That happens to me a lot.

http://www.tnt.tv/sports/nba/playoffs08_live/

I'm asking because I heard Vinnie mention it's on KCAL today and that means I don't get it out here on the deep.

That's partly because music critics seems to have a scornful attitude toward artists who cover songs they didn't write.

I don't know "Find a River" -- I gave up on R.E.M. years ago because I hated "Monster" -- but assuming it's as good as you say, wouldn't it be great if other singers could unearth it and give it more exposure?

I never much enjoyed Linda Ronstadt's versions of Little Feat songs, but she deserves a lot of credit for bringing them new fans.

P.S. Lately I'm re-stuck on Van Morrison. "It's Too Late To Stop Now," a great live album, was recently reissued. Tremendous performance by a singer with no boundaries.

Help! I need resuscitation!

The Byrds have a nice cover of "Willin'" though, from their Clarence White era. And "Truck Stop Girl."

Folks, there was a time when Linda Ronstadt was as hot as Jessica Alba. Or so we thought back in the olden days.

It's very windy here in South Pasadena, which is also rare.

Hairston's out after on AB.

Stand down.

Heh.

The green necklace.

sometimes I really do need to engage my brain before posting

(palm to forehead)

he looks so lifelike!

It was the final out in Game 3. The Dodgers won 1-0.

(missing Vin)

1. I personally don't use anything from the Fielding Bible (Dewan) or Clay Davenport. I do use some BIS data strung third-hand in my projections, but it's to determine regression populations rather than to adjust the hard data.

2. There is a difference between assigning past wins to players and projecting a player's contribution. I've only briefly dabbled in the former; it is the latter I am concerned with. As an aside, I also don't think that the term soul is particularly accurate or helpful in this discussion, and I don't mean that as some sort of atheistic objection to the concept itself but rather in order to be precise about what we are attempting to measure.

3. I don't think you've presented an argument against breaking down the game into components to express future value in terms of runs and wins. That's really been my main objection to NC, and it's been relevant every time I've criticized one or more of his moves.

4. I've written many, many more words on less significant baseball topics than 2773 or whatever Bob said that 228 was. But we all knew that, probably.

5. I agree pretty strongly that GB, FB, LD is not a particularly accurate way to divide our observations up. I still use them in my work because the full hit location data is not available. I see this as a weakness, not a flaw that invalidates the model. That is, putting each batted ball into one of three buckets does a service to analysis, but it does not do so completely. Analysts like Tango, MGL, Studes, etc. that have influenced me all basically agree that we want the data to continue to get more granular.

6. In my own projections, I regress fielding pretty heavily. I do this in part because the data streams have the flaws you are elucidating. Since I'm not a decision-maker in baseball, I've decided not to put a lot of my time into working on fielding metrics with the extant data. When people disagree with my assessments of defense, I'm pretty deferential.

7. Hitting projections are pretty well done in the status quo. Projecting a pitcher's fielding-independent numbers (K, BB, HR) is also pretty accurate. My point about the distinction between projection and prediction is that I think teams need to be able to determine expected contribution in terms of runs. The point is getting the best estimate.

8. Me personally? I don't think that any modeling of human behavior is very accurate. I don't have a high opinion of the human sciences as presently constituted. That's kind of the reason why I like baseball. Baseball is a very controlled environment, and its incredible number of variables pales in comparison to anything else I spend time thinking about. Baseball is a simple game with a rich and deep set of issues that make analyzing it interesting because reasonable best estimates are often easy to develop. I don't think we can model the day-to-day lives of baseball players very well, and I don't think any team can do that. So the human factor that is missing from my models is completely conceded. My model, as is the case with everybody else's model, statistical or otherwise, will do a bad job in terms of the human factor. The non-human elements in the model, though, are very strong, and it's my opinion that they are considerably stronger (though accurate on the same level of magnitude) than the mainstream baseball assessments. That opinion is based on data, and I don't think that it functions beyond a shadow of a doubt but merely beyond doubt. And in part, the model succeeds because it systematically uses empirical data to figure out how much weight the human factor gets, and then basically regresses the human factor completely to the mean. I think that teams and media types and many fans generally give the human factor too much weight. Even when they weight it properly, they are often using such a small sample that they should be regressing it quite a bit but do not.

