Here's the worst thing... someone here obviously tipped him off to this. So who's the traitor? My money's on crashburn. Woody, run an investigation

Actually, I guess he probably just noticed several people from bsg clicked on phorever's link and he decided to check it out. Still, Woody, run an investigation, this is the same thing we did with fugees.

the call is coming from INSIDE THE HOUSE!!

jeff2sf wrote:Here's the worst thing... someone here obviously tipped him off to this. So who's the traitor? My money's on crashburn. Woody, run an investigation

For those of you who have never seen an academic pissing contest, you now have a front row seat to a very close approximation. All that's missing is the whipping out of the vitae.

TenuredVulture wrote:For those of you who have never seen an academic pissing contest, you now have a front row seat to a very close approximation. All that's missing is the whipping out of the vitae.

hahaha.

the book thread is a mindboggler. but its always nice to see people have a civil argument. i'm learning something from it.

tangotiger wrote:

he and his gang (i was going to mention them in my previous post... i agree, they are more annoying than the man himself) will never, ever forgive bp for the paywall and secret formulas.

...

tango's forward modeling is actually too simplistic for my tastes.

A Markov chain is too simplistic?

sorry to lump you in with those posting on your blog. i do realize that for you in particular the main issue has always been the secret formulas.

results from markov chains and other monte-carlo methods can be very misleading when applied to problems with more than 4 or 5 significant contributing paraemeters. it is a little known (because the proof was originally only in russian) fact that when there is enough noise in a (relatively) small data set, very different models can fit the data equally well, and monte-carlo methods might find a good model, but not the best model. and that's when everything is linear. if things are nonlinear, the situation can be much worse. to overcome that you need to turn to advanced simulated annealing or genetic algorithms, or neural nets, combined with some good theorizing from first principles of competitive human behavior. (something at which matt has more expertise than most).

i'm concerned that even the vast database of baseball events and run-scoring outcomes isn't good enough to get a reliable model of real run-scoring/preventing skills or of the run value of an event from either inversions or monte-carlo forward modeling procedures that "fit" that data. lots of noise from the whole curved ball - curved bat interaction combined with lots of important parameter combinations that don't actually recur that often. by parameter combinations i mean pitcher type + batter type + situation, with each type needing maybe 10 (or more) parameters to describe a player thoroughly so as not to miss out on any real and relevant talent. a good fit to the past can still turn out to be just plain wrong.

y'all have a great, great dataset to play with. and these day you can build yourself a supercomputer for next to nothing. so i say throw the computational kitchen sink at the problem and see if the model space is really as simple as has been assumed. markov is a step in that direction, and a big one, but only a step. when we took the next step in seismology, we found out that those annoying geochemists had been right and that there was an order of magnitude more complexity in the earth than we had been assuming.

TenuredVulture wrote:For those of you who have never seen an academic pissing contest, you now have a front row seat to a very close approximation. All that's missing is the whipping out of the vitae.

not there yet. if tango replies in usual form, doubling the length of his previous post, with a whole bunch of model results thrown in, together with a deconstruction of the concept of statistical significance... then you've got a real nerd-fight on your hands, but still not an academic pissing contest. academic pissing only starts when matt and i start posting references to relevant academic journal articles, some of which we've authored....

phorever, do you really live in the netherlands?

jamiethekiller wrote:phorever, do you really live in the netherlands?

yes. why?

This is AT LEAST as good as when Bernard-Henri Levy got caught referencing Jean-Baptiste Botul.

Well, maybe not as funny but more informative. And since it concerns baseball and not philosophy, more fun.

tangotiger wrote:
In the end, the entire point of the debate is to advance knowledge to the point where the interested reader will learn something and can move forward in some direction.

This is the best sentence in the whole debate.

It would be more interesting if I understood what they were talking about but my Monte Carlo forward modeling skills are a bit rusty since I lost my white tuxedo back before the war.

Good stuff nonetheless.

Sorry to go slightly OT, but how does one get access to the stats you'd need to analyze things like this. I mean if you really want to go nuts and get every pitch speed and everything. I know most of that for recent years is in here: http://gd2.mlb.com/components/game/mlb/ but I'd think I'd be banned from mlb.com if I tried to download all of that.

MattS wrote:I think the reason is that you don't ever concede when they are right,

That is simply not true. Please find me an instance when I should have conceded but didn't.

and the debate generally descends into methodological assumptions.

You say descends, and I say ASCENDS!

I have a hard time with the argument that BP doesn't want their readers to learn anything. I would argue that BP is as responsible for casual baseball fans knowing sabermetrics as anybody but Bill James and Michael Lewis. Using VORP instead of RBI is a bigger gap than VORP to VORP with EqR instead of RC.

They don't want their readers to bother to learn the problems with their metrics. That is the limitation of what I say they don't want their readers to learn about and nothing more.

