Assorted links

Comments

"5. Gary Marcus on what AI still cannot do."

The Hector Levesque paper is a good read: http://www.cs.toronto.edu/~hector/Papers/ijcai-13-paper.pdf
Slides: http://www.cs.toronto.edu/~hector/Papers/ijcai-13-slides.pdf

I would be more inclined to read it (though I did read the abstract) if he showed any awareness whatsoever that the AI community tried to achieve real intelligence via iterative rules (see "Godel, Escher, Bach") and neural networks. They failed miserably. The move to expert systems and "big data" was a reaction to these failures. They wanted to achieve something useful, and I would say that they have.

Comments for this post are closed

Comments for this post are closed

4. "How often do you inspect little boys' testicles?" is what they should have asked. I've made the joke before, but seriously, should I take it as pure coincidence that the most intrusive is also the most dysregulated? Anyway, at this point, everyone should see this one coming (not just limited to this area, the marketing isn't the terrain):
"pointing to studies showing that such programs, on average, save employers little, if anything, in healthcare costs and may even increase spending by forcing workers to undergo extra testing and schedule additional doctor visits."

So, you are arguing that individuals have no control over their health status, so insurance is required to pay for the unpredictable random health problems?

Yeah, mulp, that's exactly what I argue all the time.

Comments for this post are closed

Comments for this post are closed

Comments for this post are closed

Penn Sate really should hire a nurse to do the testicle inspections.

"I heard that you were feeling ill:
Headache, fever, and a chill.
I came to help restore your pluck,
'Cause I'm the nurse who likes to ...."

You could really cut unnecessary spending on health care if every
exam began that way.

Comments for this post are closed

Comments for this post are closed

Sanduskey has the time. In fact, since the article insinuates the medical tax is to pay for the Sanduskey awards one might call it the perfect solution.

Comments for this post are closed

Comments for this post are closed

#5. Really? Pointing out that computers don't possess common sense is some sort of an insightful observation now?

Comments for this post are closed

The bit about movie reviews is ... similar to professors and students grading something like oral presentations.

One point missed in the article is that not only are the public's averages higher, but their variance is lower.

The same thing happens with grading. Students tend to think other students ... all deserve an A-. And they're very reticent about committing to any specific thing that's wrong.

Unfortunately, evaluating others is an important part of what students are supposed to learn, particularly in business schools. I've actually experimented with giving student graders points for matching my grades, not because I'm necessarily better, but to steer them into making better distinctions. The rate of matching shows that they can actually improve over the course of a semester.

This makes me think that reviewing — of movies (or orals) — is a craft that can be learned, but which isn't exercised enough.

What is the empirical basis for saying a movie is good or bad? It is after all entertainment, a way to waste a couple of hours. Maybe the empirical standard is that someone enjoyed it. If a critic doesn't enjoy it and the majority of movie goers do, maybe the critic is wrong in judging the movie as good or bad. The opinion of the critic is simply one more tick on the minus side.

"Panaflex? That's a ten point deduction, Mr. Spielberg. I specifically told you to use Arriflex cameras."

Comments for this post are closed

Comments for this post are closed

Decent comment. I give it an A-.

Comments for this post are closed

Reviewers see too many movies, so they over-favor "edgy" movies over what they regard as shop-worn formulaic fare. The problem is "edginess" is, well, on the edge, meaning there isn't much audience for it. The reason the shop-worn formulas are shop-worn is because they put butts in the seats, at least if the execution isn't horrible.

All of these are fine explanations, but isn't the most straightforward explanation that your average movie goer self-selects the movies he or she thinks he will most like, while critics have to see a lot of movies regardless of whether they expect to like them.

