User:Dredmorbius/Signal theory

From PlexodusWiki
Jump to: navigation, search

This is a sketch of an alternative classification scheme for signals (information) expressed as either signals or records and their respective properties.

It is based on ideas I've been thinking of for some months prior to Google's announcement of the Google+ sunset, and may or may not be relevant, result of which is that I'm parking it as a draft for review first.

It's quite abstract, but possibly useful.

Information Theory as Signal theory[edit]

In a nutshell, this looks back at Claud Shannon's work, treats Information Theory as instead Signal theory, and recognises (as I believe Shannon did) that there are two types of messages:

  • Messages propagating through space: signals
  • Messages propagating through time: records

The distinction between signal and information is that the first does not presume a necessary associated semantic value, while information seems to loudly suggest this. I feel that some confusion over "information theory" results from this, though Shannon is generally careful in his thinking and terms.

Signals are dynamic, ephemeral, move through channels, generally expressed as either physical force fields (e.g., electrical, magnetic) or modulations in energy (e.g., sound, physical vibration), have transmitters and receivers. Signals move through a medium, and involve changes in time.

Records are static, durable, exist in some fixed substrate, expressed as an organised state of matter, have inscribers and readers (in a technical, not individual sense). Records exist in a medium, and involve changes in space.

Both are subject to noise and/or decay, and have an encoding and decoding process.

Most information systems (better: communications systems) involve interactions and exchanges between signals and records. Signal inputs are received, parsed, processed, stored, as records, are retrieved (searched, sorted, filtered, selected), possibly re-processed, and are again transmitted (as signals).

(It's long been observed that most computers do far less "computing" than informating -- recieving, processing, storing, retrieving, and transmitting signals and records.)

That's not the classification, but the background.

Properties of signals and Records[edit]

Signals and records each have sets of properties.

Signal properties[edit]

Signal properties include:

Transmission medium, broadcast / narrowcast / channeling, bandwidth, propagation speed, fidelity, noise, leakage, interference, amplitude, range, perceptibility, frequency (bits/time ratio), transmission efficiency (bits/energy ratio), and receiver sensitivity, directness/indirectness of perceptibility.

Record properties[edit]

Record properties include:

Mutability/durability, density, ease of inscription, ease of reading, bits/mass ratio, bits/area ratio, degredation, contrast, encoding and decoding complexity.


Scale is an integral element of signals.

The speed at which a message can propagate.

The rate at which information can be encoded or decoded to or from a channel, or into or out of a medium.

The density with which information can be inscribed or read.

The duration for which a record remains readable.

The interactivity between two parties.

Channel and substrate characteristics[edit]

There are characteristics which are characteristic of and conducive to both transmission and storage media.


The characteristics of the channels themselves: Light, EMR, magnetism, sound, physical movement, chemical or endocrinological signalling. Each has characteristics of speed, bandwidth, durability (chemical signalling has characteristics of records and signals -- mercaptin as a mine hazard signal in a dark and noisy environment, ant signalling via chemicals),

Channels have included touch, sound, and light -- generally physical forces. I suppose one could come up with a thermal-based transmission system, though it would be slow and inefficient.

The essential requirement of a channel is that it be conducive to percieving changes over time. The ideal channel is one in which there is either no base signal, say, a still pond, or a uniform base signal, say, a uniform carrier wave. This recognition is interesting as it means that any phenomenon which provides a quiet or uniform base is inherently an excellent communications channel candidate. The overwhelming use of lasers not in energy delivery, as was originally widely predicted, but communications technology, derives from this.


Recording substrates have included dirt, stone, clay, sand, rock walls, papyrus, hides, wood, vellum, paper, films, magnetic tape, magnetic disk, optical disk, magnetic core, ROM and RAM storage, flash storage, SSDs, crystal block storage, in combination with marks, scratches, inks, pigments, chemical fields, magnetic fields, optical pits, and electrical charge distributions.

The essential requirement of a substrate is that it be conducive to acquiring an inscription and retaining it over time. Sand or frost acquire inscriptions, but are not generally durable. Hardened metals are durable but difficult to inscribe. Inks applied to or absorbed into substrates have proved remarkably durable, and easily inscribed. Laser-etched crystals may prove to be durable for durations of hundreds of millions or billions of years, though they're slow to write.

Intrinsic grammars[edit]

The nature of both channels and substrates gives rise to some basic elements of communication, most of which are widely understood.

Danger, alert, or emergency signals tend to be simple, highly-discernable, rapidly disseminated, and not very durable. E.O. Wilson is among those who've noted that among ants, which rely heavily on chemical-based signalling, alert signals are simple, light, small molecules, and may be shared amongst numerous ant and even other insect species. Danger signals of other animals, including humans, are similarly highly obvious and are generally not encoded to be imperceptible or unintelligible to others.

Identity signals tend to be complex and durable. Also amongst ants, Wilson notes that the chemicals which identify specific species (and possibly colonies) are complex, specific, and durable. Similarly, chemicals which mark food paths. Among other animals, including humans, the signals which identify tribal allegiance are complex, difficult to learn or emulate, and durable. These include no only language, but deep structures of culture, dress, food, conduct, beliefs, pronunciation, movement, rituals, and more.

These two examples operate at distant ranges of communication complexity (though there may be yet more complex elements than identity). But they (and other discoveries within information and communications theory) suggest some underlying principles. I'm not particularly familiar with Chomsky's work, but the notion of a universal human language grammar may be related or build from similar principles. I also suspect that behavioural characteristics, particularly in the context of information systems, employ similar logic.


Ephemeralisation of media or channel increases the rates of possible information transfer (signals) or storage (records).

The faster, lower-energy, lighter, more discernable, or more sensitive (reciever/reader) a signal is, the greater its capacity to be transmitted, distributed, recorded, and/or read.

Humans and signalling contexts[edit]

The brain as a storage and processing system[edit]

Population variability in sensitivity, storage, processing, models[edit]

Language as a shared semantic protocol[edit]

Language is a common understanding of the meanings ascribed to signals, both transmissions and recordings.