Block, SkipCaptcha, Upload, Automoderated users, edit, rollback
14,697
edits
Changes
Jump to navigation
Jump to search
Line 243:
Line 243: + + + + +
→Yearly share of gulag fatalities 1917-1988: + methodology ideas
These percentages reflect ''most likely'' percentages according to the author and are not rounded to appropriate degrees of accuracy in order to be useful for comparison or gradually lowering precision calculation purposes.
These percentages reflect ''most likely'' percentages according to the author and are not rounded to appropriate degrees of accuracy in order to be useful for comparison or gradually lowering precision calculation purposes.
: Excellent! Excellent! (<S>except the figures for 1943 probably do not include penal battallions cause, as Solzhenitsyn said, the figures are hard to come by and may never be known. All we know is they were able bodied males, likely 18 years old up, and hardened violent criminals. You'd probably have to look at comparative career, non-property crime, males in other populations to arrive at an estimate. These were disposed of in military penal battallions</s> I misread the chart, taking the figure from the wrong column). And yes, using a figure like the "unborn" corrupts methodology; it's like saying how many jobs ''could'' have been created in America if the Social Security Act of 1936 were not passed. Nonetheless, we still could place this chart in the article with a link citation, and simply say, "According to James B. Whatever blah blah blah...." and let the reader discover the notes on methodology themselves. [[User:RobSmith|RobS]]<sup>[[User talk:RobSmith|''Deep Six the Deep State!'']]</sup> 12:36, 11 October 2018 (EDT)
: Excellent! Excellent! (<S>except the figures for 1943 probably do not include penal battallions cause, as Solzhenitsyn said, the figures are hard to come by and may never be known. All we know is they were able bodied males, likely 18 years old up, and hardened violent criminals. You'd probably have to look at comparative career, non-property crime, males in other populations to arrive at an estimate. These were disposed of in military penal battallions</s> I misread the chart, taking the figure from the wrong column). And yes, using a figure like the "unborn" corrupts methodology; it's like saying how many jobs ''could'' have been created in America if the Social Security Act of 1936 were not passed. Nonetheless, we still could place this chart in the article with a link citation, and simply say, "According to James B. Whatever blah blah blah...." and let the reader discover the notes on methodology themselves. [[User:RobSmith|RobS]]<sup>[[User talk:RobSmith|''Deep Six the Deep State!'']]</sup> 12:36, 11 October 2018 (EDT)
::You caught my interest about the methodology. I suddenly asked myself, while I hadn't before, "Am I giving shares to a lot of unborn children?"
::It turns out he uses a "dummy" function that broadly sketches the proportions of the population shifts he must get from comparing all kinds of yearly data and data ranges. If he has low estimates in the same proportion to each other as the high estimates, he's already ahead of a method that uses only counts (because of the wide error range). Then he refines the various coefficients by trials of which ones fit the population size best when plugged in.
::Now, moving those coefficients he took as his basis simultaneously, the total population is fairly sensitive to small changes to those coefficients. (Hopefully) the shape isn't too distorted and can be retained for larger or smaller hypothetical populations.
::To guard this, there is probably one coefficient (close to the beginning of the series of the basis equations) that impacts the population the most. That one can be employed in shifting its value to prevent distortions caused by the other coefficients.
::The overall effect is somewhat like a mapping the earth on a flat surface; if you draw a graph of the values, you'll see the contour sizes may be a bit off, but the shapes made from the new proportions are visibly recognizable. [[User:VargasMilan|VargasMilan]] ([[User talk:VargasMilan|talk]]) 14:55, 11 October 2018 (EDT)