- Einsteinmonkey
- Registered User
-
Member for 19 years, 3 months, and 11 days
Last active Sat, Feb, 27 2016 23:39:31
- 0 Followers
- 4,233 Total Posts
- 5 Thanks
-
Nov 13, 2008Einsteinmonkey posted a message on HGeez MM, you're such a stalker (cf my wiki page). But it's obsolete info.Posted in: Ugstal Urniancepter Doggienavicenewton Bobwebacks
-
Oct 1, 2007Einsteinmonkey posted a message on Free trade can be bad?I understand what you're saying. I'm seeing it as such:Posted in: Animated Economics
Protectionist policies are in place. The government has two options: keep the current protectionist measures in place, or abolish them. Although free trade does give more choice to consumers and businesspeople in their capacity as buyers, every working citizen in their capacity as a seller loses business. This looks at each individual in more than one light - one part as a buyer and one part as a seller - so when I say "buyers" and "sellers", the same person can be both a buyer and a seller.
Further, this model is looking at the situation as it stands before the decision, which is what introduces the uncertainty. The buyers clearly gain, but the sellers lose out; but by how much do the sellers lose out? - more precisely, how much do they expect to lose, and what are their probability estimates? - and is it enough to overshadow their gains as buyers?
This question is the crux of the matter. If people do not expect their individual returns from trade to be better than their individual losses, they will obviously not like the decision to abolish protectionist policies. Smith simply demonstrates that it's theoretically possible for this to happen (which is not to say that their fears are necessarily right or reasonable).
Of course, after the decision and its consequences, there will be no uncertainty.
And as you know, I agree with you on the rights issue, though personally, I don't like mixing utility and morality. -
Jun 10, 2007Einsteinmonkey posted a message on SXSW Japan NiteTsumasaki is an awesome song.Posted in: get faded
- To post a comment, please login or register a new account.
I was digging through old PMs and found stuff I'd completely forgotten about; a glimpse into my own past. This place meant something to me, even more than I remembered.
One day, MTGS will cease to exist. It happened recently to the other MTG forum where I spent time and youth, and will happen to everything else.
When that day comes, hundreds of thousands of posts - the fossilized record of our hours spent long ago, either well or foolishly - will vanish.
When that day comes, a slice of our universe disappears - perhaps a tiny slice, but a meaningful one.
When that day comes, you will become keenly aware that the only remaining archive exists inside of yourself.
And when that day comes, you will all still hold a special place in my heart.
Aha, see, this is what I wanted from the movie.
Bane (and Jean-Paul Valley) are particular reflections of Bruce Wayne, much like Ra's Al Ghul and the Joker were in the previous movies (reflections in the sense of a funhouse mirror - distorting and exaggerating certain features). This is one of Nolan's strengths. This is why Bane means so much, and defeating him was important. And thinking about it now, perhaps this is why I was disappointed by the Talia reveal; I saw no such meaning in her character.
I do not believe making bland antagonists serves the right goal. Reducing quality may be the easy path, but never the right one. Shoddy design of characters, plots, or whatever, does not somehow enable me to focus on the underlying themes, it distracts me from them. Great design is what makes TDK Batman versus the Joker instead of Batman versus some weirdo dressed like a clown.
Bane was a little underwhelming to me. Based on the Knightfall comic arc, Bane ought to be one of the most impressive villains in the Rogues Gallery: Batman's equal (or superior) in intelligence and power. There was some feeling of those characteristics - he did break Batman's back and his spirit - but for some reason I didn't feel it quite enough.
What really annoyed me though...
There was no hint at Miranda Tate being Talia Al Ghul, unless you already knew she existed (scar and injuries or lack thereof notwithstanding), which was really ****; in fact, I didn't even know who that character was until I saw this thread. It came out of nowhere, and there was absolutely no motivation for it, and any momentum stemming from Bane was immediately dissipated.
What could have been really cool was playing closer to Knightfall, and having a replacement Batman who Bruce Wayne would eventually have to fight. I loved that part of Knightfall because Wayne recognized that it was his spirit that needed to be rebuilt, and so he concentrated on building himself, the man inside the suit, sharply contrasting with Jean-Paul Valley's heavy weapon/gadget style, which is what allowed him to win.
Presumably meant to play off of this novel. The second section in that article (and possible more by following links) seems to have a couple of implicit pointers.
I'm not certain what you mean by "solve the equation", but maximizing the log-likelihood gives you a closed form estimator, the GLS estimator. I'm not too familiar with incorporating Bayesian stats into that, but you should still be able to get an explicit result with a well-behaved prior.
Thanks, I didn't know they were the same thing either.
Where do they write tau in the paper, though? The supplemental materials use 2*pi.
Heh, the one term I neglect to link
Assuming you have some background, that formula is the likelihood function (used in maximum likelihood estimation, a widely used method for estimating parameters) for generalized least squares (a generalization of ordinary least squares that allows for heteroskedasticity and serial correlation in the errors) with multivariate normal errors.
edit: Ok, solved, I took some large videos with my camera.
What does neuroscience tell us about free will [excellent]
The summary based on Roskies 2010 is especially good. We frequently talk too generically about volition when it should be broken down into components that each invoke different philosophical and mental machinery.
What do (and don't) we know about determinism
A few people have mentioned neural networks (and by implication machine learning). It is substantially more complex than the simple types of computer programs most people have in mind, and is undoubtedly the only way a computer would be able to learn such a complex skill like playing Magic. As HH points out, Bakker does not mention this approach. There’s even unsupervised learning, where you don’t need to have the training data’s “true” classification for learning. Is this all currently feasible? Certainly not without a lot more computing ability than is available to the average person. The problem is easier if you restrict the machine to play with one deck so it doesn’t have to relearn the situation. But machines can be made to produce classical music (listen). And I always dislike arguments from ignorance. The one thing you know about a person arguing from ignorance is that they don’t know what they’re talking about.
Pattern abstraction is quite possible, and can be learned without pre-specification. At a simple level, for instance, a network trained to recognize images learns to detect edges on its own. I don't see why MTG requires more "general" reasoning than machine learning is capable of (though there hasn’t really been an indication of what others believe to be the necessary and capable levels of specificity) as compared to humans. Would the machine require certain amounts of pre-specification for e.g. generic complex strategies? Sure. But is that so different from a new player learning better, complex strategies from more experienced players? It’s hard for new players to learn a new deck; is it wrong that it’s also harder for a computer? It’s hard for people to read the metagame with precision; is it a failure if computers have trouble too? Machines may not (or may) be able to reliably tell when facial expressions indicate deceit - but can humans even do that for nontrivial tells?
I would have to think harder about the state specification problem, but I don’t think it’s insurmountable if more specific card types (e.g. on the level of “direct damage”) were heeded.