Now I ponder what "small" means, exactly.
On the one hand, having fewer identifiers seems a useful definition; it means fewer potential points of bias.
On the other, having fewer -shared- identifiers -also- seems a useful definition; it means fewer potential agents you are going to share a bias with. From the individualist perspective, you want the most accurate information for yourself, and you're less likely to regard ideas from the ingroup with sufficient criticism. But from the collectivist perspective, you want the most accurate information for your group, which suggests that the rationalist collectivist doesn't include the collective as part of their identity.
Unless I miss an alternative logic?
I don't think Graham expects one's identifiers not be shared with others. He's arguing the point more in line with your first definition, in that the more aspects you add to your personal identity, the more sources of bias one must avoid. As Graham says, "all other things being equal, the best plan is to let as few things into your identity as possible."
In this context, individualist and collectivist are labels that describe an agent's preferred social arrangements and social norms. There's nothing about keeping one's identity small that prohibits one from having a higher priority for group preferences than individual preferences.
WEIRD may be weirder than you think. We Aren't The World writes of psychological experiments on non-Westerners that give vastly disparate results from results that have been assumed to be hardwired, and the implications of this:
Henrich used a “game”—along the lines of the famous prisoner’s dilemma—to see whether isolated cultures shared with the West the same basic instinct for fairness. In doing so, Henrich expected to confirm one of the foundational assumptions underlying such experiments, and indeed underpinning the entire fields of economics and psychology: that humans all share the same cognitive machinery—the same evolved rational and psychological hardwiring. The test that Henrich introduced to the Machiguenga was called the ultimatum game.
...
To begin with, the offers from the first player were much lower. In addition, when on the receiving end of the game, the Machiguenga rarely refused even the lowest possible amount. “It just seemed ridiculous to the Machiguenga that you would reject an offer of free money,” says Henrich. “They just didn’t understand why anyone would sacrifice money to punish someone who had the good luck of getting to play the other role in the game.”
...
At the heart of most of that research was the implicit assumption that the results revealed evolved psychological traits common to all humans, never mind that the test subjects were nearly always from the industrialized West.
Edit: The actual papers this article writes about are covered in this post by Ciphergoth from a few years ago.