More complex packaging is part of it (think confusopoly of mobile plans applied to investing) and I think the paper has useful advice for "non preferential" investors, but it's not the whole story.
Is that true? VCs aren't acting the way they would if they really believed it. If they did, they would refuse funds from potential LPs, since they really can't find professionally responsible places to invest it. But that's not what they do. Alternately, they might take the LPs' money, warn them carefully that they could be investing in some real longshots, and then be willing to invest on fairly generous terms in any number of startups that look to be competently run and pursuing plausible opportunities. But that's not what they do, either.
Something doesn't add up.
There are many funds structured in a way where it's better for the fund (not necessarily for the investor) to invest 2x in questionable companies, than just 1x in great companies, lest they lose fees on the remainder.
There's more money than there are good deals in Silicon Valley, so later stage investors are forced to offer more money for less equity in order to beat other term sheets. This ends up looking like sky-high valuations, since investors that offer fair-market-valuations are unlikely to get picked. Founders naturally gravitate towards minimizing dilution.