9. While I'm frankly not a fan of any Baseball Prospectus stat, I don't know what you mean by arguing that BA/OBP/SLG tell us all. If we know how to read them, then sure, they tell us basically all we need to know about how well somebody hit over the sample. Converting it to runs (or wins) is to make it coherent. Some of the newfangled metrics don't accomplish that, and perhaps I should even say many of them. BP especially I think is guilty of sometimes looking to get a better RMSE and in so doing neglecting the theoretical building blocks of the model. Tangotiger, who personally has been a much bigger influence for me, has really worked hard against such sloppiness, and Brandon (Patriot) has repeatedly done the dirty work to show the egregious instances. However, I think you're using someone else's bath water to justify tossing my baby. I pretty strongly believe in using theoretical models instead of simple regression coefficients. Regression is a tool that can be useful in looking for things, but when you find what you are looking for you need to deploy a theoretical attitude and use rigor. If that rigor leads you to use regression within a model, that is fine. Linear weights are verified by regression analysis, but it is the underlying theoretical work that convinced me to use them.

10. Baseball is a team game, and I don't analyze teams without looking at the interrelation of the team's elements. But the batter-pitcher matchup is between two individuals. The teams impact every PA, so there is no illusion that it is just the two individuals: there is fielding, defensive positioning, game state/score/etc., the catcher, runners on base, etc. The point of breaking down a team into its individuals isn't to argue that it is not a team game or that analyzing individuals alone will suffice. It's just a question of whether or not we want to try to analyze the individuals, and I don't see how anybody could argue that we shouldn't try to if we're analyzing baseball. Very much of the batter-pitcher matchup can be understood by understanding the skills of the individual batter and pitcher in the context of the spread of skills in baseball as a whole. It is folly to argue that the .333 OBP means that the pitcher is dominating. This is only the case if the game was created with the theoretical goal of a .500 OBP. I could easily counter by arguing that with about 4.5 runs per game, there is a perfect balance: on average, the offense scores a run every other inning. Who dominates who in absolute terms is something that may be useful for historical comparison, but in projecting player contributions we are really looking to evaluate players relative to the population of available players. Perhaps you can use physics to show that the pitcher 'dominates' the batter-pitcher interaction, but this is not the sense of domination that I mean. By domination, I mean how well they do relative to the population. Excellent starting pitchers are not so dominant that they turn all hitters into bad hitters: they are so dominant that they turn good hitters into average hitters, average hitters into bad hitters, and bad hitters into terrible hitters. The intricacies of modeling one PA are such that I will never be confident in the ability to predict an at bat, and I will never have a projection for a PA that is beyond doubt. But in the aggregate, the model is well-constructed. So for a usage like giving a flexible estimate of a player's talent level in wins above replacement, I think what I've got is just fine.

I imagine this game is going to get killed in the ratings tonight by the Lakers game and AI.

Maybe not, this Lakers game sucks so far.

Getting killed.

http://en.wikipedia.org/wiki/List_of_fallacies

In particular --

* Fallacy of Nirvana (Kurt Cobain would approve?), i.e. because VORP, etc. are not perfect they are therefore useless.
* Argument from ignorance (because he can't conceive of a means to assign win shares in a team sport that therefore such a means does not exist).
* Appeal to authority

And long-winded, too!

Lakers were down 20.

I called my friend to whine.

14-0 run since. You're welcome

Weed, whites, and wine.

I spent an evening photographing Bonnie Raitt in 2000-2001. Her father, the late John Raitt, was being honoured that evening. I had a couple of moments alone with her, and all I was able to do was talk about Lowell George, and the fact that I had admired his work for so long and knew that she appreciated it, too.

She immediately turned into a blushing schoolgirl, and said that he was her "best friend." I confessed how much I loved his vocals ("honey-in-the-sludge"), his songwriting, and his slide playing -- she told me that I'd never even heard the half of it.

One cool lady.

oh no ..... is he OK?