If you first write at the bottom of a sheet of paper, “And therefore, the sky is green!”, it does not matter what arguments you write above it afterward; the conclusion is already written, and it is already correct or already wrong. To be clever in argument is not rationality but rationalization.
He makes a good point, especially about the real purpose of reason (to find truth, not defend beliefs), but I think he goes too far. Most of our conscious thought is rationalization to be sure, and I don't think we can avoid this, nor should we. Michael Gazzaniga's experiments highllight that we do it all the time, because the brain is constantly receiving signals from our non-articulate portions of the brain and trying to make sense of them. Your narrative self is like the press secretary of a large organization that it does not fully understand or control, but whose job it is to always ariculate a reason for why one feels or does something. And your mind is so hard wired for this, you don't even see that you are obviously confabulating, as when split brain patients or patients with bizarre brain lesions make up reasons why they think their mother is an impostor, or their arm belongs to someone else. If you were to question all your inner data feeds you would have no intuition, and then be as dumb as a computer.
So, in a sense we are rationalizing all the time. I don't think this implies we should embrace rationalization, and more than we should embrace being emotional because we are inherently emotional. We should merely be mindful that it is something to be managed, not eliminated, from our thoughts.
Clearly you can guess what someone will write on some daily event based on your knowledge of his or her prejudices. Paul Krugman and Thomas Sowell have worldviews that cause them to filter evidence a particular way. Both think it is an efficient worldview because it is more accurate than others, and they find the other's thoughts bafflingly inconsistent because they are based on some very primitive assumptions that are not shared. For better or worse, we all have a style, which is called a rut if it is unproductive. Mozart, HL Mencken were and Douglas Hofstadter is predictable in good ways.
It would be impossible to eliminate one's prejudices, because this is then merely a prejudice (e.g., everything is random and nothing integrates). Better to keep your deepest assumptions semi-private because if you say a key to my worldview is X, it becomes harder to change X because no one likes to be seen as fickle on their principles, it hurts one's credibility. That's why it is useful not to state them too strongly, repeat them too much, or try to serve them with every thought; you want to practice thinking without them, which should be easy given there are lots of useful frameworks existing simultaneously.
Think of your prejudices as something to manage. For example, cognitive therapy is the one therapy that does as well as the SSRIs, and it is based on changing one's thoughts based on rational evaluation, in that if you can see the irrationality of your depressive thoughts, you do not think them as much or as readily. Rationality can change your deep beliefs.