Category: SciTech
One Small Step Towards a Theory of Pitch Sequencing
by Ken Arneson
2017-07-29 17:55

Three years ago, I wrote an article called “10 Things I Believe About Baseball Without Evidence“, in which I hypothesized that it ought to be possible to develop some sort of theory of pitch sequencing. To me, pitch sequencing is the very heart of the sport, the chess match between batter and pitcher which makes the sport compelling. But for all our progress in sports analytics in recent years, a theory of pitch sequencing — what it is, how it works, which pitchers are good at it, which batters can be fooled by it — seems as distant as ever.

In this article, I hypothesized (without evidence, as the title suggests) that such a theory would involve somehow understanding that the brain of the batter makes predictions for the next pitch based on previous pitches:

I believe that before any given pitch, the batter is in some sort of Prediction State for the next pitch. After each pitch, the batter then moves into a different Prediction State.

One year after I wrote this evidence-free idea, a piece of evidence came in which supported my hypothesis.

Jeff Hawkins and Subutai Ahmad, who work for a company called Numenta which is trying to reverse engineer the brain with computers, published in October of 2015 a paper called “Why Neurons Have Thousands of Synapses, A Theory of Sequence Memory in Neocortex”.

You can read a nice layperson’s summary of the paper here. But I’ll summarize the summary even further.

Memory in the brain consists of cells called neurons. These neurons have different parts, and one of these parts is called “distal synapses”. Up until this point, nobody really had a good idea what these distal synapses were for, because they didn’t seem to do anything while a particular memory was firing. Hawkins and Ahmad theorize that this is because the distal synapses don’t cause the neuron to fire immediately. Instead, they electrically prepare the cell to fire quickly if a signal comes in from a certain direction. And it is this preparation which allows the brain to make predictions about sequences of events. Relevant quote from the paper:

“Each neuron learns to recognize hundreds of patterns that often precede
the cell becoming active. The recognition of any one of these
learned patterns acts as a prediction by depolarizing the cell
without directly causing an action potential. Finally, we show
how a network of neurons with this property will learn and
recall sequences of patterns. The network model relies on
depolarized neurons firing quickly and inhibiting other
nearby neurons, thus biasing the network’s activation
towards its predictions.”

And herein lies the physical foundation of a theory of pitch sequencing. For if Hawkins and Ahmad are correct about sequential learning, it means that there is indeed some sort of Prediction State that the brain is in before each pitch.

Once the brain has seen some sort of sequence of inputs, it prepares itself to recognize that sequence again, and to recognize and react to it more quickly the next time it appears, by being electrically primed to react through this neuronal depolarization.

At this point, it’s important to understand that we’re not just talking about sequences of individual pitches here (a curve followed by a fastball followed by a changeup). It can be that, too, but not only that.

A single pitch in and of itself is a sequence of patterns happening that the brain needs to recognize. It’s a windup, and then a release, and then a ball movement out of the hand, and then a spin which one can perhaps recognize, and then a speed and a directional movement of the ball in one way or another.

Each of these patterns and sub-patterns and sub-sub-patterns that compose a pitch are represented in the brain at the neuronal level. As a batter observes sequences of (sub-)(sub-)patterns, the brain automatically prepares itself to see those sequences again by depolarizing the neurons to make them respond faster to these patterns. Thus, from the pitches it has seen in the past, the brain moves into a sort of Prediction State about the pitches it anticipates seeing in the future.

This has the effect, as Hawkins and Ahmad put it, of “biasing the network’s activation towards its predictions”. The batter’s Prediction State has a bias, and pitchers can exploit this bias. The brain is ready to react to some patterns, which it will react quickly to, but at the expense of inhibiting a reaction to other other patterns, which it will be slower to react towards.

So if you throw three fastballs with the same speed and the same location in a row, the batter’s brain will become more and more prepared/biased to predict that pitch accurately with each subsequent pitch, and the batter becomes more likely to hit the ball hard.

But if pitchers understand what the batter’s brain is biased towards, they can fool the batter by defying that prediction. Throw a changeup to the same location, but with a different speed, and you can make the batter swing too early. The wrong neurons get fired, and the ones that should have fired to hit the ball properly are instead inhibited by the bias, and the batter does the wrong thing.

They say that pitching is an art, and perhaps at this time it is, but there is potential in this information that it could eventually be turned into a science.

* * *

This information doesn’t explain everything about how the brain processes sequencing, obviously. It’s just a initial framework for understanding how the brain learns to understand sequences of events and to predict them. And since we don’t really understand exactly it works in that general case in the brain, we therefore also don’t understand how it works for the specific case of pitch sequencing.

So if we have unanswered questions about the brain like, “how long does this cell depolarization last?” we also have corresponding unanswered questions about pitch sequencing, like, “how long does a batter remain biased towards a kind of pitch once he has seen it?”

The good news is, we can probably answer the second question without necessarily answering the first. There is data that will tell us how much better a batter gets when he sees the same pitch multiple times, either in a row, or in close proximity. Understanding the basic framework of how the brain works can help us ask better questions about pitch sequencing, and to develop useful theories about how it works, even before the neuroscientists figure out precisely how it works in the brain.

Good luck, baseball analysts.

The Data/Human Goal Gap
by Ken Arneson
2016-06-06 14:48

As I was writing a letter to my third-grade daughter’s principal in support of a change in homework policy (a letter which I’ve posted here), it occurred to me I was making a point about a phenomenon that isn’t unique to education at all, but happens in a lot of other fields, too: baseball, business, economics, and politics.

I don’t know if this phenomenon has a name. It probably does, because you’re very rarely the first person to think of an idea. If it does, I’m sure someone will soon enlighten me. The phenomenon goes like this:

* * *

Suppose you suck at something. Doesn’t matter what it is. You’re bad at this thing, and you know it. You don’t really understand why you’re so bad, but you know you could be so much better. One day, you get tired of sucking, and you decide it’s time to commit yourself to a program of systematic improvement, to try to be good at the thing you want to be good at.

