Count me in... however because the future of Construct 2 and Ouya is so damned iffy, I've just about convinced myself to go the route of Construct Classic and make a Direct X9 game for Windows instead. That'd be "good enough" as the .exe exporter for Construct 2, from what I've read, is unpredictable enough to NOT be reliable for commercial release.
Plus, Construct Classic is free and open source? Maybe the route to go about all of this is to somehow dig into the guts of Construct Classic and make IT support Ouya somehow, seeing as it has nothing to do with HTML5 whatsoever.
Very open to discussion on that subject, it requires knowledge I don't have.
I figure I'll take a little more time to answer your question. I didn't want get into a performance discussion here, but I like to talk about the subjects that interest me and this one for me is a hot topic. So I will answer it :)
First let's just accept that right now HTML5 is not up to par for mobile performance compared to native. let's end that right here. This however does not exclude the ability to make beutiful and deep game experience than just simple puzzles. In fact some old classic games can be readily done on common Android devices now with HTML5.
Classic SNES games: youtube.com/watch
Many SNES games on the list are very deep, but no where graphically complex. I know it sounds disheartening to be compared to an older console, but there is a brighter side. These games had less performance for the most part than what we can currently get on the mobile. All we have to do is take advantage of the technology and treat it differently than the PC. We can't brute force are development based on PC design standards... sadly :(
I love this game. But remove background for now. remove the OpenGL effects(ok DX) and the game really is very small on the object count. not only that the game will zoom in at smaller rooms cutting out larger environments and allow for minute details. Brilliant really. I could see C2 on Tegra 3 pulling off such a game. Then just add a simpler parralax background with less object count.
Right now with smart game design we can still make deep, fun, addictive, bueatiful(if you have the artist) games for the Ouya. We just have to be clever about what limited resources we have.
As for CC, it will be far more work to convert DX to OpenGL. However it is an option. C++ can be bound to Android JNI port in about a week providing you have the source. I've never played with CC ever.
In the mean time maybe we can look into creating a few graphical tricks :) One blazing performance trick in 3D is billboarding a 3d model. this often deals with rendering a 3d model to a 2d sprite. Then display the 2d sprite. Only re-render the 3d model again when there is a required change. This can boost performance massively. back when I used to hang around jMonkey someone did a test and went from 120fps to 1000+fps. Without requiring to render the animation every frame it left a lot of time to do other work.
Well we are working in 2d, but we can still take advantage of the same trick. We can group objects to a canvas and only update the canvas on a 30 fps or less scale. This can let us keep logic performance closer to 60fps, but keep animation smooth at 30fps and increase performance. This performance increase occurs from the benefit of not having to blit every object every frame. I haven't implement this as of yet, and likely won't for my current game; as it is for me to far in developement. However I can see my next game experimenting with this.
Also keep in mind all my testing is on a low end Tegra 2 where as the Ouya is running the top end Tegra 3. It is likely closer to 10 times the performance. Keeping this in mind. My simple ball test I posted in the "sad thruth thread" I managed 86 balls at 40fps in the Android stock browser finding that letterbox scale integer did something interesting. With the Tegra 3 we shuold be clear to make some really good games. Still no special effects, but good gorgues games none the less :)