If I'm not wrong (need a reference) it has been proved that slowing traffic on urban streets in fact raises overall traffic throughput. Am I right here?
On main roads a limit of 50mph rather than 70mph (or, more realistically, 80mph where unenforced by cameras) allows cars to travel closer together, increasing the number of vehicles per mile of road space, and this seems to outweigh the slower speed of travel. It also reduces the "concertina effect" where you can be doing 70mph one minute and parked the next, familiar to anyone who has used the M4 around reading at peak times. In slower urban areas I'm not sure whether the same applies.
If you actually think of a "unit" as being a car plus it's stopping distance, then you can fit x units on a mile of road, providing a throughput of y units per hour at z mph.
Remember: The stopping distance quadruples every time the speed doubles.
Now I accept that not all drivers do honour the stopping distance, but in town at 10mph they do have a much smaller one than on a motorway at 80mph (in general). A bump at 10mph due to no stopping distance causes little long lasting blockage, whereas a bump at 80mph can close three lanes for a few hours.
So, back to the original formula, if you compute all iterations of z (the speed) then you find that as the z decreases x increases due to the smaller stopping distance. Graphing y clearly shows that from zero mph the faster you go the more vehicles can travel through a section of road, until about 15 or 16 mph, at which point the volume starts decreasing again.
To get the maximum number of people from A to B on a motorway they all need to move at that speed (as seen during the bank holiday rush). This also has the advantages that should a bump occur it won't be long before it's cleared.