Comments for this post are closed

Comments for this post are closed

Comments for this post are closed

What interests me about Catherine Rampell's piece (#2 above) about critics' and audiences' ratings of movies is this: Ignore the fact that on some arbitrary scale the ratings are not identical. Look at her chart. It appears that (at least aggregated by genre) there is a *very high* correlation between critics' ratings and audiences' ratings. One could conclude that the relationship is given by

A = a + b*C + e

with a very high R-sq (where A is the audience rating, a and b are positive constants, and e is random variation).

To me, that is both remarkable and somewhat consoling.

Comments for this post are closed

#6. Great, solid article.

True, but I wonder if Ms Oster is a Bayesian

Comments for this post are closed

Comments for this post are closed

6. Poor Ms. Oster, it's only starting. Once you actually give birth, the amount of fuzzy science forced on you authoritatively by people who poorly understand what they are talking about multiplies geometrically.

She's gonna have fun.

One of my new policies is that I get to punch those people straight-blast square in the nose. I don't actually do so, but I am entitled to.

"I am entitled to" is the phrase that starts every item of The Narcissist's Manifesto

Andrew's version is heavy with mentions of bjs

I'm not sure what good a manifesto would be without them.

Comments for this post are closed

Comments for this post are closed

Comments for this post are closed

Comments for this post are closed

4. One big issue, the organizing protesting Prof talks about civil disobedience -- this is totally inappropriate and insulting to people who have truly engaged in that practice. Let's not compare having to pay more for a product if you don't give up your privacy to being jailed, injured, or killed to protest unjust law.

Setting aside that, though, and the joke angle, I've got to ask -- if someone signs off that he's done the exam monthly, then he gets a tumor that is advanced when the doctors start treating it, is he going to face some kind of penalty? "Surely, professor, if you had been checking regularly like you promised you'd have caught this sooner" sort of thing? Seems like this trend is ripe for blame the patient politics and will come in handy when rationing kicks in. We legitimately, for example, deny heart transplants to people still chain smoking -- there are only so many hearts. Will it become normal to deny chemo to patients who don't follow the rules like they are supposed to, once we move to government medical financing entirely?

O.k., I'll put away my tin foil hat now. For now.

Comments for this post are closed

#2 -- Didn't we discuss it here around the time everyone was busy bashing Seth MacFarlane for noticing boobs when he sees them?

You're right, that's when we agreed that context and expectations are central to an audience's evaluation.

Comments for this post are closed

Comments for this post are closed

I'm sure Penn State is taking on major exposure on their health plan, and any and all efforts they can undertake to push reinsurance to higher and higher limits saves big time money in stop loss premiums. I've done some consulting on self-funded groups and I can reassure you no matter how well-run their program is, they desire improvement.

But the new guidelines for Wellness Programs issued this summer may negate much of Penn State's programs. The article doesn't state what criterion for determining penalties, any program beyond mere participation must have reasonable alternative standards that must be made available to almost everyone who finds the program onerous. In their "alternative" program (which is usually mere participation), they can still obtain the full benefit as those who successfully completed the standard program. The result can be quixotic: A smoking cessation program starts in your office, offering a 40% premium reduction for everyone who is smoke free for 6-months (assume, please, no cheating). Someone complains that the requirement is too onerous and HR offers him an alternative smoking cessation class he must attend for 6 months. Mind you, he can still smoke, he just has to go to class. Upon completion of the class, he is entitled to the 40% premium reduction.

Comments for this post are closed

The article on small business lending had very little to distinguish lack of demand for loans because of, uh, lack of demand, from lack of supply because of banks becoming more risk averse (including more pessimistic about demand for SMEs' products).

Comments for this post are closed

Film rating has two kinds of selection bias: (1) the public goes to films they expect to like and critics go to a larger variety; (2) film critics choose their profession because they like films more than the typical person. . The observed bias--not that large--could be a net effect or due to a third factor. It tells us little about the size of (1) or (2).

Comments for this post are closed

I'm still learning from you, as I'm improving myself. I certainly liked reading all that is posted on your site.Keep the information coming. I enjoyed it!

Comments for this post are closed

Comments for this post are closed