EPO and Hobby Cyclists

[Benjamin Weiner]

Kyle Legate <[email protected]> wrote:
> Howard,

> While your points are all correct based on careful, distanced analysis of
> the available FACTS, you'll never get through to a guy who thinks that all
> terrorism is caused by a girl's ass crack.

Our problem is that Western society's corruptive influence isn't
corrosive enough. Several of the Sept 11th plotters were in the
U.S. for months before the attacks, occassionally making forays
into Florida bars (IIRC). If only one of them had been corrupted
enough to think "Hey, I'm not sure I want to die for a holy cause.
I want to spend some more time here drinking Bud Light and checking
out girls' asses," the whole tragedy might have been averted.

I'd still like to see the entire cast of "Friends" launched on
a rocket into the sun, though, but not for foreign policy reasons.

 

[Tom Kunich]

"gwhite" <[email protected]> wrote in message
news:[email protected]...
> "Tom Kunich" wrote:
> >
> > Greg, I finally came to the conclusion yesterday that what these people
are
> > doing is trying to pretend that THEY have no responsibility for the
> > terrorism in the world.
>
> Well I certainly don't feel responsible *for* it. The complaints
> about things like US trade policy and culture "stomping" as provoking
> terrorism are utterly ridiculous. I do feel responsible to do
> something *about* terrorism.

that's because the "responsibility" is so diaphanous. We believe in live and
let live. We allow "adult book stores" to be opened three doors down from
grade schools and we essentially do nothing about kiddy **** and roadside
bill boards about feminine hygene products all of which are so disturbing to
most of the rest of the world that we don't even have a good idea of just
how outraged they really are.

I worked with very educated people from all over the globe and in many cases
when prompted they have spoken of humiliation when turning on the TV, radio
or even opening a newspaper. Although they will say that they prefer the
freedom of this country, when they go back to their own countries they don't
talk about Standard Oil buying crude from Saudi sheiks, but about "Friends"
(well, actually soap operas but you know what I mean) coming through on the
satellite link to every home. In second world countries they all want to
have the luxuries and comforts that we Americans have but their religious
leaders see the moral corruption much more than the Escalades.

You don't feel responsible but you are. As am I. One of the differences here
is that I have accepted that responsibility. I have considered what we could
do and have come to the conclusion that we have been doing about as good a
job on the overall picture as we could. No one could predict Ayatollah
Komeni and in fact the Shah was a MUCH kinder and gentler governor than the
Ayatohlahs but a substantial margin. To suggest that our support of Shah
Pavlevi was wrong was to completely ignore Iran's history before and after
the Shah. In fact, what brought him down wasn't his somewhat free-handed
secret service who were butchers on the occassion, but that he was
attempting to install a government free from religious affiliations.

> We are probably in agreement that an effete (appeasement) policy is
> one doomed for long term disaster, death, and chaos. For the purpose
> of winning, it doesn't work any better with one's enemies than it does
> in a bike race. Basically we get in their face and take them out.
> That would be my approach.

I think that we have no choice. It really is "us'ns or them".

 

[Benjamin Weiner]

Robert Chung <[email protected]> wrote:
> Benjamin Weiner wrote:
> > (There are statistical methods which try to make sure
> > you get the same answer for fitting y against x or x against y.)

> Which are those, and why would one be interested in that particular
> criterion?

Well, you never know, Greg might be your paper's referee.

There's potentially hundreds of ways of fitting a straight line
to a set of points. I have some references cribbed from a grad
student if you care (but you probably know them already).
Amazingly, there is a controversy in my field right now which is
partly because people are arguing about these methods
(Tremaine et al 2003 Astrophys J 574, 740). If you care about
the values of the underlying relation, then a couple of nice
properties are invariance under changes in scale of x or y and
exchange of x and y; not all methods do this. Me, I just act
dumb and use the routine from Numerical Recipes.

 

[Robert Chung]

Benjamin Weiner wrote:
> Robert Chung <[email protected]> wrote:
>> Benjamin Weiner wrote:
>>> (There are statistical methods which try to make sure
>>> you get the same answer for fitting y against x or x against y.)
>
>> Which are those, and why would one be interested in that particular
>> criterion?
>
> Well, you never know, Greg might be your paper's referee.
>
> There's potentially hundreds of ways of fitting a straight line
> to a set of points. I have some references cribbed from a grad
> student if you care (but you probably know them already).
> Amazingly, there is a controversy in my field right now which is
> partly because people are arguing about these methods
> (Tremaine et al 2003 Astrophys J 574, 740). If you care about
> the values of the underlying relation, then a couple of nice
> properties are invariance under changes in scale of x or y and
> exchange of x and y; not all methods do this. Me, I just act
> dumb and use the routine from Numerical Recipes.

I'm less familiar now with the statistical decision theory stuff than when
I was actively studying it -- but I don't think I've ever heard much about
x-y exchange *as a purely statistical procedure.* The closest thing I can
drag off the top of my head is the principal components stuff--but most of
that is used as an aid to factor reduction rather than an end-all
statistical procedure by itself.

