Re: what is random and what is not, I would encourage anyone who is interested to learn about modern probability theory (laid out by Kolmogorov in the '30s) and stochastic processes. One nice thing about this theory is the prominence that conditional probabilities and conditional expectations take. This allows you to describe 'deterministic' and 'random' as simply two extreme points on a spectrum. Something is deterministic when conditioned on a large information set, something is random when conditioned on a small information set. Probabilities are always relative to an information set*. Something that seems random is actually highly predictable given the right information, and something that appears predictable would have appeared random if you knew a little less.
One person may consider rolling a six being p=16.67%, whereas another person who noticed the die being slightly oblong might consider rolling a six being p=20%. Another person with a supercomputer and highly accurate motion detectors might see it as p=0% (or 100%, depending on what number his supercomputer spits out). All three people can be right in their own assignments of probabilities, the different numbers just illustrate that the probability of an event is always with respect to a particular information set. Two people with different information will assign different probabilities to the same event. It's not that one is wrong and one is right - they're both assigning the correct probabilities as they see it, it's just that one knows more than the other. The more you know, the less random things become.
Trading is like that - you have some people who know almost nothing about what's going on..they don't read the news they don't read company reports they don't look at historicals; they don't analyse anything. You have other people who spend their entire life reading all this stuff and conclude slightly different probabilities for assets rising or falling.
The point here is that the uninformed trader will see everything as random because from his perspective it actually is random. It's wasn't that mortgage-backed securities actually had an objectively high probability of default as such, it was just that he didn't go out and grab enough information that would have revealed to him a high probability of default. He sees things go up and down and has no idea why. He sees some people make money and others lose money and that too looks random. He sees the smart people who made money betting against subprime and assumes they were just lucky because, after all, this event was pure randomness. Then he looks to the people who lost money on subprime and says, "well they wore the same pinstripe suits and they worked in the same bank and they drank the same fine wine, so they're clearly indistinguishable from those other guys that made money...hence verifying my position that it's all just random and unpredictable".
If he was uninformed but smart he would realize that while to him things appear indistinguishable from noise, it could be that to people better informed it would be more predictable. On the other hand if he was closed-minded he might think "well if it appears random to me then it must be genuinely random."
This is the problem here...it's easy for things to look random or unpredictable, you just have to bury your head in the sand and ignore all the information that comes your way. For things to become predictable you need to do hard work, hence not very popular.
* A good candidate for an absolute probability may appear to be the unconditional probability. However, for any probability space with measure P you can always embed it in a larger space with a new measure Q that agrees with P in the small space but disagrees outside the smaller space. So while the unconditional probability looks unique and objective once you've constructed the probability space, if the space had been constructed differently then the unconditional prob would have been different too.