Hypothetical situation with some seriously made-up numbers:
* 2x $100k for 5% investment in companies at say a diluted stake of 2% goes from $10B to 5B
* 5x $100k for 5% investment in companies who go bust
* 3x $100k for 5% investment in companies who are acquired, make 10% returns ($110k per).
Total investment: $1m
Return from non-outliers: (5 x $0) + (3 x $110k) => $330k for a net of -$570k
Return from outliers before devaluation (on paper): 2 x .02 x $10B => $400M
Return from outliers after devaluation (on paper): 2 x .02 x $5B => $200M
At this point, the losses/returns of the non-outliers are in the noise. But man, it hurts to have to explain the "loss" of your upside by such an enormous amount. I don't think VCs would shrug it off as just a part of their overall portfolio, since these are the valuations that prop up the overall performance.
One thing to consider is that a power law distribution (which is how VCs see their investments), remains a power law distribution no matter how far you zoom in, so even in the "unicorn" bracket, they expect a few of the unicorns (Uber/The Honest Co) to do most of the work.
"We thought it was worth more, and it may well be in the future. But if it's not, that's OK, that's why we have a portfolio of 150 companies; some will lose money. We plan on that happening to x% of our companies, and on average we're still lanning on seeing a #x return on the capital you invested."
In other words, this is absolutely an expected part of VC, even at the late stages.