The core software itself doesn't seem too hard to build, except that some real time features like the ETA are clearly only possible if you have massive amounts of data. Yet a simple version where orders are dispatched to a pool of drivers should be relatively easy to put together.
Restaurants can offer items for cheaper than in-store prices because online orders don't have real estate overheads (except for maybe the kitchen), and our fees will be very low / a flat rate. Cheaper food should incentivize users to accept a lower quality UX.
Drivers can be paid more than Uber eats because a small open source project should not have the same insane R&D overheads that uber has.
The only thing I can think of is that if you get big enough, Uber eats / doordash will either try to stomp you out or buy you.
