they’re right about the destination and silent about the journey. The journey is messy, parallel and necessary.
Fuck around. Find out. Then focus.
🎲
What would change if you treated this month as pure exploration?
I got my first book advice within hours of publishing.
“Price it higher. At least $4.99. You’re leaving money on the table.”
The advice was good. It was also the wrong time for it.
I priced 100 Rules to Live By at $0.99. On purpose. I wasn’t optimizing for revenue. I was optimizing for one rep. Get a book out. Feel the process end to end. Hit publish. That’s the variable I isolated.
Pricing? That’s a future pass.
Distribution channels? Future pass.
Cover design, marketing, Apple Books? All future passes.
This is how I think about building anything. Books, software, buildings. Each pass through, I pick one variable to optimize. The first pass is almost never about the thing people think it should be about. The first pass is about getting off zero and into motion.
People see a $0.99 book and think the price is the problem. The price isn’t the problem. The problem is most people with a 10-year dream of writing a book are still on page zero. I’m on page done.
After the first rep, I can run the next passes:
Test a higher price point
Expand to new platforms
Invest in cover design
Each pass isolates one variable and gives me signal on whether it moved the needle.
When you try to optimize pricing, distribution, packaging and marketing on pass one, you get paralyzed. If sales are low, which lever failed? You can’t tell. You changed too many things.
This is the same logic behind good experiments. Change one thing. Measure. Then change the next thing. If you change five things and something breaks, you’re debugging in the dark.
The deeper pattern: perfectionism disguises itself as preparation. “I need to get the price right. I need to be on all platforms. I need a great cover.” Each requirement is reasonable on its own. Stacked together, they become a wall. And the wall keeps you at zero.
I’d rather be live at $0.99 than perfect at $0.00.
More books are coming. The price will change. The distribution will expand. But each of those changes will be its own isolated pass, one variable at a time.
🔑
What would you build if you knew the first version was allowed to be “bad”?
I run a monthly event for the Trends VC community. Founders show up, get matched and have one-on-one conversations with someone new. Attendance is low. It sits outside the flywheel I’m building.
I kept it for two years because it didn’t seem like a huge investment. One hour. Once a month. Twelve hours a year. Who can’t spare that?
But I also run masterminds for Trends Pro. Eight people per group. The math looks similar. The alignment is completely different.
That was one commitment. Stack up a few like it and you’re losing a day a month. You’re bleeding out from cuts too small to notice.
Each cut looks rational on its own. Poker works the same way. I’ve been using an app to practice. Each hand has three options: fold, call or raise. In most positions, the right play is fold or raise. We “call” too often. We keep things around because the cost seems small. But calling bleeds your stack one hand at a time.
Each decision you face is a hand. Trends Tribe was a call. Masterminds are a raise. The flywheel audit forced a fold on Trends Tribe.
The filter is fold or raise. If it compounds, raise. If you’re keeping it because the cost seems small, that’s a call. Calling is how you bleed out.
A soft yes is a hard no.
✂️
Where are you calling when you should be folding or raising?
One bad meditation and the whole morning felt ruined. One missed experiment and I was “off track.”
This is the single-point trap. You evaluate yourself based on one data point. And one data point is noise.
Stock traders figured this out decades ago. They don’t buy or sell based on a single candle. They use moving averages: the 50-day, the 200-day. The longer the window, the less the noise matters. The trend is what counts.