So you decide to collect data on what you are doing, and then study that data to learn where exactly things are going so wrong. Then you’ll try some experiments to see what effect those experiments have on your results. Then you keep the good stuff, and throw out the bad stuff, and pretty soon you find yourself getting better and better at this thing you used to suck at.

So far so good, eh? But there’s a problem. You don’t really notice there’s a problem, because things are getting better and better. But the problem is there, and it has been there the whole time. The problem is this: the thing your data is measuring is not *exactly* the thing you’re trying to accomplish.

Why is this a problem? Let’s make a simplified graph of this issue, so I can explain.

Let’s call the place you started at, the point where you really sucked, “Point A”.
Let’s call the goal you’re trying to reach “Point G”.
And let’s call the best place the data can lead you to “Point D”.

Note that Point D is near Point G, but it’s not exactly the same point. Doesn’t matter why they’re not the same point. Perhaps some part of your goal is not a thing that can be measured easily with data. Maybe you have more than one goal at a time, or your goals change over time. Whatever, doesn’t matter why, it just matters they’re just not exactly the same point.

Now here’s what happens:

You start out very far from your goal. You likely don’t even know exactly what or where your goal is, precisely, but (a) you’ll know it when you see it, and (b) know it’s sorta in the Point D direction. So, off you go. You embark on your data-driven journey. As a simplified example, we’ll graph your journey like this:

statsgraph2

On this particular graph, your starting point, Point A, is 14.8 units away from your goal at Point G. Then you start following the path that the data leads you. You gather data, test, experiment, study the results, and repeat.

After a period of time, you reach Point B on the graph. You are now 10.8 units away from your goal. Wow, you think, this data-driven system is great! Look how much better you are than you were before!

So you keep going. You eventually reach Point C. You’re even closer now: only 6.0 units away from your goal!

And so you invest even more into your data-driven approach, because you’ve had nothing but success with it so far. You organize everything you do around this process. The process, and changes that you’ve made because of it, actually begin to become your new identity.

In time, you reach Point D. Amazing! You’re only 4.2 units away from your goal now! Everything is awesome! You believe in this process wholeheartedly now. The lessons you’ve learned permeate your entire worldview now. To deviate from the process would be insane, a betrayal of your values, a rejection of the very ideas you stand for. You can’t even imagine that the path you’ve chosen will not get any better than right here, now, at Point D.

Full speed ahead!

And then you reach Point E.

Eek!

Egads, you’re 6.00 units away from your goal now. You’ve followed the data like you always have, and suddenly, for no apparent reason, things have suddenly gotten worse.

And you go, what on Earth is going on? Why are you having problems now? You never had problems before.

And you’re human, and you’ve locked into this process and weaved it into your identity. You loved Points C & D so much that you can’t stand to see them discredited, so your Cognitive Dissonance kicks in, and you start looking for Excuses. You go looking for someone or something External to blame, so you can mentally wave off this little blip in the road. It’s not you, it’s them, those Evil people over there!

But it’s not a blip in the road. It’s the road itself. The road you chose doesn’t take you all the way to your destination. It gets close, but then it zooms on by.

But you won’t accept this, not now, not after the small sample size of just one little blip. So you continue on your same trajectory, until you reach Point F.

You stop, and look around, and realize you’re now 10.8 units away from your goal. What the F? Things are still getting worse, not better! You’re having more and more problems. You’re really, really F’ed up. What do you do now?

Can you let go of your Cognitive Dissonance, of your Excuse seeking, and step off the trajectory you’ve been on for so long?

F is a really F’ing dangerous point. Because you’re really F’ing confused now. Your belief system, your identity, is being called into question. You need to change direction, but how? How do you know where to aim next if you can’t trust your data to lead you in the right direction? You could head off in a completely wrong direction, and F things up even worse than they were before. And when that happens, it becomes easy for you to say, F this, and blow the whole process up. And then you’re right back to Point A Again. All your effort and all the lessons you learned will be for nothing.

WTF do you do now?

F’ing hell!

* * *

That’s the generic version of this phenomenon. Now let’s talk about some real-world examples. Of course, in the real world, things aren’t as simple as I projected above. The real world isn’t two-dimensional, and the data doesn’t lead you in a straight line. But the phenomenon does, I believe, exist in the wild. And it’s becoming more and more common as computers make data-driven processes easy for organizations and industries to implement and follow.

Education

As I said, homework policy is what got me thinking about this phenomenon. I have no doubt whatsoever that the schools my kids are going to now are better than the ones I went to 30-40 years ago. The kids learn more information at a faster rate than my generation ever did. And that improvement, I am confident, is in many ways a result of the data-driven processes that have arisen in the education system over the last few decades. Test scores are how school districts are judged by home buyers, they’re how administrators are judged by school boards, they’re how principals are judged by administrators, and they’re how teachers are judged by principals. The numbers allow education workers to be held accountable for their performance, and provide information about what is working and what needs fixing so that schools have a process that leads to continual improvement.

From my perspective, it’s fairly obvious that my kids’ generation is smarter than mine. But: I’m also pretty sure they’re more stressed out than we were. Way more stressed out, especially when they get to high school. I feel like by the time our kids get to high school, they have internalized a pressure-to-perform ethic that has built up over years. They hear stories about how you need such and such on your SATs and this many AP classes with these particular exam scores to get into the college of their dreams. And the pressure builds as some (otherwise excellent) teachers think nothing of giving hours and hours of homework every day.

Depression, anxiety, panic attacks, psychological breakdowns that require hospitalization: I’m sure those things existed when I went to school, too, but I never heard about it, and now they seem routine. When clusters of kids who should have everything going for them end up committing suicide, something has gone wrong. That’s your Point F moment: perhaps we’ve gone too far down this data-driven path.

