The Love Machine

I've been to nine weddings in the last eighteen months. Barn weddings, beach weddings, warehouse weddings. I've heard every origin story in circulation. College sweethearts, coworkers-turned-something-more, "we met at a mutual friend's game night" and, increasingly, "we met on Hinge."

Any product built to solve a problem risks solving itself out of a customer.

Match Group operates one of modern capitalism's most paradoxical business models. The behemoth behind Tinder, Hinge, OkCupid, and Match.com generates nearly $3 billion annually from ~15 million paying users. That's roughly $200 per subscriber per year, extracted via subscriptions, super likes, and visibility boosts.

But every successful match is a loss to the machine. Every "how we met" becomes a churn event. The product's purpose (to help people find love) is fundamentally at odds with its profit engine.

Hinge proudly proclaims that their app is "Designed to be deleted."

The data tells this paradox clearly. In 2025, nearly 60% of newly married couples say they met through online platforms. It takes, on average, nearly 4,000 swipes and eight months to find a match who turns out to become a partner. That time horizon is intentional.

Tinder caps free users with 100 right swipes per day; Hinge supplies just 8 daily likes. Bumble enforces time-limited first moves. That friction manifests as strategic engagement hooks designed to keep you swiping long enough to keep paying.

Match Group's genius lies not only in mechanics, but in portfolio theory. Each of its apps targets a different stage of the dating journey. As users graduate from one app's demographic, they're funneled into another. Rather than churn, what Match Group engineers is churn migration: a seamless handoff that sustains lifetime value across their ecosystem.

The algorithm perfects what behavioral economists call "intermittent variable reinforcement", the most addictive reward schedule known to psychology. You're shown people slightly above your league often enough to maintain hope, but rarely enough to keep you satisfied. The business model depends on this inefficiency.

Perfect matching would mean rapid graduation.

A 2023 study found that couples who met online dated an average of 2.5 years before marriage, versus 1.5 years for couples who met offline. Dating apps don't just intermediate romance, they elongate it.

This is the graduation problem: when customer success means customer loss.

The graduation problem extends far beyond dating. Across industries, companies have discovered that solving problems too efficiently destroys their business model.

Duolingo promises fluency but delivers perpetual engagement. Users report spending years on the app without achieving conversational ability. The company went public on the promise of 500 million users maintaining daily streaks, not 500 million people who learned Spanish and moved on.

Peloton's unit economics depend on that $44 monthly subscription long after the bike is paid off. Too much success (you get so fit you prefer real cycling) and they lose a customer. Too little success, the bike becomes expensive furniture. The sweet spot is perpetual progress without arrival.

Weight Watchers has existed for 60 years on this principle. The average member loses 5-10 pounds then plateaus, maintaining their subscription for "maintenance." The company's own data shows that most members regain the weight within two years, creating a recurring customer cycle.

Even therapist-patient relationships can fall into this trap. A therapist who resolves client issues too efficiently might struggle to maintain a practice. The incentive structure favors extended treatment over rapid resolution.

The pattern is clear: modern subscription businesses don't sell solutions, they sell managed problems.

The graduation problem scales to B2B with even more sophisticated mechanisms.

Salesforce promises to organize your sales process, but their real innovation was creating endless expansion opportunities. Just when you've mastered Sales Cloud, you need Service Cloud. Then Marketing Cloud. Then Einstein AI. Graduation never comes, just the next module.

McKinsey and consulting firms broadly operate on this principle. They're hired to solve problems, but their business model depends on finding new problems. The best clients are perpetual clients, always one strategic initiative away from not needing consultants.

AWS architects its services so that optimization is endless. You can always be more efficient, more resilient, more scalable. The platform's complexity ensures that no one ever truly graduates from needing AWS expertise.

This creates a fascinating inverted innovation cycle:

The graduation problem isn't an accident, it's the logical outcome of how we fund innovation.

Venture capital demands recurring revenue and predictable growth. A company that solves problems too well creates what investors call "customer concentration risk" and "unit economics challenges." Translation: success is bad for business.

Consider the difference in valuations. Companies with high customer lifetime value and low churn rates command premium multiples. Companies that solve problems quickly and lose customers get labeled "consulting businesses" or "one-time value creators". That’s investor speak for uninvestable.

This financing structure has profound implications for how problems get solved in modern economies. The most funded solutions are often the least effective ones, while truly transformative technologies struggle to find backing because their success threatens their own business model.

We've created an innovation ecosystem that rewards problem management over problem solving.

Some companies have cracked the graduation code by making alumni more valuable than active users.

Harvard doesn't lose when students graduate, it gains donors. The university's endowment model transforms graduation from endpoint to beginning. Alumni give $1.5 billion annually, far exceeding tuition revenue. The real product isn't education; it's lifetime network membership.

