Print Story I'm not a real doctor.
Motorbikes
By gzt (Mon Oct 01, 2012 at 10:14:22 PM EST) gzt, chess, eigenvalues (all tags)
I have a master's degree... in SCIENCE. I really love this sort of thing: http://drscience.com/wordpress/


Test in R class tomorrow. HW for other class requires extensive SAS.

Read a bit, extracurricularly, about linear algebra after looking up Greshgorin's Theorem. Had to refresh my memory a bit about eigenvalue stuff.

Yesss: I can cut some office hours because people never show up.

Larry Wasserman made a post about K++ means (a technique for using K-means that helps you pick good starting nodes), wondering aloud about a couple things, and somebody on reddit pointed to a paper by people I know in my department who did experiments showing it placed in the middle of the pack.

A friend of mine is getting married! Way cool. And apparently we're going to be invited! Way cool. She's the sort of person who has a billion people, family and what-have-you, and I don't quite know how close we really are, so I would not have been offended at all not to be invited. But pleasantly pleasant to be invited. It's right before New Year's.

I'm going to have to take over the Orthodox Christian student group at the university. But, you know, if the way they're proposing running the meetings make them the sort of meeting I wouldn't want to go to and that I think other people wouldn't want to go to, might as well turn it into something better.

I've been reading some non-statisticians talk about rationality and how it's super-rational to be Bayesian and it's so much better than "frequentism", but I'm not really feeling it, as a statistician. They're not really doing probability or statistics. Using Bayes' Theorem doesn't make you a Bayesian, using Bayes factors instead of the classical likelihood ratio makes you a Bayesian, but I don't think very many integrals are going on in a soi-disant "rationalist's" head. They certainly are in some cases, to be sure, but, really, I have one simple rule: you can't be a Bayesian if you've never done an ugly integral to calculate a Bayes factor. You definitely can't be one if you don't know how to set up the integrals or can't pick one out of a lineup with a maximum likelihood estimator.

I mean, you can believe in general relativity without knowing what a tensor is, you can believe in quantum mechanics without understanding bra-kets and Hilbert spaces, but this differences only really comes out in either ugly integrals or high-powered computing. In terms of impact on everyday life, it can't have an impact because the level you're going to use the theorem for informing your beliefs is not going to distinguish you from the frequentists - it's a theorem of frequentist statistics, too, you know. Or maybe I'm missing something, who knows? I'm not exactly good at thinking.

http://www.skepticalscience.com/news.php?n=1641 Of course, one must be skeptical of anything involving cost projections.

I just replaced my 2 gallon per minute shower head with a 1.25 gallon per minute shower head.

Finally ordered a clothes washer. The local government will give me a $100 rebate because we got an "Energy Star" one.

Still on a Soundgarden kick.

The implicit political atmosphere of the students here is extremely Democratic/Liberalist. I suspect some people are more moderate, but this is the only viewpoint expressed.

Doctor Who: that was manipulative.

Now, by the power vested in me by the state of New New York, here is my diary.

< What I want.... | Event driven. >
I'm not a real doctor. | 29 comments (29 topical, 0 hidden) | Trackback
Doctor Who by jimgon (4.00 / 1) #1 Tue Oct 02, 2012 at 06:28:23 AM EST
I've now had enough of the Angels.  The Angels, Cybermen and the Daleks.  They need to rummage around a bit in the closet and find some other foes. 




---------------
Technician - "We can't even get decent physical health care. Mental health is like witchcraft here."
Other foes by anonimouse (4.00 / 1) #2 Tue Oct 02, 2012 at 07:00:56 AM EST
I suggest watching Series 1 of Grimm. Lots of different foes, including fee fi fo fum....


Girls come and go but a mortgage is for 25 years -- JtL
[ Parent ]
+1 by Herring (4.00 / 2) #3 Tue Oct 02, 2012 at 07:06:54 AM EST
Needs
  1. New enemies
  2. Enemies that are defeated by science, not by "everybody think nice thoughts"
  3. More resources on plots, less on famous guest stars
  4. A modern-day Nichola Bryant. Everybody likes boobs.


christ, we're all old now - StackyMcRacky
[ Parent ]
I'll pass on Nichola Bryant by anonimouse (4.00 / 2) #5 Tue Oct 02, 2012 at 07:13:43 AM EST
I preferred Billie frippet Piper.


