Paul Graham's 2023 essay gives founders a four-step problem-selection framework

Why it matters

Graham's essay was written for ambitious work in any field, not as an account of YC's founder-selection doctrine. Founders can still apply its four-step framework to choosing problems, learning through early products and deciding which overlooked gaps deserve sustained work.

Paul Graham published "How to Do Great Work" in July 2023 as an attempt to identify the habits shared by people producing important work across different fields. Framed as a guide for someone working in any field, its practical core is a four-step process: choose a field, reach its frontier, find the gaps and explore the promising ones. For founders, that framework offers a way to think about choosing problems and testing whether sustained interest can support a company.

Graham's career crossed programming, painting, company-building and investing. He earned an AB from Cornell and a PhD in Computer Science from Harvard, then studied painting at the Rhode Island School of Design and the Accademia di Belle Arti in Florence. He co-founded Viaweb with Robert Tappan Morris in 1995. Yahoo acquired Viaweb in 1998 and turned it into Yahoo Store. In 2005, Graham started Y Combinator with Jessica Livingston, Morris and Trevor Blackwell. That range gives weight to an essay built around the idea that useful discoveries often come from connecting fields that usually sit apart.

Choosing the problem before planning the company

Graham starts with problem selection. He argues that strong work usually begins where natural ability and deep interest overlap, then develops through direct experimentation. A person who does not know what to pursue should make a reasonable guess and begin working, because the work itself produces information that planning cannot.

That advice maps cleanly onto the earliest stage of a startup. A founder rarely has enough evidence to design the final company from a blank document. Building a small product, watching somebody use it and discovering which part creates value can reveal a better problem than the original pitch contained. Graham's essay makes the broader case for learning by working rather than trying to reason everything out in advance.

Graham pushes readers toward self-directed projects because they expose genuine motivation. He tells them not to let "work" mean only assignments from other people and to preserve the projects that feel "excitingly ambitious" as their taste matures. For founders, the startup translation is plain: the first product should come from a problem the builder cares enough about to keep probing when the first version is wrong.

This approach places curiosity ahead of market fashion. Graham advises readers to examine questions that hold their attention even when other people find them obscure. He adds an important constraint: pursuing an unpopular belief becomes useful only when the builder can explain precisely what prevailing opinion misses. Contrarianism without technical or market knowledge produces noise.

How YC's origin compares with the framework

Graham's account of Y Combinator's founding provides a separate case study. After speaking to a Harvard undergraduate computer club, he discussed the weaknesses of early-stage investing with Livingston. Their thesis was specific: investors should make smaller bets, back technical founders and become willing to finance younger teams.

Graham committed $100,000 to the experiment. Morris and Blackwell each added $50,000, giving YC an initial pool of $200,000. The founders created a summer program because they wanted to learn how to become angel investors quickly. They launched while students were still making summer plans and treated the first batch as an experiment with standardized seed financing.

YC's defining batch model emerged from that attempt to learn, rather than from a settled plan for remaking venture capital. Graham wrote that the founders initially underestimated the importance of funding several companies at the same time. The summer structure worked, so they repeated it. The sequence resembles the essay's four steps: begin with an interesting problem, learn through practice, notice what the initial model missed and pursue the opening.

The comparison also helps explain why Graham gives unusual ideas so much room. His YC origin story says the first summer was structured as a way to learn angel investing quickly, and that the founders only later recognized the importance of funding companies in batches. The essay makes a related case for exploring gaps that other people have overlooked, provided the person pursuing one has enough expertise to explain what others are missing.

Where the advice becomes expensive

The essay openly addresses a narrow audience. Graham says his recipe assumes a highly ambitious reader, and he eventually reduces great work to ability, interest, effort and luck. In the essay, he sets luck aside on the grounds that a person cannot do anything about it. That simplification keeps the framework actionable, though it leaves less room for the conditions that determine who can spend years exploring uncertain work.

Readers in the July 1, 2023 discussion challenged that omission. Some argued that the essay underweights financial security, health and privilege. Others questioned whether a framework derived from exceptional careers can guide people whose best outcome is satisfying, durable work rather than a field-defining contribution.

Graham does acknowledge the practical constraint. He writes that people who cannot earn money doing their preferred work can seek paid work close to it or reserve time for independent projects. In his formulation, the first option may compromise the work, while the second requires fighting for time.

The distinction matters because founder advice often treats intensity as a personal choice when it is partly a resource allocation problem. Time, savings, health and family obligations determine how many guesses someone can afford. A founder with a long runway gets more attempts to find the right problem. Graham's framework is strongest once those constraints are made explicit.

A framework for finding the work

Founders get the most value from "How to Do Great Work" as a framework for problem selection. It encourages them to build before they feel certain, protect the curiosity that drew them to a problem and resist letting competitors choose the roadmap. It also asks them to preserve morale, seek demanding colleagues and distinguish productive struggle from damage.

YC's origin story offers a separate example of learning through an initial experiment. Graham, Livingston, Morris and Blackwell created the summer program to learn angel investing quickly, then recognized the importance of funding companies in batches. That history does not turn the 2023 essay into a retrospective account of YC, but it gives founders a concrete comparison for the essay's advice about learning through work.

Graham does not offer a reliable formula for greatness. He gives founders permission to start small, follow evidence and let the scale of the work reveal itself.