It's not about stripping things down, because if you used BP's methods and BP used your methods, people would use EqA and VORP because that is the way to get linked on your blog. I'm not talking about elite sabermetricians, I'm talking about still learning ones. It's easier to get noticed if you get linked on your blog.

You mean if I used a substandard metric, then people would be using those? No, if I were to produce something that had holes in it, I'd be called out on it, and my influence would decrease substantially. I will definitely disagree with you here. It almost implies something fairly negative about the readers of my blog, doesn't it?

It makes assumptions just like regression does. It makes assumptions that are not all accurate but are close enough in many case. Regression makes assumptions that can be manipulated to your favor if you know how to manipulate the correlation between the regressors and the noise term.

Sure Markov chains make assumptions. But I still wouldn't call it simplistic.

Less are subscribers because of your negativity towards BP. You did a poll and a large fraction weren't.

I think something like 30-35% of my readers. Are you suggesting I am UNFAIRLY negative? Or am I fair? If I am fair, then there is no issue is there? If I am unfair, then please cite your examples, because you are casting me in a negative light in this case.

Ultimately, I don't see how you can consider yourself a serious sabermetrician unless reading sabermetrics isn't about doing a literature review. And you can't complete the literature review of barely any topic in sabermetrics without a subscription to BP.

I disagree greatly here.

Most topics are hard without reading The Book too, but BP is part of any literature review, and seeing how you encourage that type of thing, you should encourage subscribing.

I encourage readers to read any good sabermetric study. If that means subscribing to something, then that's what I'll do. But, I'm not going to encourage more than what I would do.

How about instead of 'behind the paywall' you say, 'complete your literature review at three cents an article.' I think that BP would have way more subscribers if people didn't refuse to pay out of protest.

They have 10,000 subscribers! Baseball-Reference.com's Play Index has under 1000. I simply don't believe there is a protest. Boy, we definiteily don't see the same things the same way!

It's a typical irrationality of economic decision making where people won't pay for something out of principle like that when their utility for the articles is much higher than what they are paying for them. Many people don't want to pay for baseball on the internet. They'll spend money on hats and tshirts and everything else baseball related, but they have an aversion to paying for something online. This is exploited when you convince people they are getting ripped off.

Whoah. I hope that is the royal you, and not me in particular. In any case, I think BPro is wildly successful. 10,000 paying customers is enormous.

Regardless, the money is spent on a way of his choosing, and therefore the incentive to make money is there all the same.

Wrong. If it was up to MGL, he’d have sold The Book for no profit, or had it available for all to download for free as a PDF. He has no monetary incentive to sell The Book. His only goal was to make the knowledge available for anyone with no barrier, just like all bloggers out there. But, he’s got partners, and it not up to him.

It's an advertisement for the book in the sense that it encourages people to buy the book. That you encourage people to read it through Amazon is great, but I never figured out how to do that so I ordered a used copy.

As long as you acknowledge that a very tiny percentage of the site exists to sell The Book, then we don't have an issue. The website exists as a blog to foster discussion about baseball and sabermetrics.

phorever wrote:results from markov chains and other monte-carlo methods can be very misleading when applied to problems with more than 4 or 5 significant contributing paraemeters.

The Markov chain or simulator is only as good as the parameters it's identifying, and the way they treat themselves dependently or independently. Yes, limitations, but not simplistic.

For example, I have the perfect modeler here:
http://tangotiger.net/markov.html

It's perfect because it is simply a mathematical construction that uses recursion to calculate how many runs will score. It can't break because it uses all the principles of baseball, actual baseball. HR clears the bases, some outs advances runners, etc. This is UNLIKE all other run estimators which try to match what happens, but invariably breaks down. For example, do all run estimators for 1 run for a HR? No. But, this is an actual principle of baseball.

Now, for ease of programming, my modeler has a built-in limitation (no outs on base). Yes, it will limit you if you need to have that.

But, it acts as a baseline. And with the baseline, you can then start figuring how things are affected. What happens if you go from 30% to 50% to 3b on a single? Well, now you know. It's right there. All other modelers won't know, because they try to build something as an estimate. Not so the MArkov chain (or simulator).

A simiulator is more powerful because you can program far more dependencies. If MAtt for example finds that pitchers that give up lots of walks happen to allow fewer hits per batted ball to RHH, the simulator can handle that. The Markov chain, once you go past 4 or 5 dimensions, is a bear to program and tweak.

In any case, I don't know that we actually disagree on anything.

