The problem is with the Market application. It's horrible.
That being said, I don't have a top 5 app, not even a top 100 app for that matter, but I do make between 50 and 20 dollars a day from it. Better than nothing, and the trend is upward.
Thanks for showing interest in our project, this is exactly the type of feedback we're after!
Your first and third criticisms are very valid.
The first one can hopefully be mitigated by more groups running open trackers, so that there are more places to DDoS.
As for the second, having a central hub actually gives us a lot of room to control the network structure. The path lengths are _not_ intended to be fixed by any means. However, we wanted to get the bare minimum amount of features in place quickly so we could tie our first beta release in with the 20th anniversary of Tienanmen. With the central tracker, we can control the amount of connections per ip, run blockslists and reshape the network periodically as we see fit.
As for the third point, I don't have an answer for you yet. We're still experimenting to minimize tradeoffs in performance.
Unless things have improved since I tried it a while back, this isn't comparable. The Android thing was a straight port of the 2.5D sprites-on-raycasting engine. And it works well enough, but the G1 trackball isn't nearly sensitive enough to actually be fun.
Carmack's port is using the 3D hardware to do filtering, and he at least claims to be worrying about gameplay and frame rate consistency, something that's a real problem in the Android application.
That being said, I don't have a top 5 app, not even a top 100 app for that matter, but I do make between 50 and 20 dollars a day from it. Better than nothing, and the trend is upward.