As a programmer, I suspect I could probably type the entirity of a days work into the computer in a half hour.

If we count tablets as mobile too (they're wireless after all and plenty of them come with SIM slots) then the consumption part gets a bit better, in landscape mode you can read PDFs on them but the part of work that requires significant input would - for me - not be an option.

Some days, I'd totally agree with you; I'm not really sure of the next step(s), and have to give myself lots of time to think about things. This tends to apply when I'm entering unknown territory, and my tasks are relatively fuzzy and uncertain.

On the other hand, some tasks are extremely straight-forward (repetitive / memory-based), and more or less completely WPM and flow-bound. Even working with relatively efficient editors (using shortcuts, macros, VI bindings, etc), it's hard to type out (or otherwise input) much more than 2000 lines of code in a day. These types of tasks certainly require efficient input, and could be greatly enhanced by even better human-to-machine interfaces.

(I'm also a programmer, and these do come from my own experiences)

A lot of modern languages/platforms are able to be used as a REPL console. Same with commandline tools.

Being able to rapidly experiment with bits of code to identify the correct solution can be incredibly valuable.

I don't care if it's a bottle neck. I want an efficient means of typing, so that I can keep my focus on other stuff. Maybe I could do a lot of my typing using a terrible interface like my phone, but it would be very aggravating.

I never understand this prediction. That's a bit like saying I don't need a car because I could just dock my bicycle into some sort of enclosure with four wheels. Tablets, smartphones, and laptops/desktops were all built for different purposes and cannot be full replacements for each other, just like a bicycle can't fully replace a car without some serious sacrifices.

At this point, if you have anything worth securing, its probably a good bet that your device will get compromised in the next 5 years. Compartmentalized devices helps with that significantly, since it means only partial compromises.

What if an iPad Air could simple attach to a MB Air chassis and only serve as the display when attached?

Android is a bit better than iOS in this respect, but not much.

None of the mobile vendors have any incentive to change this, since it would mean forfeiting the App Store tax and for Apple would cannibalize the Mac market. The only way I see an uncrippled mobile device entering the market that is high enough quality to compete is if someone with none of these conflicting interests bucks the trend. Android is pretty forkable, so a droid fork that solved the security problems in a non-feudal way and that supported the sort of docking you describe would be disruptive.

Dell? Compaq? HP? A "washed up" PC vendor with stagnant market share would have nothing to lose and might have the resources to pull it off.

The App Store "tax"? Sure, Apple doesn't mind the cash. But they are first, second and third a hardware company: that's where the real money is. The reason they have no intention to allow side loading apps on iOS has to do with user experience, eliminating support headaches and security (order may be different, but these re the reasons).

The fact is your dream device would appeal to the same people who buy desktop Linux machines now. They exist, but they are a tiny part of the market. Nobody can stay in business catering to just those customers.

In practice this means that PCs and their unlocked OSes will continue to hold onto their market niche until or unless mobile bridges that cap.

In 5 or 10 years dockable tablets are going to be every-goddamn-where, especially in business. It just makes sense, and is too all around practical. And for most computer uses, even "intensive" ones, it's perfect. You get portability plus productivity in the docked configuration plus huge economic benefits. Tablets are mostly just screens, batteries, and a handful of chips, all of which are super amenable to economies of scale in manufacture. Tablets are going to be cheaper than dirt eventually, and because a tablet can be a self-contained computer it'll tend to be the default computing choice. The biggest thing missing today is primarily good software.

I would instead focus on the work done with virtual machines. If instead I had a subscription service to access a virtual machine that had the ability to run any application I wanted, streamed to my mobile device that would then display it anywhere I wanted. My mobile device could connect me to any amount of computing power I need (in reason and with a large enough budget).

Why on earth would I buy this whole separate machine to do this? My personal computing device that I carry around with me everywhere could allow me to perform any function possible, I could have a full desktop computer anywhere I wanted as long as I have a internet connection and a screen.

Gaming could take place anywhere as well. You wouldn't need a gaming rig, the processing power would be handled elsewhere whilst your device handles the graphics processing and streaming.

Internet speeds will have to increase exponentially, but are we really that short sighted to state that personal computers will never be replaced by mobile devices? Yes it may not happen tomorrow, but it will come

This is why I find machine learning and optimization such fascinating areas to watch. At a certain point, we may reach the practical limits of what human gestures, commands, and requests can tap into or do. The machine (or rather, the distributed ecosystem of machines) becomes more and more important in automating X, suggesting Y, and predicting Z.

I enjoy the prospects of VR and AR, especially in an omniconnected world. Those always seem like interesting use cases for a smart(er) phone. But I'm a lot more excited about the non-UI advances that the "internet of things" can bring us. When we free ourselves from the limitations of human comprehension, human attention span, and human neurological heuristics, we can do so much more. To me, the "large untapped areas" are all the things we won't have to tap to access (terrible pun intended). Before we can get there, of course, we'll have to connect all the devices.

At the risk of sounding hokey, naive, or unapologetically futurist, I look forward to the day when kids will say, "Wow. When you were my age, you actually had to touch things to make them work?"

Obviously flesh blood is a good thing, however, this problem space has been thoroughly explored.

The real unexplored area is that few people have noticed just how insanely powerful the GPUs in these devices are, but again the problem is in working out what they might be useful for, especially given the trend is for mobile "apps" to really be trivial front ends for web services.

Yep. After doing mobile for a while you start to notice that most apps are just listviews hitting REST endpoints. Not exactly earth-shattering technology.

Keyboard for text entry coding, digitizers for design/art/photo-work and possibly (multi)touch for richer, Smalltalk-like UIs. I've always thought three-button mouse weren't such a great idea (ergonomically) -- but a lot of the same things that work well with them (Smalltalk, ACME), should work fine with multi-touch -- as long as we evolve the GUIs a bit to take proper advantage.

I have had an iPad since the first version. I also have had various touchscreen "tablet" PCs prior. The problem is not the hardware now its the software -- and the software has gotten so, so, so much better. I can do real work from a tablet and a phone. The best practices devs first discovered and now proliferating more widely. That takes time but we are seeing the results now.