Suppose I’ve composed yet another brilliantly astute tweet but before I dispatch it to my tens of followers I ask you to guess it. Not just the undoubtedly on-point sentiment but every single word.
I’ll make it easy and just limit it to lowercase letters but I’ll also need some special characters like # and @, quote, full stop, comma and space. So let’s say on top of our 26 lowercase letters we have 6 special characters. As a self-professed maths whizz you’ll know that’s a total of 32 possible characters.
How many guesses, then, do you think would it take to predict every character of my soon to be legendary tweet?
You could start by halving the total number of options available and ask if the first character’s position is greater than 16.
What you’re asking here is whether the character is in the first or second row above.
If the answer was no then you know it’s between the a and p:
Your next question would be to ask if the first character’s position is greater than 8 and the answer might be yes:
You’re getting closer…
Larger than 12? Yes:
Larger than 14? No:
Larger than 13? Yes:
And there you have it. In 5 attempts you have the first character of my tweet. 5 x 140 translates into 700 yes/no questions to successfully guess my tweet.
What we’re doing here is gradually reducing “the multiplicity of equivalents”* or entropy. With each step we decrease uncertainty and increase predictability until we arrive at an indisputable value.
It was at Bell Labs in the 1940’s that avid juggler, unicycle enthusiast and information theory pioneer, Claude Shannon devised this way of maintaining the integrity of information from transmission to receipt. Turning information into units of irreducible, binary digits (that we all now know as ‘bits’) formed the foundations of not only the digital age, from network infrastructure to data science, but information theory as we know it.
If you’re interested in learning more about the revolutionary work at Bell Labs check out Jon Gertner’s ‘The Idea Factory’