Whatever we decide our goal of education is, I’m pretty sure that our Point G will not feature stressed-out kids who spend every waking hour studying. That’s not the exact spot we’re trying to get to. I’m not suggesting we throw out testing or stop giving homework. I am arguing that there exists a Point D, a sweet spot with just the right amount of testing, and just the right amount of homework, that challenges kids the right amount without stressing them out, and leaves the kids with the time they deserve to just be kids. Whatever gap between Point D and Point G that remains should be closed not with data, but with wisdom.

Baseball

The first and most popular story of an industry that transforms itself with data-driven processes is probably Michael Lewis’s Moneyball. It’s the story of how the revenue-challenged Oakland A’s baseball team used statistical analysis to compete with economic powerhouses like the New York Yankees.

I’ve been an A’s fan my whole life, and I covered them closely as an A’s blogger for several years. So I can appreciate the value that the A’s emphasis on statistical analysis has produced. But as an A’s fan, there’s also a certain frustration that comes with the A’s assumption that there is no difference between Point D and Point G. The A’s assume that the best way to win is to be excruciatingly logical in their decisions, and that if you win, everyone will be happy.

But many A’s fans, including myself, do not agree with that assumption. The Point F moment for us came when, during a stretch of three straight post-season appearances, the A’s traded their two most popular players, Yoenis Cespedes and Josh Donaldson, within a span of six months.

I wrote about my displeasure with these moves in an long essay called The Long, Long History of Why I Do Not Like the Josh Donaldson Trade. My argument was, in effect, that the purpose of baseball was not merely winning, it was the emotional connection that fans feel to a team in the process of trying to win.

When you have a data-driven process that takes emotion out of your decisions, but your Point G includes emotions in the goal of the process, it’s unavoidable that you will have a gap between your Point D and your Point G. The anger and betrayal that A’s fans like myself felt about these trades is the result of the process inevitably shooting beyond its Point D.

Business

If Moneyball is not the most influential business book of the last few decades, it’s only because of Clayton Christensen’s book, The Innovator’s Dilemma. The Innovator’s Dilemma tells the story of a process in which large, established businesses can often find themselves defeated by small, upstart businesses with “disruptive innovations.”

I suppose you can think of the phenomenon described in the Innovator’s Dilemma as a subset of, or perhaps a corollary to, the phenomenon I am trying to describe. The dilemma happens because the established company has some statistical method for measuring its success, usually profit ratios or return on investment or some such thing. It’s on a data-driven track that has served it well and delivered it the success it has. Then the upstart company comes along and sells a worse product with worse statistical results, and because of these bad numbers, the establish company ignores it. But the upstart company is on an statistical path of its own, and eventually improves to the point where it passes the established company by. The established company does not realize its Point D and Point G are separate points, and finds itself turning towards Point G too late.

Here, let’s graph the Innovator’s Dilemma on the same scale as our phenomenon above:

statsgraph3

The established company is the red line. They have reached Point D by the time the upstart, with the blue line, gets started. The established company thinks, they’re not a threat to us down at Point A. And even if they reach our current level at Point D, we will beyond Point F by then. They will never catch up.

This line of thinking is how Blockbuster lost to Netflix, how GM lost to Toyota, and how the newspaper industry lost its cash cow, classified ads, to Craigslist.

The mistake the establish company makes is assuming that Point G lies on/near the same path that they are currently on, that their current method of measuring success is the best path to victory in the competitive market. But it turns out that the smaller company is taking a shorter path with a more direct line to the real-life Point G, because their technology or business model has, by some twist, a different trajectory which takes it closer to Point G than the established one. By the time the larger company realizes its mistake, the smaller company has already gotten closer to Point G than the larger company, and the race is essentially over.

* * *

There are other ways in which businesses succumb to this phenomenon besides just the Innovator’s Dilemma. Those companies that hold closely to Milton Friedman’s idea that the sole purpose of a company is to maximize shareholder value are essentially saying that Point D is always the same as Point G.

But that creates political conflict with those who think that all stakeholders in a corporation (customers, employees, shareholders and the society and environment at large) need to have a role in the goals of a corporation. In that view, Point D is not the same as Point G. Maximizing profits for the shareholders will take you on a different trajectory from maximizing the outcomes for other stakeholders in various proportions. When a company forgets that, or ignores it, and shoots beyond its Point D, then there is going to inevitably be trouble. It creates distrust in the corporation in particular, and corporations in general. Take any corporate PR disaster you want as an example.

Economics

I’m a big fan of Star Trek, but one of the things I never understood about it was how they say that they don’t use money in the 23rd century. How do they measure the value of things if not by money? Our whole economic system is based on the idea that we measure economic success with money.

But if you think about it, accumulating money is not the goal of human activity. Money takes us to Point D, it’s not the path to Point G. What Star Trek is saying is that they somehow found a path to Point G without needing to pass through Point D first.

But that’s 200 years into a fictional future. Right now, in real life, we use money to measure human activity with. But money is not the goal. The goal is human welfare, human happiness, human flourishing, or some such thing. Economics can show us how to get close to the goal, but it can’t take us all the way there. There is a gap between the Point D we can reach with a money-based system of measurement, and our real-life Point G.

And as such, it will be inevitable that if we optimize our economic systems to optimize some monetary outcome, like GDP or inflation or tax revenues or some such thing, that eventually that optimization will shoot past the real-life target. In a sense, that’s kind of what we’re experiencing in our current economy. America’s GDP is fine, production is up, the inflation rate is low, unemployment is down, but there’s still a general unease about our economy. Some people point to economic inequality as the problem now, but measurements of economic inequality aren’t Point G, either, and if you optimized for that, you’d shoot past the real-life Point G, too, only in a different direction. Look at any historically Communist country (or Venezuela right now) to see how miserable missing in that direction can be.

