Big Software Companies Win by Sucking Less
To avoid disasters, giants let committees design software and leave coders to build it, giving up the chance at greatness.
paulgraham.com · Hackers and PaintersWhen Yahoo bought Viaweb, the online store building startup he had sold to Yahoo, Paul Graham, a programmer, told his new bosses he just wanted to keep hacking. He said he "had never liked the business side very much."
He soon learned that Yahoo used the word differently. To them, he writes, hacking meant "implementing software, not designing it." Programmers, he found, “were seen as technicians who translated the visions (if that is the word) of product managers into code.”
That, Graham argues, is "the default plan in big companies," and it is deliberate. Companies split design from coding, he says, because “it decreases the standard deviation of the outcome.” Only a small share of programmers can actually design software, he says, and "it's hard for the people running a company to pick these out." Rather than bet the product on one brilliant hacker, the company lets a committee of product managers, the people who decide what to build, draw up the plan and leaves programmers to carry it out.
It is a way to avoid disaster. Graham’s point is that it also avoids triumph. “When you damp oscillations, you lose the high points as well as the low,” he writes. Big companies can live with that tradeoff, he argues, because “they don’t win by making great products. Big companies win by sucking less than other big companies.”
A small startup, where designers also do the building, can therefore outdesign a giant, Graham believes. The catch is finding a fair fight. You cannot easily beat an entrenched rival on its own ground. The place to try, he says, is a new market, where no one has built fortifications yet. Microsoft, Apple, and Hewlett-Packard all did this at the start, he notes.
Questions people ask
What is a product manager?
A product manager decides what a product should do and what features it needs, then hands that plan to programmers to build. In Graham's account, that split is the problem, because the people with the idea are not the people writing the code.
What does decreasing the standard deviation mean here?
He means they make the range of results narrower and more predictable. A committee-designed product is less likely to be a catastrophe, but also less likely to be brilliant. In Graham's account, big companies accept middling results as long as rivals do worse.
If startups design better software, why don't they always win?
Not necessarily, and Graham does not claim they always do. His argument is that startups have an opening in new markets, where no one has built fortifications yet. There, a startup can win big by taking the bold approach to design.