Relay computerswhy so few 21

Eric Sosman

A similar test was staged just a couple years ago, when the American rail service Amtrak introduced high-speed "Acela" trains on the Washington-Philadelphia-New York-Boston route. ("High-speed" by American standards; "pokey" elsewhere.) A Boston newspaper sent two reporters to New York for the inaugural trip, one to return on the new train and the other to race against him on the city-to-city air shuttle. The air traveller won the race, arriving at the newspaper office back in Boston before his colleague -- but, IIRC, only a few minutes before, the victory margin being less than the normal variation one would expect from ground traffic in a city. Train and plane showed the same latency, to the precision of the measurement. ----snipped----

When doing this so-called speed tests between humans and computers, or computers vs. abacuses, what sometimes causes the "win" is the slowness of entering the number.

Asking the square root of 50 (to ten decimal places), the computer will win easily. Now, ask the computer to add five and centillion, and a human will instantly come back and say centillion five. It takes longer for anybody to just enter the two numbers, and even then, the operator (person entering the number) would have to know what a centillion IS.

By the way, it's

1000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000

(or, if you can't read all of that on your newsreader, it's a 1 followed by six hundred zeroes).

Also, ask a computer to add ten to a googolplex. I know what the answer is (googolplex ten), but the number can't be expressed as a "regular" number --- that number (or for that matter, a googol) exceeds the number of atoms in the universe. Gerard S.