The correct answer, as it seems to me in all of these examples, is to trust your data up to a certain point, your Point D, and then let wisdom be your guide the rest of the way.

Politics

Which brings us to politics. In 2016. Hoo boy.

Well, how did we get here?

I think there are essentially two data-driven processes that have landed us where we are today. Both of these processes have a gap between what we think of as the real-life goals of these entities, and the direction that the data leads them to. One is the process of news outlets chasing media ratings. And the other is political polling.

In the case of the media, the drive for ratings pushes journalism towards sensationalism and outrage and controversy and anger and conflict and drama. What we think journalism should actually do is inform and guide us towards wisdom. Everybody says they hate the media now, because everybody knows that the gap between Point D and Point G is growing larger and larger the further down the path of ratings the media goes. But it is difficult, particularly in a time where the technology and business models that the media operate under are changing rapidly, to change direction off that track.

And then there’s political polling. The process of winning elections has grown more and more data-driven over recent decades. A candidate has to say A, B, and C, but can’t say X, Y, or Z, in order to win. They have to casts votes for D, E, and F, but can’t vote for U, V or W. They have to make this many phone calls and attend that many fundraisers and kiss the butts of such and such donors in order to raise however many millions of dollars it takes to win. The process has created a generation of robopoliticians, none of whom have an original idea in their heads at all (or if they do, won’t say so for fear of What The Numbers Say.) You pretty much know what every politician will say on every issue if you know whether there’s a “D” or an “R” next to their name. Politicans on neither side of the aisle can formulate a coherent idea of what Point G looks like other beyond a checklist spit out of a statistical regression.

That leads us to the state of the union in 2016, where both politicians and the media have overshot their respective Point Ds.

And nobody feels like anyone gives a crap about the Point G of this whole process: to make the lives of the citizens that the media and the politicians represent as fruitful as possible. Both of these groups are zooming full speed ahead towards Point F instead of Point G.

And here are the American people, standing at Point E, going, whoa whoa whoa, where are you all going? And then the Republicans put up 13 robocandidates who want to lead everybody to the Republican version of Point F, plus Donald Trump. The Democrats put up Hillary Clinton, who can probably check all the data-driven boxes more skillfully than anybody else in the world, asking to lead everybody to the Democratic version of Point F, plus Bernie Sanders.

And Trump and Sanders surprise the experts, because they’re the only ones who are saying, let’s get off this path. Trump says, this is stupid, let’s head towards Point Fascism. Sanders says, we need a revolution, let’s head towards Point Socialism.

And most Americans like me just shake our heads, unhappy with our options, because Fascism and Socialism sound more like Point A than Point G to us. I don’t want to keep going, I don’t want to start over, and I don’t want to head in some old discredited direction that other countries have headed towards and failed. I just want to turn in the direction of wisdom.

“It’s not that hard. Tell him, Wash.


“It’s incredibly hard.”

The Long, Long History of Why I Do Not Like the Josh Donaldson Trade
by Ken Arneson
2014-12-01 22:22

Once upon a time, about a billion years ago, life was simple. Everybody lived in the oceans, and everybody had only one cell each. This was quite a fair and egalitarian way to live. Nobody really had significantly more resources than anyone else. Every individual just floated around, and took whatever it needed and could find, and just let the rest be.

This golden equilibrium was how life did business for a couple billion years. There was no such thing as jealousy or envy, and as a result, everyone lived pretty happy lives.

Then, one day about 800 million years ago, a pair of single-celled organisms merged to become the first multi-cellular organism in the history of the earth.

At first, these multi-celled creatures were just kind of like big blobs of single-celled organisms, and didn’t cause a lot of problems. Everybody was still kind of doing the same job as everyone else, even if they had organized themselves into a limited corporation of sorts. Most other single-celled creatures just figured they were harmless weirdos hanging out together, and ignored them.

They could not have been more wrong. For once the multi-cell genie was out of the bottle, Pandora’s box could not be closed, and the dominos began to fall. This simple change may have seemed innocent at first, but little did the single-cells know that they were the first creatures on earth to fall victim to the innovator’s dilemma. The single-celled creatures were far too invested in the status quo to change, and consequently ignored the multi-cellulars as irrelevant, and did not realize until it was too late that the game had suddenly shifted.

Continue…

How to Remove the Yahoo Sports background image on Chrome
by Ken Arneson
2013-08-29 20:33

Yahoo Sports remodeled their site this morning, and it’s awful. Mostly, I think, because the new background image on is really distracting and annoying. So I decided to zap it. Here’s how I did it, and you can too:

1. Install the Stylebot Chrome extension.

2. Once you install it, you will get a little “CSS” button on your toolbar.

3. Go to a Yahoo Sports web page.

4. Click the Stylebot “CSS” button on your toolbar.

5. Click “Install Style from Social…”

6. It should show “Loading…” for a second, then bring up “Remove background image from Yahoo Sports by kenarneson”.

You should now see a plain gray background instead of the grass field background.

* * *

I wanted to make it a plain white background like it used to be, but the text on the Yahoo Sports site is now mostly white, so that wouldn’t work. So I made it gray.

Quantum Moneyball
by Ken Arneson
2012-10-22 9:54

Just before the beginning of this sentence, this essay could go in an infinite number of directions. But now that the first sentence has been written, the number the infinite directions it could possibly go has been reduced into a much smaller infinity. Who knows what I’ll write next?

It could be anything! Gratitude to their emotions in the water! Or maybe with Zito is blind to park your dastardly actions.

I recently watched a TED Talk by Emily Levine which is like that. It rambles off in a gazillion directions, with little coherency. You could go off now in the direction of watching it. I’m not sure I’d recommend that for you, but I’m glad I did it myself, because it contained one nugget near the end which sent me off in another, more interesting direction.