Y Combinator wants its startups to graduate quickly because equity returns dwarf program fees. Successful exits generate massive returns, and graduated founders become mentors, investors, and an acquisition pipeline for the next batch. The community sustains itself through successful exits.

American Express charges annual fees but celebrates when cardholders graduate to higher spending tiers. Their business model aligns with customer financial success because prosperity increases transaction volume and reduces default risk.

The pattern among successful graduation businesses: they monetize outcomes rather than inputs, and they've designed systems where customer success directly drives revenue growth.

The graduation problem extracts costs beyond economics. When platforms profit from prolonging problems, users internalize these incentive structures.

Dating app users report feeling like they're "shopping for humans", reducing romance to a series of optimizable variables. Fitness app users develop dependencies on external validation rather than internal health metrics. Language learners plateau at "good enough" rather than pushing toward actual fluency.

We've outsourced progress tracking to platforms whose business models depend on perpetual incompletion. The apps teach us that we're always one upgrade, one feature, one swipe away from enough.

This creates what psychologists call "learned helplessness" or the belief that problems are too complex for individual resolution and require ongoing professional management. The subscription economy doesn't just sell services; it sells the belief that you need those services indefinitely.

The graduation problem represents a fundamental tension in market capitalism. Adam Smith's invisible hand assumes that individual profit-seeking behavior creates collective benefit. But when individual profit depends on prolonging collective problems, the invisible hand starts working against progress.

Consider healthcare. The most profitable medical interventions are chronic disease management systems, not cures. Pharmaceutical companies invest billions in diabetes management but comparatively little in diabetes prevention. The economic incentives point toward symptom control rather than root cause elimination.

Education technology follows similar patterns. Platforms optimize for engagement metrics rather than learning outcomes because engagement correlates with subscription retention while actual learning might lead to graduation.

Even personal finance apps profit more from users who perpetually struggle with money management than from users who achieve financial independence and stop needing the service.

This misalignment between profit and progress suggests we may need new economic models for innovation in problem-solving industries.

The graduation problem isn't inevitable, it's just a design choice. Companies can restructure their incentives to align profit with genuine customer success.

Revenue model redesign offers the clearest path forward. Instead of monthly subscriptions, companies could charge based on outcomes achieved. Dating apps could take a percentage of wedding costs. Fitness apps could charge based on measurable health improvements. Language learning platforms could command premium fees for verified fluency certifications.

Alumni network monetization transforms graduated customers from lost revenue into ongoing value creators. Former customers become referral sources, case studies, and upsell candidates for related services.

Ecosystem expansion allows companies to profit from customer success by moving into adjacent markets. A dating app that owns wedding planning, couples therapy, or family financial services can benefit from successful matches rather than suffer from them.

The key insight: sustainable business models align company success with customer success rather than fighting against it.

The graduation problem reveals something deeper about modern capitalism's relationship with human flourishing. When the most profitable strategy is to keep people perpetually dependent on your solution, we've created an economic system that profits from prolonging problems rather than solving them.

This extends the "Easy Button" dynamic I've written about before: our cultural obsession with frictionless solutions that promise outcomes without effort. The graduation problem is the Easy Button's logical endpoint: platforms that maintain just enough friction to prevent actual resolution while selling the illusion of progress.

We've moved from promising easy solutions to engineering permanent dependency on those solutions.

The implications extend far beyond individual companies. This affects which innovations get funded, which problems receive attention, and how we structure progress in society. If venture capital only funds businesses that benefit from customer dependency, we systematically underinvest in technologies that could achieve genuine breakthroughs.

The dating app economy is just one example. We've built an innovation ecosystem that rewards problem management over problem elimination, not because solutions are impossible, but because solved problems don't generate recurring revenue.

Recognition of this pattern opens space for new approaches: impact investing that rewards customer outcomes, cooperative business models that align user and company interests, and regulatory frameworks that account for the social costs of dependency-based business models.

Last weekend alone, about 12,000 couples got married who met online. Twelve thousand tiny graduation ceremonies for the platform. On dance floors from Vermont barns to California vineyards, vows were exchanged, glasses clinked, and guests cheered for stories that began with a well-timed swipe.

And somewhere in San Francisco, a data scientist quietly logged thousands of churned users. The machine had worked exactly as intended, which is to say, it had worked despite itself.

The graduation problem forces a choice: optimize for Wall Street or optimize for human flourishing. Companies that choose the latter may discover that solving problems, really solving them, creates more sustainable value than managing them indefinitely.

In an economy built on engagement metrics and retention rates, learning to let go might be the hardest disruption of all.

Up and to the right.