Pumping Heat Into the Infinite Game: Tobi Lütke on First Principles and Human Potential

Lenny opens by naming two themes that recur in Tobi Lütke's interviews and in the way people at Shopify describe him: first-principles thinking and maximizing human potential. He then asks about the 'Tobi tornado,' which Tobi describes as a large amount of change management, conflict, or real talk compressed into a short timeframe. When he sees that a project is not good, he has a conversation quickly, updates his priors, and may stop the project; the same team can then become the founders of the next version built differently.

Tobi says this can feel whiplashy, but it is also what many people appreciate because what is best ends up mattering. If he imagines a project might be wrong, either he is incorrect—and it is important to understand why—or he is correct, in which case letting people continue working on something that will not make it is unfair. Ignoring the problem would be an abdication of his CEO and founder responsibility, so compressing time is the better path.

Careers are limited—Tobi estimates roughly 40 years in the industry if you are lucky—so the goal should be to maximize the number of things you can be proud of at the end. That does not happen if everyone is optimizing a project that probably should not exist. Direct feedback, in this frame, is not about being the kindest or sweetest; it is about refusing to waste other people's limited career time.

Tobi believes there is not a single person on the planet who is close to their maximum potential. People are often much better than they think, but they are held back by undiscovered approaches, environments that narrow ambition, or habits that keep them competing with everyone else on unambitious work. He says reminding people of their own potential is wonderful, and he has a history of being right about people's potential more than they are themselves.

That belief is also the throughline of Shopify's product mission: Tobi wants the product to make people more successful than they thought they could be, and more ambitious about the businesses they build. His own path started with a snowboard store, not a plan to build Shopify. Following curiosity and optimizing each next step for maximum learning can take a person from one place to another, even if the world is full of misleading stories about human potential and progress.

School, in Tobi's view, teaches a fixed-time, variable-outcome model: learn a particular piece of math in a 12-month period and get graded on a variable outcome, regardless of how well you understand it. He says this has little to do with how people actually learn, and that following the thread reveals there is no speed limit for personal growth. Shopify has been an experimental lab for that conviction.

Tobi connects this to machine learning: overfitting is a model learning to cheat on a benchmark, and Goodhart's law is the business equivalent—any metric that becomes a goal ceases to be a good metric. School's loss function may be marks, which encourages cheating, but the right loss function for children is maintaining curiosity. He argues everyone is born extremely curious, and school fine-tunes kids in a way that can lose the neurons of curiosity by discouraging meandering into interesting topics.

Lenny asks how Shopify's core product team can operate without KPIs or specific goals, with decisions driven by taste and intuition. Tobi says Goodhart's law is real: no single metric is a complete heuristic for a complex business, because a company contains many tensions that cannot be kept in harmony by optimizing one fixed number. Shopify does not have KPIs or OKRs in the Silicon Valley sense, but it is extremely data-informed, with expensive systems that let leaders inspect every atomic part of a cohort formed 15 minutes ago.

Tobi estimates that the overlap between the most valuable product work and fully quantifiable work is maybe 20%, leaving 80% of the value space unaddressable by people who only look at quantifiable things. Shopify is comfortable with unquantifiable factors such as taste, quality, passion, love, and hate. The deep satisfaction a craftsperson feels after doing a job well can be a better proxy than a unit test passing, even if the unit test catches real problems 15 minutes later. He uses the metaphor of a cockpit: the systems provide instrumentation, but pilots still make the decisions.

The most powerful unquantifiable things in business, Tobi says, are fun and delight. If all the metrics are pointing down but everyone says they are having much more fun, he believes the metrics will start going up after some time delay; if they do not, the company adjusts course. Shopify specifically avoids OKRs partly because it wants people to defer to unquantifiable quality without being reminded of companies where the only way to get promoted is driving a metric up.

Tobi acknowledges that focusing on joy, fun, love, and delight can sound idiotic next to the aesthetic of suit-and-tie business people with PowerPoint charts and sophisticated metrics. He argues that optimism always sounds dumb or naive while pessimism sounds sophisticated, yet most of the alpha in the world is in things that are unobvious but true. Elon Musk is his example of a massively successful business person who conforms to no conventional aesthetic of sophistication.