She rambles this way and that on purpose, not completely polished and slightly unprepared, because she says she likes her talks to remain in a “probability wave” as long as possible. If you’re polished and prepared, you’ve already collapsed your probability wave into single point, and you’ve closed yourself off to new possibilities. She wants to keep open the possibility of “getting on the same wavelength” as her audience.

It’s that idea of “probability waves” that got me intrigued. She’s using ideas from quantum physics to help her understand her art. Using quantum physics as a metaphor sounded interesting, so it sent me scrambling to update myself on quantum physics and probability waves again. And now there’s a very high probability that this essay will devolve into a physics lesson.

* * *

To understand Levine’s metaphor, you need to know about the double slit experiment. This cartoon is the best introduction to it I’ve seen:

That’s kind of freaky. If you’re like me, you still don’t quite get it. I’ll add Professor Brian Greene’s explanation of the double slit experiment on Nova:

* * *

In short:

  • Before observation, a subatomic particle is anywhere in the whole universe.
  • Upon observation, a subatomic particle can no longer be anywhere. It must “collapse” to somewhere specific.
  • Where an “anywhere” ends up collapsing into a “somewhere” is based on probabilities. Some places it can end up turn out to be more likely than others. And these probabilities can interfere with each other, or amplify each other, in the way that one wave can either interfere with another wave, or amplify it.

Ok, if you’re like me, you’re still having trouble understanding the concept of “probability waves.” And when I’m confused, I turn to baseball metaphors.

* * *

Imagine that a baseball player is a subatomic particle. We’re going to pass the player through two slits, and we’ll call these slits “On-base Plus Slugging” and “Plate Appearances”.

Suppose we have a player/subatomic particle named “Kila Ka’aihue”. Let’s say Ka’aihue is projected to hit something like this in 2012:

4% chance his OPS is around .913
8% chance his OPS is around .869
12% chance his OPS is around .837
16% chance his OPS is around .811.
20% chance his OPS is around .786.
16% chance his OPS is around .762
12% chance his OPS is around .738
8% chance his OPS is around .705
4% chance his OPS is around .663

and let’s say he’s projected to get playing time like this:

4% chance he gets around 500 Plate Appearances
8% chance he gets around 450 PA
12% chance he gets around 400 PA
16% chance he gets around 350 PA
20% chance he gets around 300 PA
16% chance he gets around 250 PA
12% chance he gets around 200 PA
8% chance he gets around 150 PA
4% chance he gets around 100 PA

Before the season starts, any combination of these stats are possible. He could hit a .913 OPS and get around 200 PA. Or he could hit .738 and get around 400 PAs. Or any other combination — some are more likely than others, but they can all happen.

Some of these probabilities, however, interfere with each other. If Ka’aihue hits .663, it reduces his odds getting 500 PA, because the A’s will likely give his PAs to somebody else instead. If he hits .913, it reduces his odds of taking a path with only 100 PA, because if he’s playing that well, the A’s will want to give him a lot more than 100 PAs.

Other probabilities amplify each other. If Ka’aihue ends up with a .663 OPS, it increases his odds of ending up with only around 100 PA. If he ends up with a .913 OPS, it increases his odds of ending up with over 500 PA.

* * *

So now, let’s play the 2012 season a million times.

Each time we play, we shoot the Ka’aihue subatomic particle through these two slits, and some particular combination of OPS and PAs ends up on the back wall.

Now, if we chart the one million Ka’aihue outcomes, all the OPSes and PAs, we’ll see something similar to the double slit experiment. We’ll see some areas of high density, and other areas of low density. We’ll get lots of marks where the OPS and PAs are both high, or both low, because that’s where the odds get amplified. We’ll get gaps where one is high and the other is low, because that’s where the odds cancel each other out.

* * *

Now of course, we didn’t play the 2012 season a million times. We only played it once. And in that one, single time, Ka’aihue ended up with .693 OPS in 139 plate appearances — both low. And because of that low outcome, the A’s tried Brandon Moss and Chris Carter at first base, instead.

* * *

You can think of the whole 2012 Oakland A’s season in this way. If Ka’aihue has a low OPS, it amplifies the odds that he’ll also have fewer PAs. If Ka’aihue has fewer PAs, it amplifies the odds of Chris Carter or Brandon Moss or Daric Barton getting more PAs, until one of them starts hitting well. Which is what happened: Moss and Carter ended up in a platoon and hit well.

But if Ka’aihue has a high OPS instead, it amplifies the odds that he’ll get more PAs, and cancels out the odds of Carter and Moss getting a lot of PAs. The whole season takes a completely different path, and probably ends up “collapsing” into a completely different place.

* * *

Baseball is more complicated than just OPS and plate appearances, of course. And in the end, the stat we baseball fans are really interested in measuring on that back wall is team wins.

As the season starts out, there are an infinite number of possible ways the season can play out. Some things are more likely than others, but once we observe the season, all those possibilities collapse into one, single outcome. The 2012 A’s could have ended up with 0 wins or 162, but those are extremely unlikely paths. That would be like a diamond spontaneously jumping out of a locked safety deposit box and into a thief’s pocket. Most likely, the diamond stays in the box. Most likely, the team stays within a “box” between 40 and 120 wins.

Atomic-era general managers will understand all these possible amplifications and cancellations, and construct their teams to maximize the odds that the path their team takes collapses into a championship. The most likely outcome for the A’s was figured by pundits to be around 75 wins. And maybe if you replayed 2012 a million times, it will average to 75 wins. Or maybe, Billy Beane understood how all those waves of statistics amplified and canceled each other out better than anyone else. Maybe, the A’s season collapsing into a single, specific result of 93 wins and an AL West Division title was not quite the miracle we thought it was.