Girls come and go but a mortgage is for 25 years -- JtL
[ Parent ]
The lass... by Vulch (4.00 / 1) #7 Tue Oct 02, 2012 at 09:47:50 AM EST
...Who had been turned into a Dalek in the first episode of this series is back in the Crimble Special as the new assistant...



[ Parent ]
Ah. Good. by Herring (2.00 / 0) #8 Tue Oct 02, 2012 at 09:52:53 AM EST
Now have an an urge to cook a cheese soufflé.

christ, we're all old now - StackyMcRacky
[ Parent ]
srsly. by gzt (2.00 / 0) #4 Tue Oct 02, 2012 at 07:12:15 AM EST
Especially with the odd rewriting of the Angels every time they come up. They put the Cybermen and Daleks in once per season, it seems, but that means they take up a significant proportion of the season. They've killed off The Master. Perhaps they should start stealing from Star Wars and get some Romulans, maybe the Gorn...

[ Parent ]
Or they could steal from star trek by darkbrown (2.00 / 0) #6 Tue Oct 02, 2012 at 07:52:38 AM EST
Get some Twi'leks or Mon Calimari

[ Parent ]
Soundgarden by barooo (2.00 / 0) #9 Tue Oct 02, 2012 at 12:18:45 PM EST
Fun fact: their best song is "Room a Thousand Years Wide". 

man, i need a beefy taco now.
-gzt
that is a good song by clover kicker (2.00 / 0) #10 Tue Oct 02, 2012 at 02:53:24 PM EST
I have a certain sentimental attachment to Big Dumb Sex though.

[ Parent ]
Dr. Science by Bunny Vomit (2.00 / 0) #11 Wed Oct 03, 2012 at 09:11:56 AM EST
Nice to see he's still around.  I remember hearing his 2 minute radio bits on NPR back in '99 or so.


~~
I don't know if he's really still around by gzt (2.00 / 0) #29 Sat Oct 06, 2012 at 11:35:51 AM EST
Sure, he has a web page, but who knows how he's really doing? I don't listen to the radio. But I love that sort of thing. It's one of the reasons I hang around http://www.reddit.com/r/shittyaskscience/

[ Parent ]
On super rationalists. by wumpus (2.00 / 0) #12 Wed Oct 03, 2012 at 09:12:07 AM EST
I read way too much on a similar websites: http://lesswrong.com/ http://www.overcomingbias.com/ (one appears to be choice posts of the other) and more than a few things set off the kook meter.

Newcomb's problem: They make a big deal about this. Essentially, they set up a strawman stating that if condition A (where A is a strongly believed assumption, in this case causality) is disproved, then you should make your decisions ignoring A. Expect to see this too many times, and it shows up as a major topic with subheadings on the website mentioned. This isn't the only strawman, look up zombies (not the undead type) as well. Zombies aren't quite the major topic that Newcomb is.

Many worlds Our heroes come out and claim that the official position of physics is flat out wrong (presumably the Copenhagen system, but I don't know enough to tell if they are lying here). The kicker is claiming that their Many Worlds postulate somehow doesn't violate relativity/causality. I guess I'll have to join up to ask (note they limit starting topics to vetted members, and others say Husi is cliquish), but it seems that the Many Worlds divide either splits faster than C, or happens instantly (... in all frames of reference???). Since they aren't claiming specific laws being broken, there is still room left on the kook meter, but it is registering high.

Frequentist I can't begin to understand the argument. This might have something to do with studying EE: frequency has some very specific meanings, and I doubt I have room for more.

Note: this puts the number of seriously hard disciplines that they claim to have leapfrogged to at (at least) two (and note the arguments against each aren't at all related). This puts them in a fairly limited selection of kooks and I somehow missed it before writing this note.

