20 ideas, as formulas

Each entry has three lines: what it is, the formula (an abstract version you can reuse), and the trade-off (what you give up by following it). Most bad decisions come from applying a good idea without seeing its trade-off.

Time and attention#

1. Finitude#

• What: you get roughly 4,000 weeks, and you'll never do everything.
• Formula: value of life = Σ attention × meaning, over a fixed number of weeks.
• Trade-off: accepting limits frees you from anxiety, but it can become an excuse for low ambition. Accept the limit on quantity, not on quality.

2. Opportunity cost (the "other 20")#

• What: every yes is a no to everything else. The dangerous options are the attractive ones.
• Formula: cost(choice) = value of the best option you gave up.
• Trade-off: ruthless focus raises depth but kills optionality and serendipity. Keep a small, bounded slot for exploration.

3. Marginal value#

• What: the worth of the next hour, not of the activity.
• Formula: give the next hour to argmax ∂P(success)/∂hour. Most activities have diminishing returns.
• Trade-off: optimizing at the margin is correct but tiring to compute. Replace it with floors (minimums) plus "everything else to the main goal".

4. Explore vs exploit#

• What: try new options or use the known best.
• Formula: explore share ∝ time remaining. Explore early, exploit late.
• Trade-off: exploring too long means never compounding; exploiting too early locks in a local best.

The AI shift#

5. Complements get expensive#

• What: when one input gets cheap, its complements get scarce.
• Formula: price(X) ↓ ⇒ demand(complement of X) ↑. Code is cheap, so verification is scarce. Content is cheap, so attention is scarce.
• Trade-off: the scarce side is harder to sell and slower to build (trust, taste). You trade speed for durability.

6. Verification bottleneck#

• What: producing is cheap; knowing it's right is not.
• Formula: throughput = min(generation rate, verification rate).
• Trade-off: investing in checks (tests, specs, formal proofs) feels slow and invisible, but it's the only way to use more AI output safely.

7. Taste#

• What: a distinctive point of view in a world of AI averages.
• Formula: value ∝ distance from the average × quality.
• Trade-off: strong opinions repel part of the market. That's the point, but it caps your total audience.

8. Costly signals (trust)#

• What: signals only mean something if they're expensive to fake.
• Formula: trust ∝ cost to fake the signal × time shown.
• Trade-off: real trust signals (a face, years of track record, guarantees) are slow and personal. You can't buy them fast.

Choosing a bet#

9. Query (segment)#

• What: marketing is a query on the database of people. A segment is only real if you can run it.
• Formula: segment is real ⇔ ∃ index (a place, a keyword, a person, a marketplace).
• Trade-off: easy queries (marketplaces, search) are rented and shared with rivals; owned queries (lists) are slow to build.

10. Distance#

• What: the cost for you to learn, reach and earn trust in an option.
• Formula: searchable space = {options : distance(you, option) ≤ budget}.
• Trade-off: staying close is efficient but can trap you in a shrinking neighborhood. Occasionally jump.

11. Market hole#

• What: a gap between what people want and what exists.
• Formula: success ≈ Demand × Hole × Access × Edge × Durability. Multiplied, so one zero means a zero overall.
• Trade-off: the biggest holes are often in new markets (a bet); the safest are in mature markets (small and contested).

12. Timing and market phase#

• What: new, mature, or mature-going-through-a-shift.
• Formula: ease = demand proven × rules reset × incumbents slow. That's the "third state".
• Trade-off: early means cheap entry and education risk; late means proof but competition.

13. Positioning (one product, one mission)#

• What: a product means one thing to one audience.
• Formula: marketing efficiency ∝ 1 / number of missions.
• Trade-off: each narrow product has a smaller market, so you need several brands, and each starts from zero trust. Share the audience underneath.

14. Owned vs rented distribution#

• What: who controls your access to customers.
• Formula: durability of channel = 1 − landlord's power to change the rules.
• Trade-off: rented channels are fast and big; owned ones are slow and small but yours. Use rented to fill owned.

15. Compounding#

• What: does each week's work stack?
• Formula: value(t) = Σ work × (1 + r)^(t − week) for stacking work, versus flat for resetting work.
• Trade-off: compounding things (audience, SEO, reputation) are slowest at the start, exactly when you need cash most. Pair them with a fast engine.

16. Durability (Lindy)#

• What: old needs outlast new formats.
• Formula: expected remaining life ∝ age so far (for non-perishables like needs and ideas).
• Trade-off: durable needs attract durable competitors. Bet on the need, stay flexible on the format.

Managing risk#

17. Affordable loss#

• What: decide what you're willing to lose, not what you expect to win.
• Formula: bet if loss ≤ what I can afford, and I keep assets if it fails.
• Trade-off: small bets rarely produce huge wins alone. The upside comes from stacking survivors.

18. Kill criteria#

• What: decide when to quit before you start.
• Formula: continue ⇔ signal(week n) ≥ threshold set at week 0.
• Trade-off: hard thresholds sometimes kill a slow winner. Use them anyway; escalation of commitment costs far more.

19. Barbell#

• What: a mostly safe core plus a few small, wild bets. Nothing in the fragile middle.
• Formula: effort ≈ 85% stable core + 15% cheap optionality.
• Trade-off: the core feels boring and caps near-term upside. That's the price of not betting your life.

20. Stage gates#

• What: the next project starts when the previous one proves itself or is consciously stopped, not when the calendar says.
• Formula: start(n + 1) ⇔ milestone(n) ∨ killed(n).
• Trade-off: gates feel slow when you're excited. They're what stop a portfolio from becoming "fits everything, fits nothing".

How they connect#

The ideas stack into one chain of reasoning:

1. Finitude and opportunity cost say you must choose.
2. Complements and verification say what's worth choosing in the AI era: the scarce side.
3. Query, distance, hole and timing say where to look.
4. Positioning, owned distribution, compounding and Lindy say how to make it last.
5. Affordable loss, kill criteria, barbell and stage gates say how not to lose years while you find out.