Product Strategy and Second-Order Effects

A product decision is rarely contained inside the feature being discussed. It changes incentives, costs, customer expectations, and the competitive field around it.

Asymptotic Returns Are Not Guaranteed

"Near zero marginal cost" is the promise. Reality has fine print.

Principle

The Silicon Valley model assumes: small fixed cost → build software → each new customer is near-zero marginal cost → money printer. But costs have risen (larger teams, higher salaries), markets have saturated (no more conveyor belt of new internet users), and many new ventures have real physical/regulatory costs. The equation changed while nobody was looking.

When it applies

Evaluating a business model or startup idea
Deciding team size and hiring pace
Questioning "scale" as a magic word

Example

A delivery startup pitches "we're a tech company." But each delivery has real physical cost — drivers, fuel, insurance. That's not near-zero marginal cost. The SV playbook doesn't apply here the same way it does for pure software.

Fragmentation Is the Default

What looks like things falling apart might just be the end of an anomaly that held them together.

Principle

Paul Graham argues that the mid-20th century uniformity (economic equality, cultural common ground, political consensus) wasn't the natural state — it was an anomaly caused by WWII and large corporations. The "fragmentation" we see now (inequality, polarization, cultural divergence) is the return to the default. You can't fix it by recreating the anomaly.

When it applies

When lamenting that "things used to be better/simpler"
Analyzing why a team/org/community is fragmenting
Understanding that centralization is often temporary

Example

Your company's early culture felt unified. Now at 500 people, subcultures are forming and alignment is harder. This isn't a failure — it's the natural state. The early unity was the anomaly of a small group with shared context. Design for fragmentation instead of fighting it.

Source and further reading: The Refragmentation by Paul Graham.

The Mimetic Desire Trap

You only want it because someone else has it. Once you get it, you'll be bored.

Principle

Desire is mimetic — constructed by the people around you, not arising from within. Proust's characters only love people who are unavailable or desired by others. Social climbing, status-seeking, and romantic obsession all follow this pattern: the "inner ring" is always hollow once you're inside. Recognizing mimetic desire is the first step to escaping it and finding what you actually care about.

When it applies

Feeling envious of someone's career, title, or lifestyle
Chasing a goal and feeling empty upon achieving it
Evaluating whether a desire is truly yours or borrowed

Example

You desperately want to join the "cool" team at work. You finally get transferred. Within a month, you realize the people are vain and the work is boring. The desire was constructed by exclusivity, not by genuine interest.

Source and further reading: On Reading Proust's In Search of Lost Time by Nabeel S. Qureshi.

Moats Absorb Mistakes

Being right is optional when switching costs are high.

Principle

Companies with deep moats — network effects, file format lock-in, ecosystem gravity, monopoly distribution — can make unpopular decisions and survive them. Companies without moats cannot. The same move (subscription-only pricing, unpopular redesign, feature removal) looks like "visionary conviction" at a moated company and "suicide" at an unmoated one.

This means "customers hated it but the company was right" is often mis-told. The company was right because they had leverage, not because they saw something others missed. The lesson doesn't transfer to companies without the moat — and frequently gets cited anyway.

The inverse: if you don't have a moat, even correct decisions can kill you, because customers vote with their feet before you're proven right.

When it applies

Evaluating case studies ("Adobe did it, why can't we?")
Assessing risk of unpopular product/pricing changes
Building long-term strategy around contested decisions
Deciding whether a competitor's move is copyable

Example

Adobe's 2013 shift to subscription-only Creative Cloud was deeply unpopular. It worked because Photoshop had 20 years of file format dominance, muscle memory, and plugin ecosystems — switching was a career pivot, not a weekend project. A smaller competitor making the same move would have been abandoned in six months.

Figma later ate Adobe's UI-design lunch precisely because in that narrower domain, Adobe's moat didn't extend.

The Red Queen Effect

You have to keep running just to stay in place.

Principle

There's no such thing as a permanent lead. As you adapt, your competitors adapt too. Complacency will kill you. Sustained success comes from being flexible enough to change, letting go of what worked in the past, and focusing on what you need to thrive in the future.

When it applies

Resting on a successful product or skill set
Assuming your current tech stack will stay relevant
Feeling comfortable in your career trajectory

Example

You mastered React 3 years ago and stopped learning. Meanwhile the ecosystem shifted — server components, AI-assisted coding, new paradigms. Your skills aren't wrong, they're just no longer a lead. The race never stops.

