real question: what distinguishes information from noise? if i declare an arbitrary pattern as ordered, does it suddenly meet the physics threshold for not being able to be destroyed/created?
Comments
Log in with your Bluesky account to leave a comment
Usually a pattern, or changes that correspond to what you want to measure.
If your data isn’t clear enough, or not what you expect, checking the sensors or measurement device, simplification of the experiment, or the removal of variables that could interfere with the result is the next step.
Someone else being able to draw the same info from it as you, without being told about the content beforehand, only the method by which you arrived at it?
Short version seems to be that the point of the phrase is to say that a complete view of all information in the universe should allow it to be perfectly modeled but to be able to destroy some of that information, with like a black hole, goes against that?
Per information theory, noise has the highest entropy as it requires the most information to specify it (there's no pattern so you have to list every bit). Information in our ordinary sense is conveyed by low-entropy strings, with order and structure.
Among other things this means that doing computation can in principle be thermodynamically reversible, but *deleting* information, e.g. zeroing an array of bits, reduces entropy locally and so must be paid for by expending energy and increasing the entropy of the rest of the universe.
Likewise compression algorithms seek to represent our long strings of text by shorter strings - this only works if the original data has order and structure, and so could be represented with fewer bits. A string of random bitnoise doesn't compress.
However, this DOES mean that it can be very difficult to distinguish optimally-compressed information from noise.
(Also, you can use this logic in reverse - if an encrypted message can be compressed, it means that there is some pattern present that an attacker might be able to use to decode it.)
Yes. The distinction no longer lies in the bitstream (which now is noise) but in the combination of the bitstream and the compression/decompression algorithm - this bit of noise decompresses to text and that one doesn't.
Comments
If your data isn’t clear enough, or not what you expect, checking the sensors or measurement device, simplification of the experiment, or the removal of variables that could interfere with the result is the next step.
If yes, information. If not, noise.
So while you can make arbitrary assignments, you can't change the mathematical rules underpinning stuff.
(Also, you can use this logic in reverse - if an encrypted message can be compressed, it means that there is some pattern present that an attacker might be able to use to decode it.)
Long version: https://en.wikipedia.org/wiki/Information_theory