Lo and Behold, Reveries of the Connected World Page #2

---

So if you had a problem with anybody, you

could look them up, you could find them.

You could find who the actual person was

associated with that email address.

And still today I thumb through that

and a surprising fraction

of the people I actually knew.

For example, there were two

other Danny's on the internet,

and I knew them both. I

still know them both.

Of course now you can't even comprehend

the idea of a directory

that contains the name of everybody.

Today, we couldn't know exactly,

the directory might be some 72 miles thick.

The capacity on the ith channel

should be the traffic on the ith channel

over the speed of the ith channel

plus how much is left over,

that's how much capacity is left over,

and you split it

according to the square root

of the traffic on that channel

over the summation of the

square root over all channels.

The way the internet works,

there's no fixed route

that a message takes.

In the early days of the protocol there was

a kind of a bug and one of the computers

actually had a hardware failure

that made it believe that

it could get a message to some place

in negative time.

So, of course, every

message in the internet

did better by sending it through that

computer because it subtracted the time

net required to send the message.

And so all the messages in the internet

started getting sent through that computer,

which of course got slower

and slower and slower.

So the internet kind of

started to grind to a halt.

The mean response time

now will look like this.

It will be equal to the average path length

times the summation of the square root

of the traffic on the J channel

over the sum of all traffics

summed over all channels squared

over uc (1 minus n-bar rho).

Whatever that equation means, it tells you

what the minimum response time will be

for a network once it's optimized.

The computer was claiming that it could

deliver the message before you even sent it.

So if you had a post office like that

of course you would use it, right?

This was a simplified but

exact model at the time.

Now we have other aspects of it.

But it's basically the underlying

principles of the network,

and one of the things

we found, surprisingly,

was that the larger the network is

the far more efficient it becomes.

Like a gambling casino that certainly makes

money if you have millions of gamblers

at the slot machines?

Very much so.

You've articulated what we call

the law of large numbers.

The law of large numbers says

that a large population of

unpredictable players,

or messages,

collectively behaves

in a very predictable fashion,

a fashion we can write down exactly.

And therefore we can predict the

Rate this script:0.0 / 0 votes