[QUOTE=Shamozzle]
I think you have the right idea and as such it is an impossible question.
When is a path or a game trail a road?
If a road hasn’t been used in 500 years and is almost overgrown, is it still a road?
When two roads gradually converge but don’t ever quite touch, is it still two roads? What distance between them would decide their convergence?
When nomads travel across the Sahara, are they on a road?
[/QUOTE]
This kind of stuff is exactly why I was expecting no more than an order of magnitude estimation. I actually own some property on a non-existent road, at the corner of Cactus and Cactus. Given to me by my father, bought in not one of his most skeptical periods.
[QUOTE=Mycroft H.]
With a quick Google search I found this site. (37 kB pdf) It gives the information for 7 countries, but unfortunately it is for 1993. I’ll keep looking and see if I find anything more recent.
Summary:
US - 3,905,000 total miles, all roads (rounded figures - does not defined a “road”)
[/QUOTE]
I found this also - but how to translate miles into roads is beyond me.
[QUOTE=Ludovic]
The answer is blowing in the wind.
[/QUOTE]
How many electrons
Must go down a path
Before they make a gate switch
…
I think I can turn this into a column. I did one once on a Windows manual as written by Hemingway, Shakespeare, Frost, and a Star Trek screenwriter (The Computer on the Edge of Working Never.)
However… (surely you knew I was going to add an “however” )
In addition to the USPS roads, in our routing and mapping processes, we add all non-USPS roads that match your criteria above: Access roads, alleys, Farm-to-Market roads, Forest Service roads, etc. (This way we can guarantee our clients the most complete coverage possible. Or at least make that claim in our proposals. )
Looking at the datasets for all of our deliveries conducted since January 2006 I note:
2,195,348 USPS street segments.
117,934 “additions” (in our companies parlance)
For a non-USPS road rate of 5.4% (rounded) of USPS roads.
[/QUOTE]
Excellent. My guess was correct. The latest processors have about a billion transistors. If 2/3 of those go into embedded cache, which is a reasonable guess, and there are say five transistors on average per gate (flops have more, inverters have fewer) and there is roughly one signal line per gate (very conservative) that makes roughly 80 million signal paths. I can get an exact number for the processors my company designs, but then I’d have to get it cleared. That’s plenty good enough for a proposal.
)