Blog: Turning on Mistakes

Turning on Mistakes
Turning on Mistakes
Date: 2018-Feb-01 03:22:02 EST

One of the things I most recently changed my mind about was this: remote gaming PCs. Up until recently I dismissed these as useless fads based on the idea that graphics are generally the hardest problem and remote gaming PCs can't replace the 3d acceleration that a nicer video card can get you. This felt like a good argument, but I didn't appreciate how these can work. The challenge in rendering video isn't just getting the bits to the screen; it's doing the rendering involved. We know that fairly slow computers can still play a video of a game pretty well, that video already being prepped and ready for display. They can also play movies fast enough. What they're missing is the rendering, and what a remote gaming rig can get you is a way to hook into a video of the game and pipe your inputs over the network to where the game is running. This can work, it just means that the weak system "receiving" the game needs a dumb relay client. Whoops. There are challenges with this still - you need a fast network to have a fast enough round trip between your gamepad and your incoming video stream.

In other news:

  • The problem with this, a draft law in Poland aiming to make suggesting Polish guilt for Shoah illegal, has nothing to do with Shoah and everything to do with free speech. Disagreeing on historical fact should not be illegal, even on much more distasteful topics.
  • This is a really poor argument by Trump that fighting back against the Russia probe is not obstruction. It's wrong in the same way that arguing that something is not a plant because it is a fruit - because it is one does not mean it is not both. I think this is a longstanding fault in Trump's personality - an unwillingness to let things pass and an insistence that if he's fighting back it is necessarily excused in theory and the specifics. I think in general when people have broken worldviews, we should continually confront them on it rather than withdraw from the topic, as to do otherwise is to accept aberration and weaken the normal
  • I'm generally unfriendly to articles like this that reject testing and objective measures as part of teaching. Not because I think such measures cover all of testing, but I find the antijudgementalism problematic. There was a quote in there - "A student has only mastered something if she can do it when confronted with unfamiliar particulars", and a followup of "Think about training pilots — you would never train pilots by putting them in a simulator and then always running exactly the same set of conditions because next time you were in the plane and the conditions were different you’d die.", that makes more sense to me in another context - that of interviews. One of the big mistakes some tech companies make is that they, in the name of fighting bias, make their interviews far less effective by asking standard questions and having elaborate scoring mechanisms there to take judgement out. I reject that for the reasons above.
  • I've been reading a book by Lewis Dartnell on the theoretical task of rebuilding civilisation after an apocalypse; at least so far it entirely skirts the almost-central issue of dealing with human beliefs and perspective variance; this is a challenge at the best of times (and given how easily stupid optimism about the environment may doom our species, and how tempting it'd be to send scientists and similarly data-driven people to form a colony off-planet as insurance for our species, I worry that failing to focus on this is error.