In general, credentials are not worth much. If I don't have the technical background or knowledge to evaluate an answer given by an expert, I would be doing a disservice to myself to accept that answer on face value.Sure, I could use that explanation for learning. But still, it would be a lot better to get explanations from multiple sources.

So basically expert opinions can give a good start for learning something, but you won't learn a lot if you only go off that expert opinion.

Since the post you were referring to was about the value of ideas in math, credentials might have helped a little bit, because evaluating an idea's worth requires knowledge and experience in that problem domain. But even if that post were written by a credentialed expert, that fact would not help me a lot. I still have almost no idea how to evaluate the importance of an idea in mathematics, and learning that will take more than talking to one person.

Also, expertise and credentials are overrated. That's one thing that really shocked me while I was reading the four-hour work week. Tim Ferris suggested that you gain expertise and credentials, mentioning that it was quite easy. One of the ways he suggested was to go to some close by organization or fortune-500 company and hold a free seminar on a topic, which is easy to do because the standards for letting someone show up and do a free seminar are probably low. Then, when talking with someone, you just need to mention the seminars, and they will assume that the organization has evaluated you, and your worth will be viewed as dependent on that organization.

So basically, what I got from that was I naturally give too much deference to experts, and I don't really have the knowledge to give a thorough evaluation of someone's credentials.And a thorough evaluation of someone's credentials in an internet forum would probably devolve into a flame war.For example, the author of the article on the number 6174 has decent credentials, but it appears that he misused the definition of a kernel (See http://mathworld.wolfram.com/ModuleKernel.html).So how do I evaluate errors in an opinion where the value is determined from credentials? Do I say that it's not an error, the author just missed the details, or the credentials are not worth much?

Edit: Link to HN conversation:http://news.ycombinator.com/item?id=480398