How to be misled by polls, pols edition

The NYT has this fairly trivial piece on Hilary Clinton’s polling numbers post-Emailgate. I know I’m nitpicking here, but this paragraph bothers me a lot:

This approach is misleading for two reasons. First, CNN gives its own poll too much weight compared with polling averages, creating a perception of decline in Mrs. Clinton’s ratings that could be a result of sampling error.

What bothers me most is that the article continues. Why does this bother me? Because if there is sampling error, then no other reason matters. The poll should be thrown out. Period. End of story.

Furthermore, the author suggests that there could be sampling error. Ok…but is there sampling error? Is this a baseless accusation? If we’re lining up a bunch of “coulds” for why the polling could be incorrect, then the article could go on for dozens of pages. This strikes me as ridiculously sloppy and poorly thought out.

Is there sampling error? Or not? That’s the real story.

-JD Cross

Complex systems are complex

I use the previous phrase often and in many circumstances. While it seems redundant, trivial, and self-descriptive in an obvious way, the connotation that I have when I use it is that people often recognize that systems are complex in the inner workings and/or relationships with things other than themselves, but that they don’t really understand what it means for something to have complex relationships. For example: you recognize that a car is a complex mechanical object, but you don’t really have any idea what it means for it to be complex, nor how complex it is; Complex is often a trivially useless descriptor.

My connotation in saying that complex systems are complex is that there are profound things about such systems that are not currently understood (by anybody, even, in the example above, a skilled auto mechanic), and, in some stronger and very important cases, things about such systems that cannot ever be understood.

In introductory physics, you solve a lot of problems having to do with ropes falling off of tables and objects accelerating from rest. You might even do a fun momentum problem in which a running back collides with a linebacker. But do you ever do problems with more than two objects interacting? No. That’s a many-body problem and it is saved for physics majors to tackle in mechanics or advanced mechanics. But how many bodies constitute solvable problems versus very hard or impossible problems?

Here’s an apocryphal story of Daniel Patrick Moynihan asking a Harvard chemist this question. (also, see here.)

A while back, one of Harvard’s great chemists was discoursing on what he called the “many-body problem,” a condition in which the number of variables interacting with one another in any given situation makes that situation extraordinarily complicated and difficult to fathom. I asked in what range of numbers this “many-body problem” begins. A somewhat suspicious glance was returned. Did I really not know? Apparently not. “Three,” he replied.

In the physical sciences, complexity starts at around 3 interacting objects. I always maintain that any system that involves humans is complex. Why? Because humans don’t even understand how themselves, let alone how one human interacts with any number of other humans. We don’t understand our own behavior; sociologists don’t understand our behavior, psychologists don’t understand our behavior, and certainly politicians don’t. We don’t know what policies will work in advance, and it’s rare, and we’re exceptionally lucky if we even can collect the data to understand what policies have worked a posteriori.

-JD Cross

SUGAR!

The sugar industry and the US government were in bed with each other and the loser was everybody’s teeth.

It is not surprising to me that there are people who want to influence (lobby) what the government does, what it regulates, how it taxes certain things, what kinds of recommendations it makes. Such is the nature of power and influence. It is inevitable.

What is surprising to me is that most people that I know think the government is good and want more of it while at the same time decrying lobbying. People: problems of power and influence (lobbying) happen BECAUSE there is government. To the extent that there is more government with more power, the problems can only get worse. There is no other option.

It’s dramatized in reality in the link above, and in fiction in Evolve.

-JD Cross

Stock picking

The future is hard to predict, especially if you pretend that you can see it.

The blind forecaster beats a panel of industry “experts“. This isn’t really news, as reports like this are published as articles or books quite regularly. It is good, though, to always remind oneself of how most of economics and all of sociology is NOT at all like physics. There is no determinism. There are no equations of motions.

People, as the actors in the system, mess everything up. Second, are at least equally important, is that the number of variables in the system is fantastically large.

Can you think of any other systems with fantastically large numbers of variables for which people regularly make headlines for predicting the future?

-JD Cross

Awesomeness: robot overlords

Robots keep getting better. And mind control is coming along nicely, too.

Very cool stuff.

-JD Cross

Life imitates art: Clinton or Beal?

Hillary’s use of non-official email is taken straight out of Senator Sheila Beal’s playbook of outrageous, illegal activities.

-JD Cross

A question about net neutrality

Can somebody explain to me how Amazon buying its way to Sunday delivery is any different than a company such as Netflix paying extra to have a “fast lane” on the internet?

-JD Cross

Follow

Get every new post delivered to your Inbox.

Join 161 other followers

%d bloggers like this: