Isn't there a write up somewhere? Wouldn't it be better to link to a write up? I don't want to spend 4m19s watching some dumb video with sound in space and fancy graphics. Spoken narrative is too slow. A write up and a diagram or 2 is enough to convey principles, which is what interests me.

I am guessing, but I think it was a long time in preparation. With the UK broad gauge, the difference in gauges was large enough that track could be laid in mixed gauge, so the actual changeover could be years of dual gauge operation. For the Southern US, the gauges were too close to permit mixed gauge, I would think. An easy way to do a quick change would be to spike the sleepers [US = ties] on one side only left and right on alternate sleepers when laying track to the broader gauge. Gauge would be held by stretcher bars every few sleepers. On changeover day, the stretcher bars could be taken out and the left and right rails complete with sleepers slewed together and the unspiked sides of the sleepers spiked to the new gauge.

The Oklo natural nuclear reactors http://en.wikipedia.org/wiki/Natural_nuclear_fission_reactor were discovered on the basis of isotope ratio deficiency

But Shakespeare has well defined meaning to a considerable level of sophistication. The way to do it is to extend the language logically from what is already available and where this is not enough, invent more Klingon to express the concepts. Done well, this project has great potential to develop the Klingon language into something fully functional. Not that there is much point, of course.

So it is now only the modern English who do not have Shakespeare in their own language.

NTP has its limitations and to me the miracle is that it is as good as it is, given the lack of promises concerning packet delivery in a TCP/IP network [No guarantee on route, latency or even delivery]. If you need something more accurate, then use GPS as already suggested or one of the atomic clock backed radio services around the world and calibrate for transmission distances, which is a consistent correction for any location. This will take accuracy close to the theoretical limits and is potentially available cheaply enough that the cost should not be a bar to any application where this sort of accuracy really is justified. I just cannot see any point whatsoever in trying to do this over the internet when you cannot hope to meet the accuracy of existing means.