And with that, this essay shall hereby collapse into itself.

* * *

Disclaimer: this metaphor was presented for informational and entertainment purposes only. Baseball players are not actually subatomic particles. Quantum physics are not the most accurate way to describe the behavior of baseball players. Nor are the behavior of baseball players the most accurate way to describe quantum physics. The reader assumes all risk for all unintended uses of this metaphor, including–but not limited to–using Feynman path integral formulations to project future baseball outcomes.

QWOP as an Example of Muscle Metaphor
by Ken Arneson
2012-10-08 12:15

Jon Bois has a fun story over on SB Nation today about QWOP, the stupidest, most aggravating, hilarious video game ever made. I enjoyed the reminder about the game, because it illustrates what I wrote on Friday about muscle metaphors.

The example I used on Friday was how I’d think about shooting a free throw in order to correct my posture in other, non-basketball contexts. Using this kind of “muscle metaphor” allows your brain to build a solution for your task more quickly out of existing pathways, instead of trying building it from scratch or some other, less optimal pathway.

One of the reasons QWOP is so difficult is that it seems to defy our ability to find such muscle metaphors. It’s quite unlike any other task you’ve likely tried, and so when we first play the game, we struggle to find any sort of muscular analogy to help our brains cope with this job.

We fail miserably, usually falling flat on our faces or upside down on our heads after just one or two steps. We’re like infants all over again, trying to figure out how these muscles of ours work, kicking our legs this way and that, hoping that eventually, through trial and error, we figure out how to control these things. The “everyone is a winner” message after every failure acts like an encouraging parent, urging us to keep going.

* * *

Eventually, nearly all babies figure out how to walk. But there are often intermediate stages. Babies don’t have a lot of previous motor skills to build on, so they have to construct these pathways from scratch. So a lot of babies crawl before walking, as it’s an easier task to master. Others figure out a kind of butt-scoot, shuffling along while seated, and are satisfied with that form of mobility until they figure out the harder stuff.

QWOP has an equivalent to the butt-scoot, and that’s a kind of one-knee scoot, which works like this:

1. At the start line, press W and P together quickly six times. This will bring the runner down to one knee like this:

2. Then press Q and O together one time, to get the right thigh perpendicular to the ground:

3. Then you scoot along the ground by pressing the pairs of keys together: W/P three times, and Q/O once, over and over. Pressing W and P together three times kicks the left leg out, then Q and O together one time brings the right knee back to parallel.

4. At 50 meters you will reach the hurdle. You can just kind of knock it over. You may need to give an extra Q/O or two to get over it.

5. Then just continue until you get to the finish line at 100m:

* * *

Interestingly, though, this knee scoot method uses the opposite pairs of keys from what you want to use if you’re trying to move the QWOP guy along on two feet. If you’re walking, you want to press Q together with P, and W together with O.

But having mastered the one-knee scoot, walking becomes a little easier. Although the pair of keys we want to press is now the opposite of what we pressed before, we now have some muscle metaphors to build on. We’ve got practice now in pressing these pairs of keys together instead of individually. We’ve also got practice in developing a rhythm to our motion.

And here’s where I finally can find some muscle metaphor from my own experience instead of just the game’s. I find that when I successfully get the QWOP guy to move, I’m actually doing a kind of skipping. I do a long press first, to kick the leg out, and then I do a short little one with the same foot to adjust the leg to where it needs to be so I can successfully get the next leg moving forward. Thinking about skipping gets me in the right frame of mind to get my muscles to press the keys at the right time.

Of course, it’s still not easy even then. The part I still have trouble with now is failure recovery. If I lean forward or back too much, or stick the leg out too far or not far enough, my instincts for correcting the error seem to always be wrong. Half the time, I choose the wrong pair of buttons to push, so I make my mistake worse, not better. Splat.

Mastering failure recovery is also one of the final stages when toddlers learn how to walk. At first, they’ll fall hard to the side or face down, and as a parent, you need to be there to catch them. Eventually, though, all seem to learn how to fall onto their butts, so that they end up sitting after they fall. And finally, the failure recovery gets to be so good that they don’t fall at all.

Then, the next thing you know, your kid is old enough to drive. (My oldest just turned 15 1/2 last month). Then you have to go through this all over again with them, until they’ve mastered the accelerator, brake and steering wheel enough that you know they won’t crash into a tree anytime someone throws a banana at them.

Muscle Metaphors
by Ken Arneson
2012-10-05 18:09

The A’s making the postseason has hurt the discipline I’m trying to develop as a writer. The little brain tingle I get from reading articles about the A’s or talking about the A’s on Twitter right now is so, so tempting and hard to avoid. I’ve tried to turn my feeds off during my daily writing window, but this week, I’ve been cheating. My writing habits after one month aren’t so strongly ingrained and automatic yet that they could override the temptation of A’s talk. My articles have been much shorter this week as a result.

Translating verbal ideas into unconscious habits is difficult. So much of our routine behavior functions subconsciously and automatically, so it’s a bit of a black box. We can’t see into our brains and see what is wired to what. So we often have to try various tricks to get our subconscious brains to do what our conscious brains want them to do. A recent Fangraphs interview of LaTroy Hawkins by David Laurila had a good example of this:

I chewed on that awhile and slowly tried to transfer the concept of how I felt on the basketball court when I was shooting my jump shot. I worked that into my delivery — my arm angle and my release point. I think that was the key for me, having him translate it into basketball terms. The ball coming off your fingertips when you’re shooting a jump shot isn’t all that different from delivering a pitch.

Hawkins’ brain was already wired to do something close to the right thing when he played basketball. Thinking about his pitching in terms of shooting a basketball created a pathway in his muscle memory to the right behavior.

