The market cap is kind of a silly number that we like to look at. If somebody tried to sell 10 million bitcoins they couldn't get the spot price, the price would go down a little bit for the first few bitcoins, then there would be less byers and the price would drop, eventually the price would drop down so significantly that I doubt the actual dollar value extracted would be 1/10th of the market cap.
This is the crux of it.
Marketcap = (current trading price of coin) * (total number of coins).
The 'price' is a simplification, a single measure that takes no account of the depth of the order book or the spread of bids/asks; it takes no account of the number of coins that anyone could actually sell at that price. Kind of analogous to the 'present' being just a word to denote the separation of 'the past' and 'the future'.
Similarly the 'total number of coins' is a simplification that takes no account of any coins that might be permanently lost. And it shouldn't need to, because knowledge of the proportion of lost coins should factor into the price. You might imagine that if there is some sudden news that the number of 'lost' bitcoins is vastly higher than previously estimated, this would reduce available supply, and lead to a price increase.
Marketcap is a useful approximation, but no more than that. It shouldn't be considered a perfect barometer of the overall worth of a project.