If I wanted to pick up packages, I'd go to stores and buy things. I don't want to go to the store though, so I pay Amazon to deliver it to me.
What it comes down to for me is, I don't hate going to the store, I hate shopping at the store. Typing "16GB USB Drive" into the search box is much easier than wandering down multiple aisles trying to figure out whether that particular store put the USB sticks with the laptop accessories, the camera storage, the office supplies, on some unrelated endcap, or behind the guy standing inconveniently in front of the display I'm trying to look at.
And don't forget, the costs can be much lower when using parcel machines. For example in latvia ,it costs 2 euro to send a package from a postal machine to a postal machine, versus 3.6 euro from a postal machine to someone via courier.
And e-commerce users really hate to pay for shipping.
Based on store counts times conservative numbers on the total cost of a new Wal-Mart. They might try building a bunch of small stores, but they're definitely not going to try to become Wal-Mart.
Apple's flagship Manhattan store famously does quite well as a showcase for their products and helps to sell millions for them. Amazon could be doing that math that for every user we get into the ecosystem we can extract X in lifetime value from book, video and app purchases (again a way it could make sense).
Also, you can consider this maybe Amazon trying to become Walmart faster than Walmart can become Amazon. Having physical stores still counts for much in the retail sector, having one (or a bunch) would potentially give them leverage and more options. They _already_ have a massive nationwide system of warehouses and logistic centers, is it that crazy that they could move products into brick and mortar shops as well as the presumably much more complicated process of people's homes?