Lenny reads a quote from Glen Coates describing Tobi as a futurist obsessed with how things should be, someone who sees designs drawn primarily from the way things are as inferior to designs skating to where things could or should be. Tobi agrees. Asked about his energy source, he says the world exists at room temperature, most companies hum along at that temperature, and founders and the best leaders inject heat into businesses. His own energy source is dissatisfaction with the status quo.

Tobi argues that today is the dystopia of the future: no one who really thinks about it would want to be born 20 years before today rather than today, and that pattern holds across human history. Builders should therefore make products that lead toward progress in a small or big way. If someone pitches a good version of what everyone already does, Tobi says that was not the job, because every existing solution is highly path-dependent and often shaped by compromises that were true when a decision was made but are no longer true.

First-principles product work means understanding the power and composability of every fundamental building block available now and deriving the solution from those blocks. Tobi admits this is a tall order and no one is perfect at it. Sometimes the status quo encodes more wisdom than expected, and that is delightful to discover. But skipping the exercise and copying everyone else is an abdication of product leadership.

Tobi treats the idea that you must do things differently as an axiom rather than an opinion: if you want to do something better than what exists, you have to do it differently, though that does not guarantee the result will be better. He is amazed by business plans that fail Archimedean logic, such as making a good version of an existing thing and capturing 1% of the market. He says he becomes extremely suspicious of pitches that promise a good version of what everyone else is already doing.

Tobi says first-principles thinking is learnable and encourages people to practice thinking step by step until it becomes a habit, partly because it outperforms. Shopify itself is his first example: e-commerce software in 2004 and 2005 was built for existing retailers that needed complex, Byzantine business logic ported online. Shopify optimized instead for new businesses, making the software intuitive enough that frustrated people in dead-end careers could make progress during lunch breaks. That focus later turned out to be better preparation for enterprise cases too, because enterprise software is overfit to the sales process and RFPs.

Lenny asks for the 'Tobi algorithm' of first-principles thinking, and Tobi describes thinking in programming constructs such as pure functions and state. He compares a perfect product lead to a thermostat for high-quality product: you set the goal and then re-derive every valuable decision and foundational assumption as new observations come in. Re-running the function over updated, higher-fidelity state may produce the same answer, but sometimes an early foundational assumption changes and the entire tree lands in a different place. COVID was that kind of change: the basic Boolean of whether people are allowed to leave their houses flipped, so the whole decision tree moved.

The remote-work decision also had other fraying inputs. Shopify started in Ottawa, a city of about a million people, and Tobi argues that is not a deep enough talent pool for a company going to 10,000 people. The company was already in four or five cities and stuck in an awkward hybrid pattern where Tobi could spend 10 hours of meetings in the Ottawa office without a single other person in the room. He says most bad outcomes in a path-dependent environment come from making only good local choices, so when COVID hit and Shopify needed to staff up to help merchants survive, hiring everywhere became the better trade-off. Remote work is harder—he compares it to running a world-record marathon in Aspen—but Shopify chose to port the internet into a company rather than port the office online.

Lenny notes that at a three-day all-hands hackathon, Tobi sat at a table with headphones in, coding like any individual contributor, and no one would have known he was the CEO. Tobi says that is his happy place: clearing days of his calendar and building until midnight with remarkable people. He apprenticed as a programmer in Germany starting just before he turned 16 and has been programming professionally ever since. He keeps a Jupyter notebook open to sanity-check projections and find new insights in data, and he argues first-principles thinking starts from atomic building blocks—computers are the instruments used to create software.

Tobi says he is successful because he does not try to conform to a traditional view of what a public-company leader should do; instead, he stays in as many details as possible and tries to inoculate the business against the sunk cost fallacy. In the hackathon example, the team was working through the pros and cons of merging Shopify into one huge monorepo, sketching directory structures and tooling. A monorepo is a consequential door-A-or-door-B choice that becomes very hard later, and Tobi says it is a business-strategy decision where 'Tobi said so' helps compress time. He holds opinions strongly until convinced, and he cites Maynard Keynes: when the facts change, I change my opinion.

Asked how to disagree with him, Tobi says just disagree. He craves it, immediately trusts people more for it, and thinks they will do what they think is right rather than what is convenient; courage, he says, is rarer than high IQ or genius. When someone disagrees, he stops and looks for the divergence point, especially unstated foundational assumptions that might be incorrect—such as a regulation like Sarbanes-Oxley that he has not studied in detail. He also plays devil's advocate when everyone agrees, especially if a proposal feels predictable, and makes himself an 'end boss' to force deeper argument.

