This idea - both the problem and the solution - literally just came to me in a dream.
You hook up a car with a computer, a GPS, always-on Internet, and a touchscreen. You have a database of dangerous spots - areas where accidents are more statistically likely. In such areas, your car alerts you. So you have the equivalent of the normal warning signs you see when driving, except these are linked to real information about what can kill you, instead of the preferred behavior lawmakers, locals, and law enforcement want from you.
Sometimes that behavior is completely abritrary and meaningless; sometimes they want you to drive slow when you'd get away with driving fast. Frequently people assume those signs are only for bad drivers, and exempt themselves from the category (inaccurately). Consequently lots of people disregard warning signs on the roads as noise. However, if you're entering an area where lethal crashes are in reality much more frequent than average, this is news you can use. Especially since there are also plenty of areas with inadequate or inaccurate signage due to incompetent local governments.
Reality-based context-sensitive in-car signage solves this problem. For version 2.0, you wire the computer into your car's built-in computer, obtain MPH, MPG, all available data, and upload it - along with times and places of any accidents you have. You can then run all kinds of sophisticated statistical analyses on the data to obtain much more accurate crash prediction and correspondingly much more useful warnings.
Probably all road data will operate this way in the distant, shiny, jetpack-filled future, with physical road and parking signs becoming a thing of the past, but it's pretty cool to be able to do it here and now in real life. The only missing piece is the database; everything else on the list is already possible. Damien Stolarz made consumer in-car computers with touchscreens a commodity, monitoring your car's internals in real-time is trivial in most cases, always-on Internet is solved for major metropolitian areas, and you can buy a USB GPS unit for less than $100.
Basically, in the same way that Wikipedia matches Encyclopedia Brittanica in quality but beats it in cost, timeliness, and range of subjects, leveraging cheap commodity hardware, open-source software, and the magic third ingredient - self-organizing groups of people - allows us to provide a digital alternative to road signs which is more accurate, more relevant, more up-to-date, and less polluted with beauraucratic foolishness.