By the way Matt, regarding encouraging people to subscribe, I did say this:

http://www.insidethebook.com/ee/index.p ... _new_bpro/

I love PizzaCutter as much as a virtual saberist can love another virtual saberist.. Colin is great and I knew that Jeff was a perfect match for someone (BPro, B-R, or Fangraphs), and Idol or no Idol, it was a given that both these guys would make it to BPro if that’s what either wanted. Tommy is great and is a cross of Neyer and Phil. Eric has been excellent wherever he was. This is about as good as a Tango-approved team that BPro could have brought on board.

I mean, is there anyone anywhere that wrote something as glowing as I did? To say that I am not encouraging anyone basically is unfair.

Now, either concede the point, or be kind enough to show me where Baseball Prospectus has encouraged people to buy The Book anywhere close to what I just did. Or even at all.

tangotiger wrote:Please find me an instance when I should have conceded but didn't.

Off the top of my head, Sky Andrecheck's article on $/win and using regression. A few of us were telling you the regression would be unbiased and you were focused on randomness of the dependent variable which wasn't an issue. QERA as a useless metric seems exaggerated too given these tests. EqA tested slightly better than wOBA according to Colin's article, too. The point is that you don't tend to concede when you're wrong. Either you posted ten strong opinions a day and all of them are right or you underestimate how often you are wrong.

They don't want their readers to bother to learn the problems with their metrics. That is the limitation of what I say they don't want their readers to learn about and nothing more.

They don't break down the details. Read the comments and hear the complaints about gory details. We all say metrics have limits and reguarly point out exceptions. It's a different audience.

You mean if I used a substandard metric, then people would be using those?

Some people would, yes. Not everyone understands the details, and people default to trusting you on it. If you don't think that has to do with what you do make bloggers more famous (myself included), that's a naivete about human nature that I don't think is consistent with your political or sabermetric beliefs.

Ultimately, I don't see how you can consider yourself a serious sabermetrician unless reading sabermetrics isn't about doing a literature review. And you can't complete the literature review of barely any topic in sabermetrics without a subscription to BP.

I disagree greatly here.

Midseason trade value calculators a la BtB are idiotically wrong. Yes, idiotically, as in they aren't even close-- the marginal value of a win is MUCH higher for a team that is still in a playoff race in July. My first paid BP article about Roy Halladay last July broke this down. But everyone went on using them anyway, because they don't treat BP as required reading. The stuff that Colin and Russell are doing at BP looks like it will be essential in evaluating managers and fielding. Ignoring it is incomplete. Nate Silver evaluated tons of different topics that you see bloggers reinventing the wheel trying to redo right now. Davenport Translations are some of the most important MLEs out there. Other projection systems may have caught up or be catching up or whatever, but what Nate Silver did for projection is very significant and the articles he wrote should be required reading for anyone trying to do projections. There are tons of very important articles on a variety of topics that people should read.

They have 10,000 subscribers!

I don't know if the actual number is public.

Regardless, the money is spent on a way of his choosing, and therefore the incentive to make money is there all the same.

Everyone has goals other than forwarding knowledge. MGL wants knowledge to be widespread, but to think he is indifferent between being right and wrong is to call him unhuman. Everybody would rather be right than wrong. Certainly people who argue sabermetrics like we all do.

tangotiger wrote:By the way Matt, regarding encouraging people to subscribe, I did say this:

http://www.insidethebook.com/ee/index.p ... _new_bpro/

I love PizzaCutter as much as a virtual saberist can love another virtual saberist.. Colin is great and I knew that Jeff was a perfect match for someone (BPro, B-R, or Fangraphs), and Idol or no Idol, it was a given that both these guys would make it to BPro if that’s what either wanted. Tommy is great and is a cross of Neyer and Phil. Eric has been excellent wherever he was. This is about as good as a Tango-approved team that BPro could have brought on board.

I mean, is there anyone anywhere that wrote something as glowing as I did? To say that I am not encouraging anyone basically is unfair.

Now, either concede the point, or be kind enough to show me where Baseball Prospectus has encouraged people to buy The Book anywhere close to what I just did. Or even at all.

I thought Will Carroll wrote a blurb? Didn't Ken Funck cite you a bunch of times on SOMA?

That's one example. You have since said you didn't see evidence of change.

Barry Jive wrote:or seeing Tom $#@! TangoTiger post a gigantic $#@! response on your diddly poo Phillies board

Honestly, I don't even know who Tom TangoTiger is. I thought this was some new poster. I guess we should be honored he posted on our "diddly poo Phillies board."

MattS wrote:Off the top of my head, Sky Andrecheck's article on $/win and using regression. A few of us were telling you the regression would be unbiased and you were focused on randomness of the dependent variable which wasn't an issue.

6MM$ per win is wrong, so I can't concede there. I DID correct myself later in the thread because we were talking about 2 or 3 things at the same time.

QERA as a useless metric seems exaggerated too given these tests.

Did I say useless? I must have simply said that it used the wrong denominator. If I said useless, then I was wrong, and i'll be happy to correct myself on my blog.

