You've simplified the problem too much. If there was one central cafeteria and orders were limited to certain neighborhoods with largeish time windows, then maybe, but that is not really what it looks like. You can easily sign up to drive for one of these services and see the reality of it, but be prepared to stand around waiting for peoples' orders for around minimum wage once you factor in all the expenses and dead-heading time.
Frequently there is a section of town with a bunch of takeout restaurants in the same block. And people want to order takeout for lunch around noon and dinner in the early evening. The app could then assign drivers the customers in the same area as one another, if there were more of them.
The people making deliveries are doing unskilled labor. It was never expected to pay well. The question is if it can work at all.
The apps absolutely already do merge any orders they can. They can't control conditions at the restaurant or on the road however and so a lot of the job is really being a flunky and waiting in line and in traffic which are both quite high around lunch and early evening. It scales very poorly and you probably don't want your order sitting in someone's car for half an hour while they wait for another customer's order.
As far as paying well, I don't think that was ever really expected, but these companies just seem like losers for all but those who managed to cash out in the IPO.
