From the Mind of Trevor Stone

I started reading Paul Graham when he wrote A Plan for Spam, and I wrote a masters' thesis examining several variants on Bayesian spam filtering. He generally writes insightful articles about creating tech startups, in large part because he's a domain expert on startup companies.

Graham's latest essay, on income equality is, however, mostly useless. (Perhaps because he's writing about economics and society, about which he is not a domain expert.) He published a simplified version of his argument boils down to the claim that economic inequality is purely a measurement and an outcome. He argues that economic inequality is not inherently bad and that we should instead focus on the problematic subset of causes of inequality. There's a grain of truth in this, but Graham totally ignores the outbound edges from economic inequality in the graph of social ills.

Some specific fallacies in Graham's essay:

Straw man

Graham seems to be arguing against the position that less wealth inequality is always better than more inequality. The end state of such a position is zero inequality, in which all people have the same amount of wealth, which is basically extreme communism. He says "You can't end economic inequality without preventing people from getting rich, and you can't do that without preventing them from starting startups." I'm not aware of anyone who actually holds that position. Even the Occupy Wall St. movement, a melting pot of some fairly radical ideas, wasn't advocating for the top 1% to hold precisely 1% of the wealth; they just thought the richest 1% should own significantly less than 50% of the wealth. The non-vacuous position Graham fails to argue against is the case for reducing income equality, not eliminating income equality.

Anecdotal fallacy

The long version of Graham's essay focuses on startup founders, with Mark Zuckerberg (Facebook founder) and Larry Page (Google founder) as anecdotes. Startup founders are probably disproportionately represented in the top 20 billionaires, but I suspect that they make up a smaller fraction of the full 1% cohort. Even if, as Graham argues, major wealth acquisition for startup founders is socially beneficial, that does little to support his argument that income inequality in general isn't problematic if most of the wealth is concentrated in non-startup hands. Graham's reliance on anecdote is so strong in this piece that he dismisses economic statistics as a way to analyze the situation.

Appeal to consequences

Graham suggests that reducing economic inequality would reduce or eliminate startup culture. Graham basically takes it as a given that startups are good, and therefore concludes that attacking economic inequality would be bad. There is plenty of room for both. Furthermore, startups might not contribute that much to wealth inequality. Initial startup funding generally comes from venture capital firms and individual wealthy investors. A moderately successful startup typically gets bought by a larger company, enriching the initial investors, the founders and early employees, and potentially the shareholders of the purchasing company (if the market reacts positively to the news). Wildly successful startups usually create wealthy founders when the company goes public and the stock market places a high value on the company. In both of these cases, the story is mostly about the already wealthy moving money around, some of which goes to a relatively small number of previously-not-wealthy folks. Even here, Graham doesn't address whether the existing wealth disparity between successful founders, ordinary tech workers, and folks in less-lucrative is better or worse than other potential wealth distributions. Should employees hold a greater fraction of startup shares? Should IPOs be taxed to support poverty reduction efforts? Graham's essay gives no guidance on such matters.

Graham's essay proposes an odd argument of inevitability, too. He cites the exponential curve of technological growth as evidence that economic inequality has historically and will continue to grow exponentially. This seems factually inaccurate: the western has significantly less wealth inequality today than it did under feudalism. I suspect too that technological and economic progress in the post-war era was greatly facilitated by the destruction of significant amounts of wealth which (naturally) disproportionately impacted the rich.

Graham points out the "pie fallacy"–that there's a fixed amount of wealth to go around–and spends much of the essay talking about creating wealth. However, he ignores the fact that many important components of wealth are finite resources for which pie-division is a very important concern. The most notable of these is land, a finite resource whose supply and demand imbalance is being felt particularly acutely in Paul Graham's back yard: Silicon Valley where even educated and skilled workers are finding it difficult to afford housing. A more subtle somewhat-finite resource is consumers. A society in which few people have disposable income is one in which building new enterprises becomes increasingly tough. The lower rate of income inequality in post-war America is an important example (though Graham tries to dismiss it) because well-payed workers play an important ecological role in a growing economy, providing a wide base which can buy new products in turn funding the creation of more new products. Perhaps such an arrangement is unstable: from a relatively equal distribution wealth will naturally accumulate with the institutions and individuals who reliably generate successful business. But perhaps there's another part of that natural cycle in which the wealth becomes too concentrated and the system destabilizes, leading to destruction and redistribution of wealth, starting the cycle anew. If that's the case, should we pursue a "controlled burn" approach of intentional wealth redistribution or should we follow a "forest fire" approach when wealth redistribution comes with little warning and dramatic upheaval?