the deck
Craft · Dan Koe · future/proof · 10 Jan 2026

If you have multiple interests, don't waste the next 2–3 years

Range was sold to you as a weakness. It isn't — it's the edge, if you give it a vessel: a way to channel every interest into work you own. Seven ideas, from why the generalist wins to the system that compounds it.

Reformatted for internal use · animations are CSS & SVG, no external assets
Idea I

The three ingredients — and the death of the expert

The man whose whole life is spent in performing a few simple operations… generally becomes as stupid and ignorant as it is possible for a human creature to become.— Adam Smith

Specialisation won the Industrial Age because a pin factory of one-step workers made 48,000 pins where one all-steps worker made 20. Schools were built to feed that machine — punctual, obedient, one narrow function. That system does not need you specifically to do the task. If pure specialisation makes people stupid and dependent, what makes an individual smart and sovereign?

  • 1Self-education — direct your own learning; a different result than school demands a different method.
  • 2Self-interest — concern with your own flourishing (Rand's sense: neither predator nor doormat). Cheap dopamine is a corporation's interest, not yours.
  • 3Self-sufficiency — refuse to outsource your judgement, learning and agency.
GENERALIST self-education self-interest self-sufficiency
Interest motivates education · education enables sufficiency · sufficiency clarifies interest.

What this section means

The generalist isn't a dabbler — they fall out of this triad naturally. Every founder or creative you actually admire understands enough of marketing, product and people to direct the ship and adapt when it turns.

Idea II

You're living through the second renaissance

Study the science of art. Study the art of science… Realize that everything connects to everything else.— Leonardo da Vinci

The last moat is an opinion — a perspective only you hold, because only you lived your life. Every interest you've pursued leaves a residue: more connections, a more complex model of reality, so more problems you can solve and value you can create. Specialism halts that process. Your edge is in the intersection, not the expertise.

This is the Renaissance pattern. Before Gutenberg, knowledge was hand-copied and scarce. Within 50 years, 20 million books flooded Europe and — for the first time — one person could pursue several domains of mastery in a lifetime. Da Vinci painted, sculpted, engineered, dissected. AI is the new printing press: it collapses the cost of knowledge again.

your edge domain A domain B
Psychology + design. Sales + philosophy. Fitness + business. The pure specialist can't see it.

What this section means

Shiny-object syndrome was trying to tell you something. Stop apologising for range — start compounding it. AI can only think what you tell it to think; the direction is the scarce part.

Idea III

Turn multiple interests into a way of life

Four things are now true at once, and they need to combine into one life:

  • AYou have many interests but can't just learn forever.
  • BYou love interest-based self-education but do it in the margins of a career.
  • CYou want to be self-sufficient but don't feel you have value worth paying for yet.
  • DYou must adapt fast — nobody knows what work looks like in five years.

The combiner is attention. When anyone can write anything or build any software, the ones people know about win. So become a creator — not necessarily a "content creator", but someone who creates for themselves instead of for a paycheck. Every business is a media business now; treat social platforms as the mechanism that gets your interests in front of people.

What this section means

Reframe "learning" as research and it becomes the job. You're already spending the hours — spend them learning in public and you have the foundation of a business.

Idea IV

Turn yourself into a business

The barrier collapsed: a laptop, an internet connection, free (skill-based) distribution, and tools that replace teams. If you've ever helped someone with your interests, you're qualified. Two ways in:

Path 1 · skill-based

  • Learn one marketable skill
  • Teach that skill through content
  • Sell a product/service for that skill
  • One-dimensional — you niche down, chase profit over interest, build a second 9–5

Path 2 · development-based

  • Pursue your own goals → that's the brand
  • Teach what you learn → that's the content
  • Help others get there faster → that's the product
  • Focused on the eternal markets (health, wealth, relationships, happiness); you become the customer avatar
multiple interests niche down a second 9–5 you are the avatar
Path 2 contains Path 1 — you still get good at every marketable skill on the way.

