I like my MBP for this as well, but the writing is on the wall. The Mac App Store is entirely optional
now but I don't have any reason to believe it will continue to be that way. The App Store on iOS is
not optional and Apple sees some benefit over Android because of that (well they define it as qualities), installing from 'outside' the Sandbox has only been getting worse.
So how about a development environment for Android in the 'cloud' ? How cool would that be, your files are in iCloud so you know they are safe, your tools are always current and compatible since you're just talking to them over web API, and you can pay for a months worth of access to the compilers and debuggers for cheap instead of wasting all that time to put together an open source environment or spending big bucks for the MSDN Library equivalent service. You can watch movies on your MBP while you are building the Android manifest and running regression tests on that thing in some cloud instance that spun up just for that one task and will go back to idle when its done. Click a button and blam! it's submitted to the App Store.
You may not realize it but that is the vision that is driving moving away from a computer that you hold and do things with, vs a 'viewer' into a service somewhere that is doing the lifting and "adding value" in other ways.
If you read the thread on the Lisp epiphany [1] then this is the same kind of thing. But rather than data is code and code is data the epiphany here is that if you have enough network bandwidth and its available 24/7, then it becomes just another bus on the backplane and where your "computer" is and where your "screen" and "keyboard" are becomes irrelevant. At the office we have folks who put their desktop computer in some room, out of the way, and using X have their 'screens and keyboard' on their actual desk, they do this because the desktop machine is modestly noisy (we have quite cases but its not silent) and really on a gigabit LAN you can't tell that your machine isn't under your desk anyway.
That is coming to the world, faster than you expect, and because it will be insanely more convenient for most people supporting the few people who want to do this 'locally' will fall by the wayside.
Step 1, you figure out how to present a interaction rich UX
Step 2, you figure out how to move most of the data into the network.
Step 3, you create tools that present the UX, process data in the network, and deliver the result.
Google Docs is crushing Microsoft office, services like Box.net, DropBox, and S3 are capturing more and more of the data.
Free startup idea: Github + APT, add a button to a GitHub like service that says "Build this for Ubuntu-x86-64-12.10 and give me the link to the package." Click the button, wait, then post the link to a modified APT
apt-get install <URL>
All of the pieces for this exist, a couple of weekends coding can put it out into the world. Now if the packages can be built in the cloud, why does your machine need to be a 'computer' again? Build it then 'add this feature with apt' is a pretty powerful thing.
Probably won't happen for a couple of years at least but it will happen, too many people want it too, and too few will demand local compilation to keep it supported.
[1] http://news.ycombinator.com/item?id=4765067 - "The Nature of Lisp"