,
10 tweets,
3 min read
Read on Twitter
1/ The diction used by these African drummers to communicate across vast distances is a perfect example of how knowledge about the statistical structure of language leads to an understanding of information density. scientificamerican.com/article/the-ta…
2/ The redundancy layered into the drum beats meant to protect the message from degradation across time and space also works in reverse. What can be said with the same amount of information? What can be understood from any particular set of data? More than we might imagine.
3/ The same process by which information is added in order to reinforce the strength of a message can be used to extract information without loss of meaning. What is the minimum amount of information needed to communicate any given message? Less than we might first imagine.
4/ The answer depends on the message. An average English article can be compressed by about 50%. Photographs are also compressible: light pixels and dark pixels come in clusters. Video is more compressible still, as differences between nearby frames are relatively insignificant.
5/ This is why the broadcast of sportscasters sitting at a table is more compressible than the action happening on the court. The position of the anchors is more predictable than the motion of 10 bodies scrambling for a ball that is bouncing in any and every direction.
6/ A degraded message has less information, but has any of the message been lost? It depends. In the message: “It is Tuesday, July 9th, 2019” the word “Tuesday” is redundant. It is extra information. Lose it, and you don’t lose any of the message.
7/ So, how much is enough information? Do you need a newscaster to tell you that a fire is raging if you can see the smoke rising from the hills? What do you need in order to be certain? How much information is needed in order to arrive at an absolute, final conclusion?
8/ Some of the greatest scientific innovations have come from the realization that an answer is staring you right in the face. Busy measuring and computing, you are gathering and generating reams upon reams of new information, when everything you need is already in front of you.
9/ It was on page six of the @hashgraph white paper that I came to this same realization. Contained within the transaction history of every ledger exists the near sum total of all the information needed to come to aBFT consensus about the state of that same ledger. 
10/ @leemonbaird solved consensus not by creating something new, but by redeploying something old - a thirty-year-old voting algorithm - with the knowledge that the votes had, in a sense, already been cast. His genius was knowing where to find them.
