Good example.
I know of some large complex enterprise-wide mainframe systems developed in the 1980s at a cost of a few million dollars (the former "Big Eight" consulting firms made out great). Anyway, these systems are under consideration for redevelopment away from "green on glass" to GUI to make for an easier user interface and additional functions.
Estimates to rebuild these systems to do more are coming at less than the original development price, and that ignores the big inflation of the 1980s. What this means is that computer development productivity has increased tremendously since 1980.
It's not hard to see why, so many things are "canned" instead of "roll your own" and developer's tools are much more powerful and user friendly. Computer hardware is so much cheaper that programmers don't have to worry about bits and bytes as they did back then. (For example, mainframers generally need not worry about packed-decimal or binary fields as once was mandatory.) An enterprise wide mainframe in
1980 could have as little as 8 megabytes of "RAM" and maybe 1 gigabyte of disk. Today a wrist-watch has that (well, not quite, but you see what I mean.) Likewise with communications, in 1980 you were lucky to be using one shared 9600 line and you had to keep your data transmission packets small to avoid flooding out the line. Developers don't have to spend time squeezing stuff in as they did in the past.So yes, redoing the Internet network won't be an easy job, but it certainly won't be the sum of the parts that created through now. Geez, back in ARPANET (?) days someone had to first fill first out paperwork before making a long distance phone call to his counterpart, they don't have to bother with that day.
Further, the Internet consists of various component, for example, email and WWW. They could be redeveloped independently of each other. When rebuilding a highway, they don't shut the old one down and rebuild it all at once, they do it in stages so traffic can keep on flowing.
Actually, it does. Scientific theories don't become "facts" until an idea is published and debated among scientists, who test out the theories in their own labs and will debate them. Perhaps it is not "public" debate, but an open debate is vital to science.
There is a big gray area between "science" and "art". Certain scientific principles can be proven and yield predictable consistent results every time. But other scientifiic principles yield only statistical probability. That is, in 100 experiments, a certain result is expected to appear so many times, but an any individual experiment can't be predicted. That can have great impact depending on the application, for example, medicine.