Keyboard with E-Ink keys. The keys can change based on the language you are typing in, and can change to icons in different applications, like Photoshop. The screen technology is inexpensive but there needs to be a reliable way of connecting the keys to the PCB. First Steps: solve the connection problem, make a prototype.
Touch sensing keyboard keys that can be used for gesture recognition. Basically, each key has a low resolution touch sensor in the top of the key cap. User can swipe with the tips of their fingers across the keyboard to perform gestures, such as positioning the cursor. If placing the touch sensors in all the keys is too complicated and costly, than placing a sensor in the space bar should do the trick. Swiping your thumb left and right across the space bar would cause the cursor to move back and forth. First Steps: try this with a prototype with the sensor in the space bar, check patentability.
Keyboard with built in headphone jack and BT or USB connection. This one is simple, and it's annoying me right now. I am working on my desktop right now, and I wish I could plug my headphones in, but my desktop is on the floor under the desk. A keyboard with built in BT headphone connection or USB connection would be perfect. First Steps: make a prototype using USB AMP and a keyboard with a USB HUB. Do market research to see if people want this.
Ok, so since the problem described above is annoying me right this second, option two. An iPhone (smartphone) app and a companion app for the computer that allows you to turn your smartphone into basically a wireless speaker or AMP. Problem is latency. First Steps: research to see if this could be done with an acceptable latency, meaning less than 100ms.
Chrome plugin that allows you to clip the we into a single front page. The plugin would allow the user to select an element (usually DIV) on any given page. That selected element will be snapshotted and added to the "start page." The idea is that you can have one start page that is constantly updated with snapshots from all kinds of sites you would normally check. The snapshots would be automatically updated every so many minutes. Clicking on one of the snapshots would open that page. First Steps: see if this exists and if it's possible to do this in Chrome.
Open source chrome plugin that allows you to click its button which would cause the plugin to submit current web page's URL to a remote server. The idea is to use it in different other projects. I wanted to use it for my YouTube download server that takes videos and adds them to a podcast on my local server. First Steps: see if this exists, if it's possible to do (this could be blacked by Chrome for security reasons), write this and post it on GitHub if it doesn't exist.
A film adaptation of the Deathworld by Harry Harrison. Seriously, I have no idea how this book has not been made into a movie yet. It's got everything, a likable protagonist, a one with nature type of message so popular in Hollywood, and for some reason it appears to be in the public domain. First Steps: write a screenplay based on the book and pitch it.
Thermoelectric heat pump that cools the air before it enters a computer case to increase CPU performance and reliability. This is something I actually wanted to do with one of my computers. Either do a water cooling system with cooler, or an air cooled system that injects cold air. I like air cooling in general, because cold air circulating also cools the other components, such as caps and memory. First Steps: make a water cooling prototype. This can even be made into a space heater. "Keep your PC cold and your room warm!"
As long as we are on the subject of cooling, it makes a ton of sense to run data centers next to the ocean or on large cargo barges in the ocean for cooling purposes. The ocean water can be circulated through heat exchangers to cool down the cooling liquid. First Steps: see what people in the industry think about this, never really researched this.
A website where people can post common problems in their industry and other people can propose good (and probably bad) solutions to those problems. I have noticed, especially talking to some industry people, that in many cases a good solution is just a matter of looking at a problem differently. Usually the really great and revolutionary ideas occur at the intersection of different disciplines. First Steps: see if a site like this already exists, if not, make it.