Bayes' Bastards

Bayes' Bastards, also known as the Bayesian Conspiracy, was a group of scholars and activists formed near the end of the Eighth Caliphate. On a basic level, their goal was shifting the rational paradigm from the largely frequentist traditional rationalism to Bayesian rationalism, arguing that the perspective where probability expresses a degree of belief in an event produced measurably better results as opposed to the standard view at the time; that is, probability as an expression of the relative frequency of said event. Much more broadly, they brought about radical social changes which culminated in the collapse of the Eighth Caliphate, as well as serving as the main organizational body of the Ninth (also referred to as the Caliphate of Flowers).

Bayes' Bastards originally arose as a circle of Eighth Caliphate Rationalist scholars dedicated to modelling the process of human reasoning using statistical models. The founder of the circle, Abn al-Hakim Shahrukh, was concerned with the state of then-modern decision theory and desired a rigorous mathematical model which could be used to quantitatively measure the rationality of a given agent. This goal was accomplished within several decades, throughout which the circle remained relatively small and consisted mostly of statisticians, logicians, and game theorists. The theory they produced is now known as the Shahrukh-Bashur-Uday metric, after Shahrukh and two other core members of the circle, Samir Bashur and Waleed Uday.

It was at this point that that the circle began working on the problem which would lead to their rise to prominence: that of applying the Shahrukh-Bashur-Uday metric to human reasoning. In order to accomplish this, experts from many different fields were invited to the circle, including but not limited to evolutionary psychology, cognitive archaeology, Bayesian statistics, and chaos theory. Their approach involved studying both the abstract processes involved in human reasoning and the physical neural structures responsible for producing them in order to form a multi-level model of the same. At the time, even the most powerful supercomputers were still unable to fully simulate a human brain, so this strategy was designed to act as a workaround while remaining a close approximation of the actual thing. To delve into the specifics, while fascinating, would unduly prolong this section, so I instead refer interested readers to Darius Sinclair's The Bayesian Homunculus, which offers a detailed account of the above. Suffice it to say, then, that they completed the model: the Bayesian homunculus was born.

One thing was immediately made abundantly clear upon combining the Shahrukh-Bashur-Uday metric with the homunculus: humans are not rational animals. This statement may seem obvious, especially to one with a background in the subject, but remember that this was the first time anyone had been presented with such an accurate and rigorous proof of the fact. It is one thing to modestly proclaim "We are none of us rational!", but quite another to look into a system mathematically proven to be dangerously flawed and see a mirror of oneself looking back. Nor was that the worst of it: when the homunculus was made in the image of a traditional rationalist, there was no significant difference between the value given by the Shahrukh-Bashur-Uday metric and that given for an uninitiated human. The fact that the circle did not immediately fragment is a testament to the formidable mental prowess of its members. Not only did they expose their own flaws, but they revealed that the framework on which much of their lives were constructed was shaky at best, on the verge of collapsing at worst. Indeed, the circle did not come through this revelatory period entirely unscathed: some members fell into a deep depression, and there were several suicides (which should illustrate how central rationality was to their lives). Despite this, the core group was preserved, and their minds were too brilliant to let this stand. In short order, Bayesian rationality was proven to be optimal, and in slightly longer order a plan of attack was formulated. The circle would keep their experimental results from all except those vital to effect the radical social change required for traditional rationality to be overturned, and endeavour to institute broad reforms in pursuit of this goal. Thus were Bayes' Bastards born.

The changes brought about by the Bayesian Conspiracy are both manifold and, I am sure, familiar to the reader, so I will just touch upon some of the more major points. Security of communication between members was effected through the Aorta System in the early days of the Conspiracy, with a dedicated secure intranet being constructed later. Nevertheless, some information leakage was inevitable, which in combination with the fact that Bayes' Bastards were actively opposed to traditional rationality contributed to the eventual collapse of the Eighth Caliphate. This was largely due to two factors: first, many of the reforms espoused by the Bayesian Conspiracy attracted widespread support among the general populace, and secondly, the Eighth Caliphate Rationalists were a core part of the Caliphate's power structure. The eventual proliferation of the Bayesian Conspiracy's experimental results caused deep divisions among the upper echelons of the ECR, severely impairing its function and by extension that of the Eighth Caliphate's governing body. After the collapse, the Bayesian Conspiracy served as the organizing presence behind the establishment of the Ninth Caliphate or Caliphate of Flowers.

Finally, both the Shahrukh-Bashur-Uday metric and the Bayesian homunculus proved extremely important in the development of thinking machines later in the Golden Millenium, specifically in the context of providing a vital framework for the construction of Friendly AI and associated superintelligences. A modified version of the S-B-U metric also forms the basis for the technique of utility mapping, which is one of if not the single most important tool for creating Friendly AI; indeed, almost every AI currently in existence has architecture based on these two concepts. Notably, the Jupiter Brain superintelligence independently reproduced both results within three microseconds of its activation (whether this was due to inspection of its own neural structure or simply a result of a logical reasoning process is still an open question).

-Excerpt from Rationality Throughout the Ages by Ánthos Daskydox