EqA tested slightly better than wOBA according to Colin's article, too.

I said EqA is pretty good, and breaks down only at the bottom end (pitcher-level hitting). I said that EqA is enormously complicated, when it's basically just linear weights. You are misremembering.

The point is that you don't tend to concede when you're wrong.

I concede when I'm wrong. You still haven't showm me anything. Just that I have too strong an opinion on occasion, or that I am not clear on why I have that opinion. That's as far as you can go. If you want to go further, you'll have to quote me.

They don't break down the details. Read the comments and hear the complaints about gory details. We all say metrics have limits and reguarly point out exceptions. It's a different audience.

This is different when you new guys came on board. My complaints regarding BPro practices are not necessarily ongoing.

Some people would, yes. Not everyone understands the details, and people default to trusting you on it.

But they default because others vouch for me. It's like being Roger Ebert.

If you don't think that has to do with what you do make bloggers more famous (myself included), that's a naivete about human nature that I don't think is consistent with your political or sabermetric beliefs.

Call me naive then. I truly believe that eveyrone who reads and posts at my blog do it for the pure knowledge transfer, and nothing more.

If you are suggesting that you, or others, come to my blog to somehow make yourself a name, and even not be completely truthful in your posts in order to advance a personal cause, I would be highly disappointed.

Otherwise, I believe that everyone posts there for the greater good.

Midseason trade value calculators a la BtB are idiotically wrong. Yes, idiotically, as in they aren't even close-- the marginal value of a win is MUCH higher for a team that is still in a playoff race in July.

Yes, that is totally correct. That's been shown like for a few years right?

My first paid BP article about Roy Halladay last July broke this down. But everyone went on using them anyway, because they don't treat BP as required reading. The stuff that Colin and Russell are doing at BP looks like it will be essential in evaluating managers and fielding. Ignoring it is incomplete.

Matt, Eric, Pizza, Colin, and the other new guys are DEFINITE required reading. There's no question about that.

At the same time, Dave Allen, Max Marchi and several others are also DEFINITE required reading.

But, each area acts like its own universe, not acknowedging each other. I see the comments at BPro in articles about how some things are so new to them. Well, that's a bad job at BPro for not highlighting what's happening elsewhere.

And equally bad job at non-BPro site for not highlighting BPro. Studes, however, is different, in that he has OFTEN posted to BPro articles (he's a subscriber). BPro does not repay him in kind in the least.

Can we agree that all sides can do a better job of promoting everyone? If so, how would you recommend we help?

I for example offered to do a "Mike Silva" type of artcile for Bpro, a series that was pretty widely appreciated by the newbies. They turned me down. It would have been really really good.

Nate Silver evaluated tons of different topics that you see bloggers reinventing the wheel trying to redo right now. Davenport Translations are some of the most important MLEs out there.

Nate's done great work, and if people are reinventing it, then they shouldn't. AT the smae time, why can't BPRo simply open up the archives for past articles? I mean, people don't subscribe for the archives do they? And once they are open, we can link to it, and get more traffic to you. Why can't that happen?

And I use DT MLEs all the time, like I do for Rally and ZiPS. Is someone saying otherwise? If they are, then I'll be happy to straighten them out.

I don't know if the actual number is public.

It is now. It was 10,000 near the start of BPro. And, a little bit of figuring out by me, using public data. tells me that they might be somewhere close to 20,000.

Everyone has goals other than forwarding knowledge. MGL wants knowledge to be widespread, but to think he is indifferent between being right and wrong is to call him unhuman. Everybody would rather be right than wrong. Certainly people who argue sabermetrics like we all do.

All I said is that he has no monetary incentive. Whatever incentives he has is not money-based. That's all I'm saying.

I thought Will Carroll wrote a blurb? Didn't Ken Funck cite you a bunch of times on SOMA? That's one example. You have since said you didn't see evidence of change.

Yes, Will wrote it.. on his blog, not at BPro. Dan Fox as well. Would you believe that they actually took out a mention of The Book as honorable mention by Dan Fox in his book review of Echoes of Green (or whatever it was called). I have proof. And Joe Sheehan took out a mention of me in his Jack Morris article? Joe wrote to me specifically after to say that it was his call to take out the reference.

I mean, I even ask for the simplest thing like linking to my Fan projects, something that BackSheGoes and SBNAtion and Fangraphs and HArdballTimes and Rob Neyer do without quetion? BPro? Sheehan did it several years back, and that's it.

What is SOMA?

***

I'm willing to bridge whatever gap exists. My hand is extended. Tell me what you want me to do, and tell me what you are prepared to do, and let's do it.

Let's move forward, instead of standing still with all this yapping.

Bakestar wrote:the call is coming from INSIDE THE HOUSE!!

+1 favorite commercial ever