With emphasis on that word "exactly," here is Gary Gorton’s superb paper The Panic of 2007.
Go ahead and read and read and read and the more you feel confused the more, in fact, you are being instructed. You are confused because it is confusing. Then I got to p.45 (!) and I almost split a gut (and cried, simultaneously) when I read the sentence:
Now we come to the first information issue.
It then goes like this:
What is the loss of information? The information problem is that the location and extent of the (2006 and 2007 Q1-2 vintage) subprime risk is unknown to anyone. It is very hard to determine the location of the risk, partly because of the chain of interlinked securities, which does not allow the final resting place of the risk to be determined. But also, because of derivatives it is even harder: negative basis trades moved CDO risk and credit derivatives created additional long exposure to subprime mortgages.
His examples show this in detail but I do not know a simple way to blog it. Scroll to pp.23-30, and p.35 for a dose of how these securities were structured.
Gorton is also highly critical of "mark to market" (p.62) and he pinpoints the collapse of certain parts of the REPO market (p.66) as a critical development. He ties it all in to Hayek and Grossman and Stiglitz and discusses how we ended up having assets with non-transparent, non-backwards-translatable prices and what that means for economic calculation. He contrasts a private clearinghouse (and monitoring) vs. rules of accountancy and how we ended up relying too much on the latter.
Starting on p.67, there is a sustained and mostly convincing argument that securitization has not been much at fault.
If you are interested in the nuts and bolts of the current financial crisis, and its origin in 2007, this paper is essential reading.