So can we use component analysis to separate the April wheat from the April chaff? Could greater computing power bring PECOTA into that discussion? Perhaps a list of possible predictors, things that could be checked? (schedule thus far, health factors via Will, I'm sure there must be numerous other 'usual suspects' that could be rounded up and looked into)

George Santayana, from The Life of Reason, Vol.1: "Those who cannot remember the past are condemned to repeat it."

You are the rudder of sanity. It's a thankless task -- everyone wants instant "analysis" and to discern the big surprises before everyone else -- even if it is only from a handful of events. Your relatively long-term outlook and measured take on the game is a welcome antidote to the aimless hyperbole readily available elsewhere.

Stay the course and continue to write about the perils of "knee-jerking" in an entertaining manner. That's one of the things I've come to enjoy about April baseball.

There is an implicit acknowledgement of the small sample size problem in traditional lore when, as you noted in an earlier article, a win in April is not treated as seriously as a win in August. (Though of course it would be better to look at factors such as strength of schedule when deciding if a team's April record is indicative of a sustainable level of play rather than just dismissing it entirely.)

One key problem with repeating yourself ad nauseum is that after a while, I tend to skip over articles that seem to cover the same old ground, and can easily miss any new aspects contained within. I'd rather you work more on fleshing out new points of discussion, rather than go over what I've read many times before (in the same month, to make it worse).

I love your approach for this article, Joe, and I love that you used your own quotes.

I don't know that it's your "never ending pleas" that people object to so much as the meta-journalism that's making up more and more of your output. Most of what you have to say about the media ("they get hysterical over steroids," "they see patterns where there are none") is so typical of media-criticism about all media (as opposed to just baseball media) as to come off as kind of facile. Everyone knows that mainstream media in this and all other countries simplifies the story to the point of parody and appeals to the lowest common denominator. We don't need you (who, I am the first to admit, is a top-notch baseball writer) doing your best Fire Joe Morgan without the wit impression. It just seems like you're banging your head against a wall that your target (and somewhat captive audience) climbed a long time ago.

So, who are the biggest surprises that you don't expect to be maintained? As a Ranger fanatic, there are some surprises that have made the AL West look a bit upside down. Will they be sustained?

1) Millwood managing an ERA of 2.78 and averaging over 7 innings per start with a BABIP of .220

2) Andruw Jones hitting .393 all year. Ok, that won't be sustained.

3) Frank Francisco not allowing an earned run. That probably won't be sustained either, though stranger things have happened. Anyone know the longest scoreless streak by a reliever? Francisco is up to 26.1 innings without an earned run since August 22 of last year. He hasn't allowed an earned run or unearned run (though he has allowed 2 inherited runs to score). Eck's best during 1989 was 22 innings.

We all have a tendency to accept as true the trends that confirm our own hypthoses, or those we would like to believe, while rejecting as small samples the trends we don't like or don't want to believe.

For example, as an Orioles fan, I am certain that Adam Jones' April is the beginning of the emergence of a star player, while Pie's April is confirmation of the suspicion we had that he will never hit in the big leagues.

Then again there are the trends you can count on because they are a continuation of what has come before. Markakis really is that solid, and the O's starting pitching really is that bad.

Sample size is always a bit quirky. Players can have a fluke month, or a fluke season. How much of a season is determined by skill as opposed to luck? Even if you took a player who is well below replacement level, they can still have that one hot month where they look like the best player in the league. Conversely, a star player can have a bad month or a bad season and it is uncertain how much of that is due to a decrease in skill or to bad luck. We can make some guesses or hypotheses, based on the data, on how much "might" be attributable to luck, but we can't ever really seem to "know".

To what extent do Bayesian approaches help you out here?