From reading the BBC news article: Calls for 'speed-limiting' car
It talks about a call to add voluntary speed limiting devices to cars. Voluntary is good, and it does suggest that there would be the possibility of overriding the device. That allays my concern that you would find accidents occurring precisely because the device would prevent you from adding that burst of speed to get out of an otherwise unavoidable situation - but only a bit. The time you need the burst of speed is the time when you need a split-second reaction - and if you had to disable the device, you've lost that reaction time.
A while ago I had the idea that a speedometer device could be devised that highlighted the speed limit on the display, perhaps by reading a signal broadcast by speed limit signs. I have something like this already on my sat-nav. It has a record of the speed limits in place (exactly the method suggested in the article), and highlights when my speed (as calculated by the satnav) exceeds the stored limit.
There are a few advantages to my method, I believe. 1) It reads the limit directly from the road, so temporary limits would be recognised, and out of date data in the digital record would be reduced. 2) Having the display directly in front of you would make it more obvious. 3) The satnav calculation of speed is not the speed calculated by the car's built-in system. The built-in system is more accurate - the satellite system is not an instantaneous measurement, but is averaged over whatever the sampling rate is, and includes any inaccuracies in the calculation of position.
Of course, there are downsides. If the limits were broadcast (or, more likely, queried by the devices) from the roadside, it would be possible for rogue limit indicators to be placed by mal-intentioned individuals. However, I do not see that as a huge issue - at worst you'd find that it affected only some signals along some roads. The issue could also be ameliorated by combining the data received from the roadside with data from a satnav-like stored database.
It would also mean that an initial implementation cost was required to upgrade current speed limit signs. Again, a database would mean that cost could be spread over a reasonable period - new roadsigns would have the transponder, and the older ones could be fitted in a long-term upgrade.
I do not know exactly how the transponder would work - I imagine it would be an integral part of the sign, rather than a separate device fitted to the sign. My initial concept had been to have the in-car devices read the signs themselves - optical character recognition should be able to cope with the standard lettering. My other thought was having markings on the road to represent the speed limit that were read by the in-car device.
Anyway, I am sure the government will do something that some group finds objectionable...as ever!