Homeostasis
From MgmtWiki
Full Title or Meme
A particularly life-like description of Stability of living systems.
Context
The attempt to form analogies between computers and life systems continues. Even the word "computer" was originally applied to humans that performed mathematical calculations long hand, or with monstrous, clanking desktop machines.
Norbert Weiner was established as a professor of Mathematics at MIT when he began a collaboration with Aururo RosenBlueth originally with the Harvard Medical School. In 1948 Weiner published "Cybernetics or Control and Communications in the Animal and the Machine",[1] which is as good a description as possible about their collaboration and about the plethora of terms from the life science that invaded computer science.
To Quote from Dr. Weiner:
- A great group of cases in which some sort of feedback is not only exemplified in physiological phenomena but is absolutely essential for the continuation of life is found in what as known as Homeostasis. The conditions under which life, especially healthy life, can continue in the higher animals are quite narrow. p114
- Small, closely knit communities have a very considerable measure of Homeostasis: and this, whether they are highly literate communities in a civilized county or villages of primitive savages. p 160
- Any organism is held together in this action by the possession of means for the acquisition, use, retention, and transmission of information. p 161
- [In a discussion of the control of electric generating systems with obvious analogies to cloud computing] parallel systems had a better Homeostasis than serial systems and therefore survived, while the serial system eliminated itself by natural selection. … This possibility of Self-organization is by no means limited to the very low frequency of [electricity generation or human brain waves]. Consider self-organized systems at the frequency of infrared or radar [frequencies]. p202
- In connection with the effective amount of communal information, one of the most surprising facts about the body politic is its extreme lack of efficient Homeostasis. There is a belief, current in ay countries, which has been elevated to the rank of an official article of faith in the United States, that free competition is itself a homeostatic process: that in a free market the individual selfishness of the bargainers, each seeking to sell as high and buy as low as possible, will result in a stable dynamics of prices, and with redound to the greatest Common Good. … The evidence, such as it is, is against this simple-minded theory. The market is a game … strictly subject to the general theory of games. … Where there are three or more players, the results is one of extreme indeterminacy and instability. … There is no Homeostasis whatever. p 158ff
After his collaboration with Dr. RosenBlueth on living systems, Dr. Weiner was tasked with solving non-linear problems for WWII. His piece de resistance was the design of the analog computer that controlled the heaviest guns for the largest warship (the Missouri) that fought in that war and continued in active service until 1992. Homeostasis serving life and death.
Wiener was not the first to apply feedback to learning. Edward Thorndike is 18
Problems
- As soon as three systems[2] are connected, it is harder than it might be expected to create any system that is self-regulating and self-sustaining.
- Weiner described feedback system that kept big guns on the Missouri battleship rock-solid steady on the target. That sort of feedback works only when the target is well defined.
- Most problems of Identity in particular, or control systems in general, is that the target cannot be so well defined. It is quite hard for a computer to classify pictures of dogs. The computation system that can do this is called Artificial Intelligence.
- Training works for most computer learning systems for identifying people (or dogs) in pictures, but the imposition of other rules, Assurance of Identity of humans, or quality of training of dogs, is substantially harder.
- With indeterminate problems, solutions need to be constantly evaluated to determine if they are performing as required by the currently rules, which may not be the ones they were trained for.
- The wiki page on General Theory of Living Systems describes the similarities between living systems and computing system. Also known as the comparison of Carbon-based Life Forms to silicon-based life-forms.
Solutions
- In multi-computer, multi-networked systems, Homeostasis can only be achieved by components that evolve independently, but are only evaluated by how they contribute to the overall welfare of the totality of the system. In the philosophy of real-world ecosystems, this is known as "Blind Variation and Selective Retention".[3] The assumption is that the Selective Retention is based on some objective measure, which has proved to be a problem when humans are involved in determining the metric.
- Game theory may work to engender inequality, but not comity. There the metric is clear enough, but it might not work to the advantage of all players. Capitalism is one such game.
- Components must be evaluated in an ecosystem of components of roughly equal Complexity. As the richness of systems grow, the number of levels of ecosystems grow as well.
- In general, creation of Homeostasis at one level of Complexity will not help provisioning it at a higher level.
- In terms of Privacy, the systemic privacy of data is not, in general, improved by focusing on the privacy of the components of that system.
- In terms of Identity, the Assurance of the identifier is never determined by looking at each component of the identity system in isolation from the entire ecosystem.
- Robust solutions for Homeostasis need the ability to learn on-the-job and continuously, as the ecosystem itself is constantly changing around each individual, even if the individual is not.
Computerized solutions in order of Complexity:
- The Steepest Descent algorithm measures the slope at the point where the last solution was found. If not zero it will move down the curve to find the lowest (or up to find the highest) valued solution.
- If no slope can be calculated, a Hunting algorithm will search the ecosystem near to where the last solution was found to see if a statistically better one is found. If so it will try that next time.
- A Trust Vector provides a means for creating a sequence of tests of identifiers, attributes, behaviors and inferences that can be statistically evaluated to give Authorization to access valuable resources.
- Bayesian Identity Proofing is a method for using deep learning to improve identity solutions over time.
