FVerbaas posted at 10:38PM Thu, 27 December 2018 - #4342497

If 'bot censorship' is not an acceptable word in this case we need a new one. For the puropse of the discussion let us use the word 'sensapping' for automated judgement of images for compliance with the rule that governs the realm. Let the software used be named 'sensapp'.

Yet a war broke out on the word 'censor'. Nobody questioned the right of the authority of the realm to impose rules on publications in the realm. Question was how we feel about using a sensapp to decide whether a publication in the realm can be tolerated. The nature of the publications (images) is such that the parameters that drive the rules are in terms of human perception, not in hard numbers. The technology used to solve this sort of problems is artificial intelligence. Artificial intelligence by definition uses fuzzy logic, which is an extrapolation of what the verdict of a group of experts would be if they judged to formulated rules. In the algorithm that is actually applied, there are no formulated rules only the result of training. The database that was used to train the AI is not published. The question is particular sensitive because the output of the AI is a yes/no black/white decision. Of course if ratings go between 0.0 and 2000, and publication can be only if end rating is at least 1000, 1000.0001 is more than 1000 and can stay, and 999.9999 is less than 1000 and must go. If the amount of 'flesh color' on the image counts in the degree of 'pornographicness', changing the reflectiveness of a side prop can make the difference.