Bingo. Finance is just about allocating risk. When it goes wrong, it's either because 1) the risk taken was misunderstood, and/or 2) too much risk was taken.
Fraud falls into category 1 (as do many other things, like correlation of loan defaults), and leverage into category 2.
The mechanisms by which risk is allocated are generally quite simple.
I would say opaque rather than complex. The details of a CDS aren't complicated, nor is the Eurodollar system. But it can't be inspected from the outside, and even the Fed is reliant on running complex models in an attempt to model monetary policy impact.
Agreed, financial market structure is often intentionally obfuscated by incumbents. The worst public example is undocumented equity order types [1].
The Fed's monetary policy models are another story. If you drill down on the internal politics of the FOMC, you start to see the "research" as Kabuki theater which exists to justify whatever decision the big people upstairs want.
It is this political dynamic, rather than the reduced compensation, that deters most talented people from public service.
Back in '08 we heard about how "complex" CDS are. There is nothing complex about a CDS.
Whenever something bad happens in the financial sector, there is always a simple story behind it. Usually a combination of fraud and leverage.