Bayne's Theorem (without frequency nonsense): imagine the following dialog.

Bayesian Master: "Behold: I have flipped a coin heads for 50 times. What will the next be?"
Simplicio: Heads?
BM: And why is that?
S: Because the odds of you having a two headed coin are noticeably higher than a trillion to one. BM: You've discovered the secret of the universe on your own! I bow to you.

Note that getting the exact probability (assuming you know both probabilities going in) is rather useful. The catch here is that you would assume that if this is in fact the secret of the universe, most of the rest of the websites claiming such would be spent on determining P(X) and P(X|Y). They do not. In fact, I'm not even sure it is on there (you don't want to know how much of their recommended reading I've plowed through: the local library hasn't been kind lately), unless it involves determining "priors" and comparing them with the dreaded frequencists. I am reminded of a leftist evangelical critique of the Left Behind series. The hero spends two sentences reading the Gospels, agrees wholeheartedly (without specifics) and has his entire life changed without ever referencing them again. The similarity isn't quite there, but it is too close for comfort.

In the end I will have to find somewhere to try to get an answer about how many worlds doesn't violate relativity (although I don't really have enough physics to avoid being snowed). The cognitive bias parts look important, but a successful kook works by making the stuff you don't know look accurate and the stuff you do know look less critical. Time to look elsewhere to check cognitive biases.

On the other hand, their deconstruction of Harry Potter (and reconstruction with a much more consistent world) is not to be missed.

Wumpus

yes, I am indeed talking about lesswrong by gzt (2.00 / 0) #13 Wed Oct 03, 2012 at 10:47:38 AM EST
"Many Worlds" is a valid and somewhat popular interpretation of quantum mechanics, but Yudkowsky gets a lot of basic physics wrong and, in the actual physics community, it is by no means a slam dunk. So their fanboi-ism of it is not rational.

Newcomb: You mean they didn't trivially solve a difficult philosophical problem just by being rational? Oh noes!

Frequentism and Bayes. If you've ever taken a statistics course, you probably learned a "frequentist" interpretation of statistics (and you also learned Bayes' Theorem). You have null hypotheses and you have p<.05 and that kind of crap. Somewhere underneath the hood - you didn't get to this unless you needed even more theory - you have maximum likelihood estimators, sufficient statistics, a bunch of crap like that. Okay, never mind, I'm not going to get into this. I would just say that a lot of the naive ways of thinking about probability, confidence intervals, that kind of thing, are more Bayesian in their terminology than frequentist, but there isn't a practical difference until you're doing very ugly integrals and solving computationally intensive problems. So... they're silly.

I have the opposite opinion about their Harry Potter. Except for the rape chapter.

[ Parent ]
and then... by gzt (2.00 / 0) #14 Wed Oct 03, 2012 at 10:48:22 AM EST
...and then there's the whole singularity/"friendly AI" thing.

[ Parent ]
No statistics courses. by wumpus (2.00 / 0) #15 Wed Oct 03, 2012 at 11:50:55 AM EST
One probability course (random variables). Probabilities should plug into Boolean logic better, but I'm not surprised that the rest of the claims are pure fanboyism.

"difficult philosophical problem" I'm not seeing the difficulty (although I caught glimpses trying to explain why it wasn't), unless you want to claim that ignoring causality is a similar agony to what Kepler went through when he admitted that planets didn't stick to circular motion.

"Yudkowsky gets a lot of basic physics wrong". My knowledge of relativity and quantum physics are about as weak as you can still get a bachelors in engineering, but I'm pretty sure I caught him sweeping that bit under the rug. Other mistakes shouldn't surprise me.

Wumpus

[ Parent ]
I was slightly unfair by gzt (2.00 / 0) #17 Wed Oct 03, 2012 at 01:45:29 PM EST
Not a "lot" of basic physics - he gets one thing remarkably wrong that he may have corrected by now about the polarity of reflected light. Everything beyond that is quantum and relativity and difficult and beyond my pay grade (and Yudkowsky's, manifestly).

