While doing some reading, I recently came across a paper, Weighing the fog of war: Illustrating the power of Bayesian methods for historical analysis through the Battle of the Dogger Bank. The paper focuses on the 1915 Battle of the Dogger Bank, a clash between the battlecruiser forces of the Royal Navy and German Kaiserliche Marine. The battle was a minor victory for the British - they sank the German Blucher and heavily damaged the Seydlitz in return for heavy damage to HMS Lion, but confused signalling meant that the British force broke off the battle before a decisive conclusion had been reached. It's commonly assumed that the battle would have ended in a British victory had the fighting continued. The paper attempts to test this by using a Bayesian simulation of the battle, and concludes that, in fact, the British were more likely to lose more ships than the Germans.

On the one hand, this is an interesting attempt at applying some of the techniques I work with on a daily basis to a topic of history I'm greatly interested in. As an astronomer, my work focuses on applying Bayesian statistics to simulate radio emission from dust clouds - this paper uses them to simulate one of the most important naval battles of WWI. On the other hand, I feel there's some serious issues with their assumptions, and hence with their model. For example, the model does not simulate hits causing flooding. Instead, they model ships as a collection of turrets, that are destroyed when all turrets are knocked out. Given that flooding caused the loss of Blucher at Dogger Bank, and the sinking of Lutzow and the disablement of Seydlitz and Derfflinger at Jutland, this omission would seem to favour the Germans. They also assume that British munitions handling did not change between Dogger Bank and Jutland, arguing that

[Beatty's] extensive report to the Admiralty of 9 February 1915 recommended changes, but not in gunnery training or ammunition handling, which did not appear to be causes for concern in terms of the perceived results of the battle.

This is an assessment most historians of the period would disagree with. While Beatty did not directly recommend changes in munitions handling, he did recommend that the rate of fire of British ships be increased to the maximum possible. This meant that many crews compromised on safe munitions handling, making their ships far more vulnerable at Jutland than they were at Dogger Bank. The authors of the paper also rely heavily on Philbin's The battle of Dogger Bank: The first dreadnought engagement, January 1915. While this is the most recent treatment of the battle available, it has not received good reviews. Given all this, I'm not sure whether or not applying statistical techniques in an attempt to simulate history is a particularly useful endeavour - a few wrong assumptions or inputs, and your results will be irrelevant, and it can be very difficult to check whether or not an assumption is correct.

I'd be interested in hearing other people's views on this topic. Are statistical simulations of history a useful way to explore possible outcomes, or are they doomed by the necessary simplifications and assumptions? And does anyone have any papers that do this sort of simulation more convincingly?

I noticed several academic books have "in the X imagination" in their title. Such as Out of the East: Spices and the Medieval Imagination by Paul Freedman, Saracens, Islam in the Medieval European Imagination by John Tolan, and The Peasant Kingdom: Canada in The 19th-Century Russian Imagination by J.L. Black.

What is the origin of the "in the X imagination," in the title, how common is it, and why the word imagination instead of a different word?

I'm struggling with a certain attitude reflected by modern historians - especially non-military modern historians - about the military realities of the early 19th century. In works that touch on my current subject, there is often this stubborn insistence on discussing training in regards to regular soldiers and militia.

The reality was that many regulars would have been no better trained than their militia counterparts, owing to the absence of any kind of regularity of training or even a stable training methodology for the American military. The attitude of "superior training" to me seems to be symptomatic of taking the modern military status quo and projecting it into the past with little examination.

This leads to the kind of militia-derisive axioms I often butt heads with. Though there's certainly a case to be made for the later-war superiority of regulars, and especially those with specific training conditions (such as the months-long training regime of Winfield Scott prior to Chippawa), but I find the "training" projection hard-coded in historiography and difficult to examine.

Perhaps other military historians here have dealt with similar issues?