That struck me, because I’ve used basketball as a kind of muscle metaphor myself in order to change a physical behavior. I have a bad habit of locking my knees when I stand, which transfers weight away from my legs to my back. It causes me back pain if I end up standing for too long. Somewhere in my life, my muscle memory became wired to stand this way, and it’s very hard for me to avoid this habit this without conscious thought. In trying to fix this, I found that there was one situation where I naturally stand with the correct posture — when I prepare to shoot free throws in basketball. So now, when I need to get my posture correct, all I do is say the word “free throw” to myself, and my body corrects itself.

Finding the right cue can take a lot of experimentation, and often differs from person to person. It can be even harder when the behavior you’re trying to change isn’t a simple motion like throwing or standing, but more complex social behaviors. Charles Duhigg wrote a book called “The Power of Habit”, and he has an interesting preview of it here, where he explains how he broke a bad habit he had developed:

I’m really trying to accomplish two things with my experiment of writing daily: first, to improve the quality of the actual words I put on the page, and second, to develop the habits of behavior around that writing that put me in the proper context to write efficiently and effectively. Obviously, this week has shown me that I still haven’t figured either one of those things out yet. Because that part of our brain is a black box, it takes a lot of time and trial and error to figure out a system of cues and rewards that result in a productive set of behavioral habits. I’m still very much a work in progress here.

We We We All The Way Home
by Ken Arneson
2012-09-27 17:46

Yesterday, I mentioned in passing how I enjoy baseball on two levels: one level in rooting for my team, and another in the aesthetic quality of the game. The day before, I defended the idea of cross-pollinating new scientific ideas with older fields of human endeavor, to see what comes out of the mix. So today, let’s make a new hybrid.

How can we explain the psychological attraction in rooting for a team? Why, when we’re watching two teams that we have no previous attachment to, do we often find ourselves rooting for one team or another anyway? And how is this different or separate from the aesthetic joy of watching a game?

* * *

As I write this, I am watching Ian Kinsler bat against my favorite baseball team, the Oakland A’s. On the rooting level, I want him to fail and flail badly. But on an aesthetic level, I admire Kinsler. His at-bats, the way he takes bad pitches and fouls off good pitches until he can get a good pitch to hit, are probably the most consistently good at-bats I’ve seen from any player since Rickey Henderson. If our enjoyment of sports were only about rooting interest, I should be incapable of appreciating Kinsler at all. If our enjoyment of sports were only aesthetic, I wouldn’t have a reason to want to see him fail.

Can baseball fandom be fully expressed in a mere two-dimensional chart, with rooting on the x-axis, and aesthetics on the y-axis? No, of course not. For instance, suppose the A’s pitcher were Bartolo Colon. Colon was suspended in August for performance enhancing drugs, but let’s say he’s served his suspension and now he’s pitching. Do I still root for him to succeed? Yes, he’s on my favorite team. But now there’s a moral dimension on the z-axis added to the mix, too. We can go on. Fandom is complex.

* * *

But still, we want to talk about it, so we need to model it. Do we need modern science to do so? Not really. For example, Aristotle, addressed such issues over two millenia ago. Here’s a paragraph on Aristotle’s aesthetics, from a 1902 version of Encyclopedia Britannica:

Elsewhere he (Aristotle) distinctly teaches that the Good and the Beautiful are different (heteron), although the Good, under certain conditions, can be called beautiful. He thus looked on the two spheres as co-ordinate species, having a certain area in common. It should be noticed that the habit of the Greek mind, in estimating the value of moral nobleness and elevation of character by their power of gratifying and impressing a spectator, gave rise to a certain ambiguity in the meaning of to kalon, which accounts for the prominence the Greek thinkers gave to the connection between the Beautiful and the Good or morally Worthy.

Not sure if Aristotle meant Good and Morally Worthy were separate things or the same, but I’ll assume they’re separate. So applying Aristotle to my example above, the A’s are Good, Ian Kinsler is Beautiful, but Bartolo Colon is Morally Unworthy.

* * *

Aristotle’s three dimensions are a kind of model of this aspect of human nature. And since this model is still being discussed 2,000 years later, we can certainly say that this model has a certain level of usefulness. But does this model accurately map to the actual structure and organization of the human brain? Can we explain this structure in terms of evolution, that there were some sort of selective pressures which led to this behavior?

Aesthetics and morality are huge subjects, so I’ll pass on those in this blog entry, and just focus on the rooting aspect.

Group behavior has always been a bit of a tricky subject for evolutionist to explain. It’s easy to explain selfish individual behavior: it’s behavior that’s directed towards passing your genes on to the next generation over the genes of your rivals. The prevailing explanation for most of the last 40 years or so has been kin selection: unselfish behavior towards your kin helps pass more of your genes along to the next generation. Any sort of unselfish behavior toward people who are not your kin is just sort of a side effect of unselfish behavior towards your kin.

But that’s an unsatisfying explanation, particularly if you apply it to team sports. Why do I go to the Coliseum, dress up in green and gold with thousands of other A’s fans, 99.999% of who are not my kin, and cheer the team together with them? It’s really hard to make a convincing argument that I’m doing it to pass my genes on.

The alternative explanation is group selection. Group selection is a theory that fell out of favor in the 1960s, but in recent years has been making a comeback. In his recent book, The Social Conquest of Earth, E.O. Wilson argues strongly in favor of group selection as an explanation for human social behavior.

Under group selection theory, human evolution happens in two dimensions. There’s a selfish dimension that pushes individuals to promote their genes over others within their group. But there’s also a dimension that pushes us to behave in ways to promote the genes of the group over the genes of rival groups. In times of war or drought or famine, those groups who behave in ways that encourage cooperation instead of selfishness survive to pass their genes on more than the groups whose individuals behave more selfishly.

