Wow. So that’s almost 200 Tbits/sec. It looks like the current fastest cables are on the order of a few tens of Tbits/sec.
Just for fun, how would today’s shipping industry fare? A large container ship might carry 12,000 containers and cruise at 25 knots (~28 mph). Say 10,000 lbs per container, that’s 60,000 tons, or 75 million drives with 300 exabytes (that’s 10^18, quintillion) of data. Moving at 28 mph it will cover 4,000 miles in 6 days. That’s…
580 terabytes per second. Or 4.64 petabits per second.
It took some of the greatest minds of 19th century mathematics to fully elucidate why the first transatlantic cable worked so badly. In particular, I have heard that Riemann’s theory of the characteristics of partial differential equations was motivated by this problem.
The key, as some have alluded to above, is the recognition of the importance of resistive loss in a system that, at low frequencies, is largely capacitive. If you think about a network of series resistors periodically shunted to ground by capacitors, you’ll realize that the signal does not propagate at a fixed speed. The first capacitor charges exponentially in time, generating a voltage that charges the next capacitor, and so on and so on.
If you analyze this, you find that the signal essentially diffuses from one end of the line to the other, rather than propagating. It is like the flow of heat. If you double the distance, the time goes up by a factor of four, rather than a factor of two. It is like trying to communicate to the inside of a turkey by pulsing the temperature of the oven up and down. If you are deep within the turkey, this is a very slow way to communicate, and if you pulse the temperature too quickly, the signal will be averaged out.
Since the transmission line highly distorts the signal, you can’t just transmit quickly and wait a long time for the signal to reach England. You have to transmit each dot or dash extremely slowly, otherwise the symbols will overlap at the recieving end (intersymbol interference). It took the genius of Oliver Heaviside to figure this out and recommend the introduction of coils periodically (loading coils) to increase the inductance per unit length, thus allowing the system to actually propagate the signals at a constant velocity, rather than making the signals diffuse across the ocean. People thought he was nuts, because everyone knew that inductance was a bad thing, and could only make matters worse.