Tobi argues decision making is understudied, even though Shopify makes millions of tiny decisions every day—whether to write code a certain way, add a unit test, or skip one. The company is not hiring engineers or accountants primarily for narrow specialization but for people who make excellent decisions within their areas. He wants the product to feel like it was made by one mind, not like a television remote with a Netflix button slapped on, and he wants decisions made together with clear lines of who needs to convince whom rather than by democracy.

Tobi often future-casts: look at the future and think backwards, asking what decision a future self would want made now. He did not build Shopify to flip it, despite opportunities to sell, because the journey was too interesting and because consolidation of all companies into a few owners would create too much of a monoculture in thinking. He likes that Shopify is different, even though he says it is a terrible place to work for many people and the best place in the world for some—people should be able to window-shop for a place whose beliefs fit them like a glove.

At a 100-year horizon you cannot specify a software product, but you can specify the mission and the things that will survive for 80 years. Shopify exists to make entrepreneurship more common, which Tobi calls precious: economics depend on businesses, yet most of the world does not celebrate entrepreneurship as a courageous and glorious act, and many people never meet anyone who has tried it. Because the world is path-dependent, a company can push the path toward the things it values, and long-term focus is also powerful for decision making.

Tobi points to Shopify's long partnership with Stripe as an example of playing an iterated prisoner's dilemma: both companies were small and decided to assume the other would win its market and work together. In every turn, coordination is more valuable over long periods than a momentary defection, but most companies fail the corporate marshmallow test by pulling future profits forward or defecting on partnerships. If the timeframe is 100 years, he says, there is no question that coordinating is the correct strategy.

Product pitches that promise more money often amount to putting future profits forward at a discount, and Tobi resists them because a long time horizon and customer trust are more valuable. He describes the best companies as a cockpit full of monetization dials and levers; his job is to add as many levers as possible and then not pull them. Shopify takes a very small stake of merchants' sales, so it is incentivized to make merchants as successful as possible. That is why he calls positive-sum games with customers, on a long enough timeline, the ultimate growth hack.

Tobi uses chess as an imperfect analogy because business does not have perfect information, but he likes that chess has two independent games: the positional game of developing pieces and gaining influence, and the tactical game of conversion puzzles and A/B tests. Business lionizes tactics—changing a button color and watching conversion rise—but tactics are not where most durable value comes from. Position is about what territory you occupy, how much merchants trust you, whether they want more or less of you, and whether your product subsumes other software. If you do the positional game well, the tactics are yours; if you extract all the value through tactics, you get 'tacticked' out of your position. Objective number one is simply: don't die.

Tobi frames company building as an infinite game in which founders create finite, winnable games for people that serve the mission's infinite game. He recommends James Carse's Finite and Infinite Games, though he warns it is hard to read and thinks Carse did not fully appreciate the grandness of his own idea. Shopify itself pushes from behind and does not want the credit; Tobi says there is too much grandeur in wanting to front it all. That is why Shopify helps people build their own businesses rather than starting with a marketplace that leases components at prices matching their margins.

Tobi treats good UX as almost a moral obligation: when software goes bad, it makes people feel dumb, and machines should not negatively influence people. Shopify's tax configuration is essentially invisible because the software can know the taxes and take that complexity away; every time a complex thing becomes simpler, more businesses exist on the platform. Entrepreneurs are often unsophisticated and scared, and a broken button or confusing API can make them close the browser and decide they are not cut out for it. Tobi says courage is rarer than creativity or genius, so lowering the amount of courage needed is enormously valuable; he points to Instagram's out-of-network 'try run' for reels as a profound UX idea that lowers stakes.

Tobi's career advice is to build a talent stack at the intersection of your unique strengths and curiosity, much as Brian Armstrong combined economics, coding, and cryptography before Coinbase. He cites Kevin Kelly's thousand true fans and a pirate-themed poker set to argue for making amazing things for some people rather than something everyone merely tolerates. He also points to Shopify hiring PhDs in dynamic-language garbage collection and JIT compilation to make Ruby fast, then merging the work back into Ruby Core so everyone benefits. He argues people should think of themselves as products: employment is a double opt-in subscription where you claim you will be worth more than you cost. For product leaders, the final rule is that product quality reflects how much the people who built it give a shit; leaders must be exothermically infectious about caring, and founder mode works because a founder can protect the people pumping heat into the system.