First, let's have a unification of user interfaces. As it stands right now, novices find it incredibly challenging to tell when to left click, when to right click, when to click once, and when to double-click. They can't tell the difference between their "desktop" and their "web browser", and if you step back and think about it for a moment, it doesn't make any sense that they should have to.

I would also like to see the notion of everything in a computer being a metaphor for something in real life come to a blessed end. There's no reason that computers need to have a "desktop", and "files" and "folders" don't make much sense to novice users. Most of them are totally incapable of organizing their information in a useful way, and inconsistencies with file save and open dialogs don't help this. I often hear from people who just need help finding the file that they know they saved, but can't find on their computer. I've also had to reconcile vast hierarchies of folders for users that had been saving different versions of the same file to different locations.

I would like to see a new internet-distributed file system, where data is separated into regular chunks, and then those chunks are saved in multiple locations around the internet in a fast rootless node structure. Public chunks are unencrypted; private chunks are encrypted. To access all of your information from anywhere in the world, you simply sign in to a portal from any computer; your login decrypts a small chunk file which contains encrypted references to all the rest of your data. This would make the very idea of a "backup" completely obsolete and would solve data portability and storage issues for anyone with a broadband internet connection. It would also -- at least for a while -- completely halt viruses and malware.

I want to see consumer devices become more upgradeable and more modular. At least once a week I have to explain to a customer that their entire motherboard (or, often, laptop) needs to be expensively replaced, because the DC circuit failed, or a graphics chip overheated (thankyouverymuch HP).

I think there needs to be a serious effort to upgrade the communications infrastructure in the U.S.; I'm aware of the challenges presented by the geography in this country and current and past building practices. However, much of this build-out has already been paid for [1]. Instead, customers find themselves having to call tech support every time they think their email has stopped working, only to be told that their computer is currently in the process of downloading a 10MB attachment from someone.

I believe that there needs to be a much greater importance placed on performance in software. I think that the current commonly-accepted principles in software development -- ship early, ship often, and hardware is cheap so don't spend too much time making it fast or small -- is wrong-headed, and I think that's obvious to anyone who actually interacts with their customers on a regular basis. The fact that products like McAfee and Norton can have such massive impacts on system performance that the customer is left wondering what died and went to hell in their computer is a problem that needs to be addressed.

This is just for starters. I could go on like this for a long time. I think that all new construction should be wired up for gigabit, right alongside phone & power; I'd like to see cars with upgradeable powerplants; etc.

It's not that I think that current technology isn't improving, or that it's bad necessarily, but I do think there are many problems that it presents that its developers really aren't even aware of, or that they care to address. We keep getting more and more time sinks in the form of shiny new "social" networks where less and less of substance is shared in each iteration, while basic principles of design and infrastructure continue to languish in the shadows.

[1]: http://www.newnetworks.com/BroadbandScandalIntro.htm -- not the very best reference, but it's getting harder to dig this story up anymore.