Good morning on this the third day of December in the year of our Lord 2016.
I had an interesting discussion with a friend about the bayesian interpretation of statistics. I was doing a little research for a video which I’m working on (answering a question from a friend) and what I studied in math was what would often be described as the frequentist interpretation. I’m still a little skeptical of the bayesian interpretation, but much less so as I learn that it’s an interpretation of statistics which completely punts the assigning of probabilities. It calls them “priors”, as in “prior assumptions”, and says nothing about how we arrive at them. Basically, it turns statistics from a math problem that doesn’t apply to the real world (frequentism) into a quantification of our ignorance. Perhaps the clearest example of this is using bayesian statistics to gauge how surprised we should be by an outcome; surprise being, in this case, a measure of how much work we should put into re-examining our priors.
This is a far more reasonable thing than the descriptions of bayesian statistics I had heard before. I should note that those sources were not reliable ones, so I did hold off on judgment. And I think the problem with them was common to how people use classical mathematical probability: they want it to be a way of turning ignorance into knowledge. The desiderata is: garbage in, gold out. Which is to say, what is desired is alchemy for data. The ability to get a lot for little work. And that desire is a perennial temptation.
In other news, I’ve been working through the Vulkan tutorial. I’m still a ways away from being able to display anything on the screen, but I’m up to the point where I’ve found an available graphics card and a suitable queue family from which to request a queue to use for submitting commands. Having already read through the tutorial once, Vulkan is very verbose to set up—the tutorial took abut 800 lines of C++ to get a single, motionless triangle onto the screen—but a lot of that involves making decisions appropriate to your project, which you encapsulate into functions which are much easier to work with, so once you’ve done all this setup work, actually using it for the main graphics work is not significantly harder than other easier, less verbose APIs like openGL. And I do like the approach of having skimmed the tutorial first, then going back and doing it slowly to learn how things go. (And since I’m using the lwjgl (Light Weight Java Gaming Library), there’s a bit of translation work from the C++ of the tutorial to how lwjgl does things.) Fun stuff.
And it’s been a long time since I’ve done any bowmaking, but my problem is that since my third child was born it’s very hard to get an hour or two to myself when it’s OK to make baby-waking levels of noise. When she’s a bit older, I’ll get back to it.
One thing I’ve learned over many years of having a large number of hobbies, is that it’s important to be OK with putting some things on hold for a few years. It’s probably going to be four or five years until I take up knitting again, which I haven’t done much of in the last five years. And that’s OK.