[nathias]

That's not really the concept of rationality I or the article is talking about (consistency, non-bias etc.) knowledge or it's use isn't the same, it's more like the a priori knowledge where the concept is immediately applied by universal rules etc. This specific rationality is good in formal games where the rules are universal and the concept should be immediately applied, but doesn't work for empirical contexts (life, science, engineering, etc.).

[kragen]

They aren't really different concepts of rationality; consistency and non-bias are about not fooling yourself, so that you can come to the conclusions that the available evidence would justify. That's how people as a group can empirically acquire knowledge about the world. Of course, for individual people, social aspects are often even more important, since learning from someone else's experience can be much cheaper than learning from your own—as in the case of alpine sweetvetch; but even resisting deception and knowing whose opinion to listen to benefit from consistency and non-bias. Indeed, perhaps even more so, since the alpine sweetvetch isn't trying to emotionally manipulate you into believing it.

[nathias]

You use bias as a dirty word, but it's really just a weight of the opinion, there is no knowledge without bias. As for consistency, it comes secondary to categorization, it's easily abused for 'foolish' consistencies that can also be created with framing effects.

[kragen]

Sorry, I'm using "bias", "consistency", and "rationality" in the statistical, logical, and philosophical senses, respectively. So, I think, is the article. Your use of different definitions for those words probably explains why you reached conclusions that read as obvious nonsense to me. You might think about rereading the article with those definitions in mind.

I can also recommend reading about alpine sweetvetch.