An argument of why they won't buy webOS from HP -- IMO, Amazon shows its focus on recent launch. They didn't release an Android phone, but just an Android tablet to build their empire of content. Hence they don't need differentiations on hardware/OS level, instead they want a hardware/OS which they can: push their cost way down; cheap enough to get people on board; and sell their content.
They may not care about what the user thinks of the OS, but if they do many of these devices, they are going to want a unified OS they control. They built a custom Linux for each Kindle, and now a heavily-modified Android? It's great stuff, but there's a lot to be saved if they can bring it all into the same code base.
One shouldn't spend months in basement coding, instead go and execute something simple, release, get metrics and feedback and learn how to iterate with an aim at product fit. With experience and good examples of launched products, that would get you an amazing team and good track record.
cohort analysis is the best investment for us so far. without cohort, we never know if our iterations are contributing our goals -- usually it means keeping track of retention for identifying product/market fit.