Well, the problem with this philosophical problem is that everybody thinks it's easy, but half of people think it's one answer, half think it's the other. Whatever. They cut through a lot of Gordian knots in other philosophical problems, it gets sloppy.

[ Parent ]
damn interweb cliques. by the mariner (2.00 / 0) #16 Wed Oct 03, 2012 at 12:18:47 PM EST
i can't tell you how many i've really wanted to post about someone's intertron theorizing but was prevented from doing so by some kind of moderation or waiting period.

well, okay, i probably could... 

[ Parent ]
the devil you say by nathan (2.00 / 0) #18 Wed Oct 03, 2012 at 09:28:00 PM EST

I suspect some people are more moderate, but this is the only viewpoint expressed.

Since they are rational, educated academics, I doubt any of them are "moderate," much less "deformed, hate-filled troglodytes who eat mud."



nah, probably the PC thought police. by the mariner (2.00 / 0) #19 Wed Oct 03, 2012 at 10:50:41 PM EST
they're just keeping their heads down, protecting their reputations.

[ Parent ]
do we really need to resort to conspiracy theories by nathan (2.00 / 0) #20 Wed Oct 03, 2012 at 11:04:15 PM EST
When it's perfectly clear what's going on?

Face it -- it's vanishingly unlikely that even one person, other than maybe that annoying guy with a Ron Paul kiosk in the students' union, would not be perfectly in accord with the party of educated people. Not merely midrange partisan positions but ideological vanguard ones, of course -- these kids are not merely educated, they are very educated.

[ Parent ]
however, conversely... by gzt (2.00 / 0) #21 Wed Oct 03, 2012 at 11:13:59 PM EST
...there are horse barns not two blocks away from our building. has a member of the enlightened party ever ridden a horse (except as part of an expensive hobby)? there are strong spatial effects in political alignments.

[ Parent ]
presumably these kids aren't riding horses by nathan (2.00 / 0) #24 Wed Oct 03, 2012 at 11:33:09 PM EST

Some very bad people would suggest that they are riding those horses secretly so that they are not seen to be riders. This is, of course, a monstrous suggestion, since riding a horse is animal slavery and contrary to all enlightened ethics.



[ Parent ]
I don't know if anybody rides these horses. by gzt (2.00 / 0) #25 Wed Oct 03, 2012 at 11:44:00 PM EST
They're research horses. Maybe they get eaten or something. Sounds like something the Other Party would do. and the spatial effects just bleed all over. and somebody just started up a reading group for financial time series - obvious infiltration of the party of bankers and private equity rather than right-thinkers. i must be double-vigilant, somebody might tip their hand.

[ Parent ]
i'm not sure what to make of this post by nathan (2.00 / 0) #26 Wed Oct 03, 2012 at 11:56:28 PM EST
In my experience, bankers are private equity people have quite good thoughts. Perhaps some of those narrow-minded small-town bankers have bad thoughts, but who cares about them?

[ Parent ]
*bankers AND private by nathan (2.00 / 0) #27 Wed Oct 03, 2012 at 11:57:22 PM EST


[ Parent ]
right, but the public thinks they don't by gzt (2.00 / 0) #28 Thu Oct 04, 2012 at 08:45:16 AM EST
so non-bankers emulating bankers may weigh more than a duck.

[ Parent ]
ya, so educated they can't do an honest day's work by the mariner (2.00 / 0) #22 Wed Oct 03, 2012 at 11:25:53 PM EST
it's no wonder you can't find two cents between them. no worries, just pick up another loan and buy a round on the taxpayers' dime.

[ Parent ]
Their earning potential is greatly amplified by nathan (2.00 / 0) #23 Wed Oct 03, 2012 at 11:31:07 PM EST
By a college degree. After all, if it weren't, who'd borrow $100k at 7.9% to go to college? Especially since the BACPA of 2005 made all those loans nondischargeable.

[ Parent ]
I'm not a real doctor. | 29 comments (29 topical, 0 hidden) | Trackback