tag:blogger.com,1999:blog-8304928500646903522.post7568685202668513282..comments2024-03-28T18:32:05.933-04:00Comments on bensozia: Links 14 January 2022Johnhttp://www.blogger.com/profile/01037215533094998996noreply@blogger.comBlogger2125tag:blogger.com,1999:blog-8304928500646903522.post-56299420290959068612022-01-14T20:30:00.512-05:002022-01-14T20:30:00.512-05:00Using big data and machine learning we can now mak...<i>Using big data and machine learning we can now make accurate scientific predictions without having any idea what the underlying theory might be. This makes many scientists uncomfortable; they want to know why. But does that make sense? And does machine learning show that our theories don't explain the world nearly as well as we would like to think?</i><br /><br />Fundamentally, science is about empirical observation first, and explanations for observations second.<br /><br />If you can make accurate predictions based on observations, that's useful - not as useful as understanding WHY those predictions hold true, but it's something.<br /><br />And there are lots of things we can accurately predict, but still don't really understand. Anesthesia is a classic example - we have a massive amount of information showing that it is safe and effective, but it's all data that was gleaned through trial and error, without us ever fully understanding WHY it works.<br /><br />Gravity is another example - it's clear that there's this force we can observe, but what creates it? We describe it as a fundamental attribute of matter (and energy), but WHY is that the case?<br /><br />Why are any of the fundamental attributes of anything what they are? We very quickly run into a sort of philosophical wall we cannot surmount - even if we can conclusively describe the way something is, if you keep asking what causes things to be that way, eventually you reach a point you can no longer explain except to say "That's just the way things are". We can only explain so much, and then things just have to be accepted without being to point to a yet further "cause" behind them.<br /><br />All of that said... I do still think there's a substantial danger in not understanding the underlying reasons for things, in that it can promote acceptance of a model that can produce ~mostly~ accurate results given available information, but which is ultimately flawed in a subtle way.<br /><br />Take for example the theory of phlogiston - it produced VERY accurate predictions about objects combusting under glass! You could very nicely explain away how the flame would die out because the air within the vessel became saturated with phlogiston, with accurate predictions about the rate at which this process occurred, etc.<br /><br />But what those very accurate predictions got wrong was that instead of the combusting object giving off something, it was in fact absorbing something - if you weighed an object after burning it (with sensitive enough equipment), you would find that the mass of the object had actually very slightly increased (after accounting for loss of soot, etc). Up until the point that this detail was noticed, people assumed they understood what was going on, because their predictions were "accurate". But their understanding was incomplete, and thus wrongheaded - and it required specific observations that people didn't really have much reason to make for it to be proven wrong.<br /><br />That's the major potential danger of not understanding the underlying reasons for your predictions being accurate - you could think you have a better understanding than you actually do, and at best delay your arriving at a proper understanding, and at worst produce some sort of awful result through the actions you take based on your flawed assumptions. Eventually, that sort of risk bites you in the ass.G. Verlorennoreply@blogger.comtag:blogger.com,1999:blog-8304928500646903522.post-33138010027150766642022-01-14T15:45:03.431-05:002022-01-14T15:45:03.431-05:00Study of English horse bones finds that medieval w...<i>Study of English horse bones finds that medieval war horses were not very big by modern standards, 14 hands and shorter.</i><br /><br />I decided to compare numbers, although there's some variance in the numbers I have been able to find.<br /><br />The average medieval Englishman was somewhere around 67 inches (5' 7") tall. This is seemingly a measure of the general population, as calculated from skeletal finds.<br /><br />The average American "Doughboy" in WW1 was apparently about 67.5 inches (5' 7.5") tall - only 0.7% taller.<br /><br />I've found other numbers which show that the average "soldier in WW1" (across all nations, perhaps?) was actually shorter than their medieval counterparts - 65 inches (5' 5") tall. If accurate, I wonder if this is the result of mass conscription (in particular pulling in the very young); or perhaps widespread malnourishment (both from conditions preceding the war and those caused by it); or perhaps simply variances in regional heights (comparing Englishmen to a pool of all soldiers containing many other ethnicities in the rest of Europe); or some other factor.<br /><br />Horses in WW1 came in a wide variety of sizes - there were shortages of horses throughout the war, and people had to take what they could get. Many horses in WW1 were also used differently than medieval "war horses" - the majority were just workhorses used to move materiel, drawing supply trains or towing artillery. But some were "warhorses" in the sense of being ridden into combat, carrying both cavalryman and pack.<br /><br />In both cases, both American and British policies existed showing preference against horses under 15 hands - although I have found reports that smaller horses actually fared better overall, in large part because they were better suited to short rations. Overall, there seems to have been a preference for "medium sized" horses of about 15 hands or slightly over.<br /><br />The jump from the medieval value of 14 hands to the WW1 era value of 15 hands is an increase of about 7% - ten times as big a difference as that between the height of medieval men and WW1 Doughboys.<br /><br />Interestingly, WW1 horse heights match pretty closely to the heights of horses in the Napoleonic Wars. I wonder where (or rather, when) the extra height crept into horses used in warfare. Perhaps it had something to do with the advent of cannons and artillery, and the added necessity of horses to pull such heavy materiel? Or perhaps it had something to do with improved breeding methods developed over time? Or perhaps it is something to do with differences in terrain - where battles tended to be fought, clearances of bogs and marshes allowing heavier horses to be more useful, or even just better road networks influencing things?G. Verlorennoreply@blogger.com