Here are a few very minor little ideas for how to improve various technologies. I’ll update it as I think of more!
Clocks
Gadgets which display the time, should almost always display it, unless there’s a very good reason not to. For example, Windows and OSX are good about displaying a clock at most times, but not during their login screens. There’s no reason not to display the time on the login screen. (An example of a good reason to suppress the time would be if the user was playing a game which took up the entire screen)
Again, cell phones. In a perfect world, a cell phone should double as a fully functional pocket watch. And yet, some phones suppress the time if an alert is active. And some phones won’t display time at all if they’re out of their service area. I guess the idea is they can’t figure out what timezone they’re in, but they could at least display the time based on the last known time zone. (If the programmer is that paranoid about changing time zones while out of service, put a question mark next to the time)
YouTube, Media Player, etc.
In virtually all media players, if you want to change what time you’re at, you have to slide a slider on a fixed-length horizontal bar. This is a very imprecise way to control what point you’re watching/listening to. In addition to the sliding bar, make the time display (the one which reads something like “1:22 / 4:37″) be editable. If the user wants to jump to exactly 2:05, let them click where it says “1:22″ and type in “2:05″.
Captcha Text
Captcha Text is that thing in every form you fill out where you have to verify you’re human by typing some squiggly letters into an input box. Brilliant idea, a bane to spambots everywhere. Here’s the thing. If you enter something else wrong in the form (like you choose an existing username or your “confirm password” doesn’t match), you’ll typically have to enter the captcha again after fixing the form. There’s no point to having to refill a captcha a second time. If you were able to fill it in correctly the first time, you’re either human, or such a well-programmed bot that the 2nd one isn’t gonna make a difference anyway.
Monitors
Laptop monitors, and any other monitor integrated with a computer, should take a lesson from classic desktop monitors: an off/on switch. It should be possible to turn the monitor off without turning the computer off. With a laptop, it’s usually not even enough to load a black-screen screensaver: the monitor will still emit light.
Shorter URLs
Many (I’d go so far as to say almost all) URLs contain unnecessary subdirectories, a lingering remnant of the very early days when the typical website was a static directory of real files. Take the English Wikipedia, for example. The URL for their “technology” article is: http://en.wikipedia.org/wiki/Technology. There is no reason for the “/wiki/” subdirectory in there. It would be better if the URL were simply: http://en.wikipedia.org/Technology. Probably better for search engine optimization, too (but that could just be superstition).
UPDATE: From the email, Gwern pointed out a couple reasons for longer URLs. See more details at (1) below.
ATM Machines
Maybe there are some privacy concerns involved… but it would be kind of cool if an ATM machine remembered customers’ choices from previous uses, and offered a shortcut key to repeat them. For example, say I go to an ATM machine and make a $200 withdrawal. In doing that, I have to select “withdraw”, then “checking account”, then “type another amount”, then finally type in “200″ (actually I have to type in “20000″, since it goes “$.02″, “$.20″, “$2.00″, “$20.00″ and finally “$200.00″– which is a whole other thing in itself, since we can’t withdraw in penny-granularity anyway!) Then I come back a few days later to the same ATM. This time, after entering my PIN, there ought to be a “Withdraw $200 from Checking” button right at the beginning. 
UPDATE: sik0fewl pointed out in the forums that there are Diebold ATMs in Canada, used by TD bank, which have exactly this feature, called “My Favourite Withdrawal”. Sweet, I feel like one of those mathematicians who independently discovered a great theorem months after a different mathematician.
Auto-Form Fill
Modern browsers like Firefox are set up to remember things you type into forms, and when you start typing in a new form, a list of previous things appears that you can choose from. Unfortunately, at least in Firefox, there’s no intuitive way to remove things permanently from the list. Over long periods of time, some of these lists can fill up with quite a lot of junk. It would be best if you could highlight a choice on the list, and then right click, and a right-click menu would show up with some options like “Edit”, “Delete”.
UPDATE: sik0fewl pointed out that this is already possible in Firefox. Highlight the option you want to delete, and press the delete key. Wow, I wish I’d known that like a year ago The tech improvement idea is officially revised: the feature exists, but please make it more intuitive 
I will update this in the future as I think of new things, and/or as my ideas are integrated into the technology of the world
NOTES
(1) The first reason to use longer URLs is, an URL system with no subdirectories might conflict with things like robots.txt, sitemap.xml, etc.: common files which live in the root directory and have a very specific purpose & protocol. For example, Wikipedia has an article on “robots.txt”, but also has its own robots.txt. The two should obviously be kept separate. The second reason is the danger of setting up the more advanced, shorter system on one webhost, and then being surprised by a new webhost which doesn’t give the necessary .htaccess privileges or whatever.
Great points. Programmers should carefully make sure no “robots.txt” issues arise. It’s not a problem in WordPress: wordpress url names end with “/”, so a mischievous blogger could create an article with URL “mysite.com/robots.txt/” but this would be distinguished by the “/”. Wikipedia uses uppercase for filenames, and that would suffice to distinguish its real “robots.txt” from its article “Robots.txt”. Even if an URL system was so generalized as to lack even this much structure, programmers could still hardcode necessary URLs as “illegal”. (Incidentally, for all the care Wikipedia takes with ensuring people can look up obscure things like “Robots.txt”, you still can’t look up “#”. Go ahead, try it, the results are seriously broken, especially if you click “Search” instead of “Go”. And this flaw is non-trivial over at the English Wiktionary.)
As for the restrictive servers, I don’t think that’s such a big problem. Running Wikimedia requires PHP and a database, both of which are far, far bigger privileges than .htaccess. Besides, if a giant project like English Wikipedia is forced to move to such a restrictive server, I think URLs will be the least of our troubles
FURTHER READING
How to Improve Blogspot
Reinventing the Wheel
The Halting Problem