I don't think there's anything hard about computing this, but in the normal case companies report all the money they collect as revenue, and then subtract costs to show their profits.
In the marketplace case this is deceitful because the marketplace owner doesn't really have a way to bring the costs of the product down, so it's much more realistic to only consider their margin as revenue.
In the UberPool case I think there's a reasonable argument to be made that since they are paying drivers a flat fee, but charging users based on dynamic pricing and packing a variable number of people in each vehicle Uber has more flexibility in terms of how it provides the service to customers, the top line number of how much they are charging users is more reasonable. Uber cuts their costs on an UberPool ride in half every time they put two groups in the same car.
You would really want to see the data broken out by category if you were an investor, since mixing the two types in the revenue number is sort of meaningless since we don't know the split, but we're not exactly in a position to demand financials.
Still, I think this data shows a much rosier picture than HN wants to paint about how "Uber is losing money on every ride"/"Selling 2 dollar bills for $1" etc. the 2.8B loss is huge, but when you look at it compared to the $20B bookings number, you can see that they're not significantly subsidizing rides, but rather using their war chest to compete on price. They only need a 15% price increase to become profitable, which shows their prices are in the right ballpark, even if they're not profitable right now.
I do wonder if this is essentially a move to try and compete more effectively for talent that, besides being outraged, may be starting to believe HN about how likely Uber is to fail.