I just looked up the black hole mass-velocity dispersion stuff. It's
interesting for two reasons:

1. When I gave my "straightening scatterplots" lecture about a month ago I
had to search around for data I could use for examples. Next year perhaps
I'll use some black hole data.

2. I can only see the abstract of the Tremaine article (2002) so this is
perhaps both unfair and premature but this reinforces something else that
I occasionally mention in my lectures: that the statistical techniques
developed in different fields are optimized to handle different kinds of
issues. Econometric techniques (and techniques developed in the other
social sciences) tend to be much more focused on dealing with the
consequences of crappy data than the techniques used in the physical
sciences, which focus more on the efficiency of the estimator.

 

[Benjamin Weiner]

Robert Chung <[email protected]> wrote:

> I'm less familiar now with the statistical decision theory stuff than when
> I was actively studying it -- but I don't think I've ever heard much about
> x-y exchange *as a purely statistical procedure.* The closest thing I can
> drag off the top of my head is the principal components stuff--but most of
> that is used as an aid to factor reduction rather than an end-all
> statistical procedure by itself.

Maybe you'd be interested in this:

http://bayes.wustl.edu/etj/articles/leapz.pdf

A rather long saga, which like Proust's could only be brought
to a conclusion by the death of the author.

> I just looked up the black hole mass-velocity dispersion stuff. It's
> interesting for two reasons:

> 1. When I gave my "straightening scatterplots" lecture about a month ago I
> had to search around for data I could use for examples. Next year perhaps
> I'll use some black hole data.

> 2. I can only see the abstract of the Tremaine article (2002) so this is
> perhaps both unfair and premature but this reinforces something else that
> I occasionally mention in my lectures: that the statistical techniques
> developed in different fields are optimized to handle different kinds of
> issues. Econometric techniques (and techniques developed in the other
> social sciences) tend to be much more focused on dealing with the
> consequences of crappy data than the techniques used in the physical
> sciences, which focus more on the efficiency of the estimator.

Oh, sorry. Damn journal publishers. You can get the full text here:
http://arxiv.org/abs/astro-ph/0203468
Many of the references can also be found in full-text version by
searching here:
http://arxiv.org/find/astro-ph

The abstract doesn't capture the statistical-method smack-talk
that is in the body of the paper, although one of their points
is that the difference in statistical methods is not as big as
their colleagues/mortal enemies have claimed. Science and
rbr have a lot in common actually.

It does seem that different fields have very different needs
from statistics. However, in astronomy, we have no shortage of
crappy data, or data with substantial selection effects.

 

[Robert Chung]

Benjamin Weiner wrote:
>
> Oh, sorry. Damn journal publishers. You can get the full text here:
> http://arxiv.org/abs/astro-ph/0203468

Thanks. From Table 1:
http://anonymous.coward.free.fr/temp/velocity-dispersion.png

 

[Ewoud Dronkert]

On Tue, 27 Apr 2004 11:12:43 +0200, Robert Chung wrote:
> Thanks. From Table 1:
> http://anonymous.coward.free.fr/temp/velocity-dispersion.png

So does log(v)-m have an exponential signature?

 

[Robert Chung]

Ewoud Dronkert wrote:
> On Tue, 27 Apr 2004 11:12:43 +0200, Robert Chung wrote:
>> Thanks. From Table 1:
>> http://anonymous.coward.free.fr/temp/velocity-dispersion.png
>
> So does log(v)-m have an exponential signature?

http://anonymous.coward.free.fr/temp/velocity-dispersion.txt

 

[Ewoud Dronkert]

On Tue, 27 Apr 2004 13:41:54 +0200, Robert Chung wrote:
> http://anonymous.coward.free.fr/temp/velocity-dispersion.txt

Thanks, just what I needed to keep my mind off the beautiful weather
outside.

 

[Robert Chung]

Ewoud Dronkert wrote:
> On Tue, 27 Apr 2004 13:41:54 +0200, Robert Chung wrote:
>> http://anonymous.coward.free.fr/temp/velocity-dispersion.txt
>
> Thanks, just what I needed to keep my mind off the beautiful weather
> outside.

In that case, I'll mention that it took me three tries to find a
linearizing transform that works better than log(velocity) vs. log(mass),
better in the sense of goodness-of-fit (which is an excellent example of
why I'm not a fan of using goodness-of-fit as a model selection
criterion). It took me three tries because I screwed up the first try, so
a guy like you ought to be able to do it in two.

The Milky Way seems out of place.

 

[Ewoud Dronkert]

On Tue, 27 Apr 2004 15:45:19 +0200, Robert Chung wrote:
> The Milky Way seems out of place.

It not being a black hole, fortunately, may have something to do with
that. Just guessing though; I hate astronomy, hate it, hate it. (One
less than cyclingforums.com).