fausto
wang
webb

In other words, when he explains why a player got a yellow light, he's not explaining why he thinks the player should have a yellow light, he's explaining why the system gave him one.

More specifically, I have little doubt that he takes pitch count data into count, writing as he does for the folks who brought us the Pitcher Abuse Points system -- and as the author of the definitive book on the subject.

If not, I'm curious as to why 2 years ago, they were described as being pretty terrible, to today as being described as good as it gets.

Providing a lot of raw information does not really lend itself to great writing -- there's rarely a grand argument he's trying to make -- but I quite enjoy reading Will's stuff.  Nate uses logic and reason but also has the benefit of dealing with raw, hard data.  Both guys are pioneers in their areas, but Will's path is more uncharted by far.  I mean, PECOTA is damned impressive, but Bill James was doing projections based on objective data 20 years ago.

Nobody has ever really attempted to cover medical issues, let alone to systematize them, the way that Will has, and while it's unlikely he'll have the kind of impact Bill James has, he is in his own way just as new and radical of a baseball writer.  No doubt he's learned a lot on the fly, and the results aren't always pretty.  Hell, for any of us writing about baseball, the results are rarely pretty, particularly when you start making predictions.  But I'd rather have a guy admit to having to reverse himself than too stubborn to do so.  Wouldn't you?

he would have to have data that changing positions (from a fit 22 year-old SS to a still fit 22 year-old 2B, specifically) leads to a greater injury risk

Exactly, that is what the data shows.  But remember, it's not like he's saying it's a massive risk.  Cabrera has this risk factor, plus some injury history, and he's still only a yellow light.  I didn't expect it and wasn't happy about it, but I find it hard to argue with two mild risk factors making a yellow light.

I have the same skeptical attitude toward that guy with the proprietary defensive metrics, but there are a few differences for me, the first being that that other guy is also really smug.  The second is that there are other defensive metrics out there, while nobody is really doing what Will is doing.

The third comes from being a regular reader, by which you get an incredible sense of the amount of research and validation that goes into what he writes.  He gathers information from bona fide experts (not claiming to be one), processes it, and shares it -- all in massive quantities.  And there is nobody else doing it.

So the upshot is that this particular system is a small part of what he does.  And while I'd like to be able to scrutinize it as you would, that disappointment is minor compared to my appreciation of his expertise, his work, and the information and insight that nobody else is even attempting to provide.

But, I've come to peace with the situation because of two things: one is basically everything jay has already mentioned. Two is the biggest thing, however: it's just baseball. Even if these guys are talking out their rear-ends, what's the worst that could happen? I say find some rational, respected writers and stick with those guys. At very least, it gives us cause to argue and discuss the finer points of something we all love. It's not like we're shooting people into space based on Will Caroll's research data.

It's one thing if this is viewed as another in a long line of blogospher bloviations - it's quite another if you view it as serious research.  If it's just so much hot air then, yeah, it's cute and all - interesting even - but that's all it is: hot air.  If, on the other hand, it's serious research, then it fails on a number of levels, the most important of which is the "propriety" nature of the data - always the first cover for BS - along with a complete absence of statistical analysis coupled with a lack of retrospective criticism.  In other words show me the data, how you crunched it and the historical accuracy of your predictions.

Like I alluded to upthread, if you wanna see how this stuff is really done by degreed professionals who are recognized by the scientific community as experts in their field, take a look at the Framingham study.  It's all over the Internet and there's nothing "proprietary" about it.  That's what serious research on health predictors looks like.