1) Users don't care 2) It's complex, so it costs quality developer time. Even then there's no guarantees. 3) The old school approach is cheap/easy/fast. Addit

1) Users don't care 2) It's complex, so it costs quality developer time. Even then there's no guarantees. 3) The old school approach is cheap/easy/fast. Additionally enabled by fast/quality clouds that allow scaling relatively easily. 4) Increasing number of clients do not accept incoming connections because of IPv4/Nat or because of cellular/mobile/wan connections that don't accept incoming. 5) It's a competitive market for apps, web services, and related. Any increase in latency or increase in support calls is prohibitive. If even one and 10 people need support for port forwarding that's a deal killer. 6) in an increasingly mobile world incoming connections and anything that hogs battery (bandwidth or even just being awake) is a disadvantage. 7) there's no easy money in P2P. No monthly data plans, centralized towers, centralized servers, etc. Sure a mesh network of smartphones with millions of clients could do cool things. But where does AT&T/Verizon make money? Without AT&T/Verizon is Samsung going to make a p2p/mesh network phone if they need to sell millions to break even? As an example skype years ago with mostly desktop/laptop clients was largely p2p (just login was centralized). With increasing numbers of tablets, WAN connections, and smartphones they switched to central servers. So sure, you might be able to spend a man year and get an awesome, robust, and performant solution. But your competition will have spent that time actually making users happier and steal your market.