What this section means

Don't be a YouTuber, a personal brand, an influencer. Be you — somewhere your work can be discovered, followed and supported. Pursue the goal, validate it on yourself, then help the past version of you.

Idea V

Brand is an environment

Not a profile picture and a bio. Brand is the little world you invite people into — the accumulation of ideas in a reader's mind after 3–6 months of following you, illustrated across every touchpoint. And underneath it, your brand is your story.

  • 1Spend a day writing where you came from — the low points, the experiences, the skills, and what actually helped.
  • 2Filter every idea, post and product through that story so the brand stays cohesive (not so you talk about yourself constantly).
  • 3List 5–10 people you respect online; note the patterns in their bio, picture and content; take your own spin — then stop overthinking it.

What this section means

The hard part is believing your story is worth telling. It is. And brand basically is content — so the next piece is where the work really lives.

Idea VI

Content is novel perspectives

The internet is a firehose; AI adds noise; trust and signal matter more than ever. Curate the best ideas in one place, in your own words. The best speakers keep 5–10 core ideas top of mind and repeat them — writing a truckload of content is how you find yours. Aim at the intersection of performance (others will care) and excitement (you care).

old / rare books curated blogs heavy-hitter accounts signal filter idea museum 1 idea → many structures
A representation of the mind you're trying to build.
Step 1Build an idea museum — a swipe file. Write every useful idea down the instant it lands. Format doesn't matter; the habit does.
Step 2Curate by idea density — 3–5 high-signal sources (old books, curated blogs like Farnam Street / Navalism, a handful of great accounts). Discovery takes months.
Step 3Write one idea 1000 ways — break 3 posts down to their structure (Claude is good for this), then rewrite a fourth idea in each. That's how you stop staring at a blank screen.

What this section means

Idea density compounds into a library so good people can't help but open your emails and pass your ideas around — which makes you less dependent on the algorithm.

Idea VII

Systems are the new product

We're in a systems economy. People don't want a solution to their problem — they want your solution, the one you built by getting a result for yourself. Dan's 2-Hour Writer started as a personal constraint: write everything I need in under 2 hours a day.

NEWSLETTER daily posts thread → carousel → YouTube → blog → products
Make everything newsletter-centric and you have one thing to worry about.
who you are

Brand

Your story and worldview — the world people enter.

what you make

Content

Curated, idea-dense perspectives in your own words.

what they buy

Product

Your system for reaching a goal you already reached.

what compounds it

System

One workflow where every output feeds the next.

The one-line version

Stop learning to feel smart. Give your interests a vessel — brand, content, product, system — and the learning becomes the work.

Range is the moat — if it has a vessel

The missing piece was never focus. It was a container that lets every interest pay rent.

— Dan Koe · future/proof · 10 Jan 2026
Apply it · 16 years in tech

16 years of residue. Give it a vessel.

You've done the self-education for a decade and a half — across infra, data, product, ML, and whatever you tinker with on weekends. What's missing is self-sufficiency: a container that turns that cross-domain range into work you own and can't be replaced from.

Idea I → applied

You have two of the three ingredients

Self-education: 16 years of it. Self-interest: you already build things nobody assigned. The gap is self-sufficiency — your judgement, income and direction still route through an employer. Closing it doesn't mean quitting; it means starting to own an audience and a body of work in parallel.

The move

Stop treating your range as something to hide in interviews. It's the thing a lab — or your own venture — actually can't hire twice.

Idea II → applied

Your intersection is 16 years deep

A fresh PhD goes deep on one axis. You can stand where distributed systems × evals × product sense × [your outside interest] overlap — and that composite view is exactly the "opinion" the essay calls the last moat. It's also what a frontier-lab hiring manager remembers.