Source and further reading: Mental Models: The Best Way to Make Intelligent Decisions (~100 Models Explained) by Shane Parrish.

Remove Friction, Get Nonlinear Effects

Making something 10x easier can cause 1000x more of it to happen.

Principle

Evan Williams' formula: take a human desire that's been around forever, and use technology to remove steps. Blogger, YouTube, Substack, Ableton — each removed a small amount of friction and got an explosion of creativity. The relationship between friction reduction and output is nonlinear. This applies to tools, processes, and any system where latent human energy is bottlenecked by barriers to entry.

When it applies

Designing developer tools or creative platforms
Simplifying internal processes or workflows
Evaluating product opportunities

Example

Making video creation 10x easier (smartphone cameras + YouTube) didn't produce 10x more videos — it produced millions of creators who never would have started. The same will happen when game-making tools get easy enough for non-programmers.

Source and further reading: Video Games are the Future of Education by Nabeel S. Qureshi.

Sales Is Qualification

Your job is not to sell. Your job is to determine fit — then selling happens effortlessly.

Principle

Most sales pain comes from skipping qualification and jumping to pitching. The 80/20 rule: customer talks 80%, you talk 20%. Listen until you can "play the movie" of their current process in your head. Identify their top 1-3 pain points. If they're not a fit, tell them honestly and refer competitors — this builds massive goodwill and often generates referrals. The actual "pitch" is just confirming their needs back to them with proof points.

When it applies

Any sales or persuasion conversation
Pitching your team's work to stakeholders
Job interviews (you're qualifying mutual fit)

Example

A prospect needs paper payment support, which your product doesn't offer. Instead of forcing the sale, you say "we're not the best fit — talk to [competitor]." They're pleasantly surprised, remember you, and refer three friends who are a fit.

Source and further reading: How To Sell by Nabeel S. Qureshi.

Strategy Tax

Sometimes the product is bad because it's serving a goal that isn't the user.

Principle

Not every product decision is a product decision. Many are strategy decisions wearing product clothing — serving advertising inventory, platform lock-in, executive-level narratives, cost reduction, acquisition optics, regulatory positioning, or a sister product's roadmap. The team often knows the user will dislike the change. They ship anyway because the decision was made above them, on a different scoreboard.

Recognizing strategy tax matters for two reasons:

As a user or critic: you stop expecting product logic to explain every move. Once you see that a feature exists to protect ad revenue or to justify a reorg, its shape makes sense.
As a builder: you notice when you're being asked to ship strategy tax and can decide, explicitly, whether to push back, route around it, or comply.

The danger is ambient strategy tax — where a hundred small decisions each serve non-user goals, and the product slowly becomes unusable by composition. No single decision was crazy; the composite is indefensible.

When it applies

Diagnosing why a product feels like it's fighting you
Reviewing roadmaps that have non-obvious prioritization
Understanding competitor moves that don't fit product logic
Deciding which features to defend against pressure

Example

Google Reader shutdown (2013) to push Google+. Reader wasn't killed because it was bad; it was killed because it was legible to the "social strategy" narrative as a competitor for attention. The product decision was a strategy decision.

Tumblr's 2018 adult content ban to preserve App Store placement. Skype's Snapchat-chasing redesigns in 2017. The 2020s Twitter/Reddit API pricing moves. Each defensible on some non-user scoreboard. Each indefensible on the user scoreboard.

The Ten-Year Hollowing

Great companies need bold, hands-on leaders who live and breathe the stuff they make or they'll eventually hollow out.

Principle

When a company replaces a product/engineering CEO with an operator (finance, logistics, marketing), the culture erodes invisibly for about a decade. Profits from past glories mask the rot. The operator optimizes what exists but can't generate what's next. Product taste and engineering ambition can't be delegated down the org chart — they must be embodied at the top.

When it applies

Evaluating leadership transitions at companies you work for or invest in
Noticing when "efficiency" initiatives start replacing "what should we build next" conversations
Understanding why a once-great team or org feels like it's coasting
Resisting the temptation to promote the best operator over the best builder

Example

Boeing, Intel, and Apple all installed non-engineering CEOs around the same era. Within ~10 years each: Boeing got the 737 MAX disasters, Intel fell behind on chip nodes, and Apple shipped Vision Pro to empty rooms and fumbled AI. The numbers looked fine — until they didn't.

Source and further reading: The Great Falls of Boeing, Intel, and Apple by David Heinemeier Hansson.

Closing principle

Strategy becomes clearer when you ask what a decision makes easier, what it makes permanent, and whose scoreboard records the result.