  • +Name your two or three strongest domains and the non-obvious thing their overlap lets you see.
  • →Write from that overlap, not from the middle of any one field where you're outgunned.
Ideas III + IV → applied

Learn in public — it's the same hours

You're already reading papers and prototyping. Reframe it as research done in public: notes, evals, teardowns. That builds the exact signal a lab screens for and the foundation of an independent path if you want one. Take the development-based route — pursue your goal (an offer, a shipped system, going solo), teach what you learn, help the engineer you were three years ago.

The move

Don't be "an ML influencer." Be an engineer whose work is discoverable. One public artefact a week, tied to a real goal.

Ideas V + VI → applied

Write the 16-year story, then build the museum

  • VStory: a day spent writing the pivots, the failed bets, what each stack actually taught you. Everything you publish filters through it.
  • VIMuseum: a swipe file of the ideas that matter to you across your domains; 3–5 high-signal sources; practise writing one idea several ways until you have 5–10 you can repeat.
Idea VII → applied

One system, everything downstream

Pick a hub — a newsletter, or a repo + build log — and make every output feed it. An eval becomes a post becomes a talk becomes a section of a guide. Under ~2 focused hours a day, the body of work compounds while the day job stays intact.

Staying a pure specialist

  • Deepen one stack, hope it transfers
  • Learn privately, publish nothing
  • All eggs in the employer basket
  • Range shows up only as an apology

Giving the range a vessel

  • Publish from the intersection weekly
  • A swipe file → 5–10 repeatable ideas
  • An audience that would back your next thing
  • One hub where every artefact compounds

The moat you already dug

Sixteen years is a lot of residue. It's only wasted if it never leaves your head.

framework: Dan Koe · application: yours to run
Apply it · starting fresh

New to tech, many interests, stuck in tutorial hell

This is Dan's own story: years of learning, dopamine from feeling smart, life unchanged. The fix wasn't picking one thing — it was a vessel that lets every interest become work.

Idea I → applied

You're mid-triad — and missing the container

Self-education: happening (maybe too much). Self-interest: it's why you're here. Self-sufficiency: not yet — no value you can point to that someone would pay for. The essay's answer isn't "niche down"; it's build the vessel so the learning stops being an escape and starts being the job.

The move

Set a hard cap: for every course you start, one thing shipped in public. Learning without an output is the trap.

Idea II → applied

Your non-tech interests are the point

Psychology + design. Sales + philosophy. Fitness + business. The intersection is where you're not competing with a thousand pure specialists. For an AI-adjacent path, that might be LLM tooling × the field you already know — build for the domain you understand.

  • 1List your interests. Circle the two with the most overlap and the most pull.
  • 2Make your first projects live in that overlap — not in a generic tutorial lane.
Ideas III + IV → applied

Learn in public = your portfolio and your business, at once

You don't need permission or a credential. Take the development-based path: pursue the goal (a paid role, or an independent income), document the journey, help the person one step behind you. That trail is the portfolio a hiring manager wants — and the seed of something of your own if the job market drags.

The move

Ship a small thing every two weeks with an honest writeup. Boring on day one; a body of work by month six.

Ideas V + VI → applied

Write your story, start the swipe file

  • VStory: yes, even if you think it's boring. Where you came from, why you're switching, what you've picked up. It's the through-line for everything you post.
  • VIMuseum: jot every idea that grabs you the moment it does. Find 3–5 dense sources. Rewrite one idea in three structures — that's how the blank screen stops winning.
Idea VII → applied

Build one loop, not ten accounts

One hub — a simple blog or newsletter. Each project → a build log → a few posts → a thread. Every output feeds the next; you maintain one thing, not five.

Tutorial hell

  • Course after course, nothing shipped
  • Dopamine from feeling smart
  • Learning entirely in private
  • Niche into one hot framework → a second 9–5

A vessel

  • A project every fortnight + writeup
  • Work that lives at your intersection
  • Building in public — feedback and a trail
  • One hub where it all compounds

Don't waste the next 2–3 years learning

Learn in public, from your intersection, into one system. That's the whole move.

framework: Dan Koe · application: yours to run