Under group selection theory, the behavior we see in team sports makes much more sense. We naturally form emotional attachments to our groups, because we were evolved to do just that. As E.O. Wilson points out, every single animal that exhibits social behavior (including the one Wilson is expert in, ants) evolved its social behavior to protect and defend a nest. So we root, root, root for the home team, and find it extremely irritating when invading Yankee fans come into our home nest and chant for their team, instead. The joy we feel when our group wins, the pain we feel when our group loses — those are emotions that evolved in our brains to promote the genetic survival of our groups.

* * *

Note I said “our groups.” Jason Wojciechowski has an article today (Baseball Prospectus, $ required) on the use of the word ‘we’ in reference to team sports. Is it appropriate for fans to use the word “we”, or should that be limited only to the players on the team? Jason tries to define that line somewhere in along the lower level employees of the team. I don’t think that works (which Jason ultimately acknowledges).

Former Baseball Prospectus writer Kevin Goldstein used to rail against fans using ‘we’ on Twitter all the time. At one point (which I can’t find now — Twitter search sucks) — he argued that you don’t say ‘we’ to refer to your favorite band, so why should you do so for your favorite team?

I strongly disagree with Kevin here. A band is different from a team. You like the band primarily because of the aesthetic experience it provides you. But as we’ve seen here, the aesthetic experience is only a small part of the experience of watching baseball. Sports are the most popular activity on earth right now not because they provides an aesthetic experience alone — but because they have gone beyond that and tapped into the a primal root of human evolution: the network of emotions that group selection has hardwired into us.

The reason professional sports is a profession at all is because it creates the feeling of ‘we’. That feeling is the main point of team sports. We-ness is the product.

To have a business that sells a product, we, and then to deny those customers the use of the very word that best describes the product–that’s madness.

Neurohumbug
by Ken Arneson
2012-09-25 19:25

Early in my life, I really didn’t have any sort of vision for a career. I just kind of drifted towards whatever opportunities came to me. I had an aptitude for computers, partly because my dad, who was an electronics technician, understood that they were the Next Big Thing. In 1980, he bought a TI-99/4, hoping that I would fiddle with it and learn from it. I did. And so as I grew up, the opportunities that fell into my lap happened to be with computers, because whenever there was some computer stuff that needed to be done, I seemed to be the guy who could figure it out.

Then in 1994, I was asked to set up a web server. Immediately, I knew. It was like walking up a big hill and just staring at your feet the whole time, and then suddenly you reach the top, see the view, and you suddenly realize the world is a whole lot bigger than the size of your feet. The Internet was going to be huge. It was going to be exciting. I decided I would bet my career on it.

I was far from the only one who understood that the Internet was a Big Deal. Looking back on it now, it’s clear that I was right THAT the Internet would be huge. It’s also clear that neither I nor anyone else had any idea whatsoever HOW it would be huge.

And so the dot-com bubble came and burst, and there were plenty of Pets.com and Webvan.com examples, where my generation made all sorts of big bets on the THAT, and completely missed on the HOW. The Internet would indeed change our lives, but it wasn’t going to be by giving us new ways to sell dog food.

* * *

About 10 years ago, I came to a similar epiphany with neuroscience. I had taken a class at UC Berkeley in the late 80’s that was primarily about aesthetics. The class asked, what made this work of art a classic, but that one forgotten? The question stuck with me for years, but I never could find an answer that made any sense to me. But one day in the early 2000’s it struck me that the answer wasn’t in the artwork, it was in the brain’s interpretation of the artwork. So I googled the word “neuroaesthetics”, wondering if there was such a thing. It turned out there was an International Conference on Neuroesthetics was being held in Berkeley just a few months later. I decided to attend.

I discovered that neuroaesthetics is a baby science, where everyone, including me, was excited THAT we can try to understand art from a scientific point of view, but at the same time, a science where no one really has any clue as to HOW understanding the brain will help us understand art. It seemed to me like looking at a jigsaw puzzle without knowing what it’s really a picture of yet. You start out by looking at this detail and that one, and seeing if any of the pieces fit together at all.

It’s taken about 10 years, but now people are trying to take this information and attach it to their existing models of human activity, to see how this changes the picture we thought we were looking at. Some of these attempts will probably turn out to be the equivalent of attaching the Internet to dog food. But we don’t learn that these things don’t work until we try and fail. Watching this process unfold is as interesting to me as watching the dot-com craze play itself out.

And like any craze, the bubble will eventually pop. Perhaps the first sign of that pop was when the leading journalist covering this neurofever, Jonah Lehrer, was found guilty of various forms of plagiarism. Since then, there has come a natural backlash against trying to apply brain research to all these forms of human activity. The most scathing attack came a couple weeks ago by Steven Poole in the New Statesman:

An intellectual pestilence is upon us. Shop shelves groan with books purporting to explain, through snazzy brain-imaging studies, not only how thoughts and emotions function, but how politics and religion work, and what the correct answers are to age-old philosophical controversies. The dazzling real achievements of brain research are routinely pressed into service for questions they were never designed to answer. This is the plague of neuroscientism – aka neurobabble, neurobollocks, or neurotrash – and it’s everywhere.

Indeed, there are flaws with many of these models that use brain studies for supporting evidence. I’m especially skeptical of those that use brain scans that show the brain “lighting up” in response to this or that stimulus. That’s like trying to understand how a computer works by making note of when the hard drive makes a noise when it spins. It can tell you a little bit about how a computer works, but not nearly enough to build an accurate model from.

I also am suspicious of any model that claims that there are “4 kinds of X” or “7 different Y”, such as Jonathan Haidt’s five six moral foundations. In computer programming, there’s an axiom that you design for cases of 0, 1 or N. You make sure your program can handle it when there’s no data. If there’s one specific thing you’re trying to solve, it’s OK to write something that handles that one specific case. But if you’re going to be handling a number of cases that’s above one, then you abstract your program to a level that can handle ANY number of cases, not just the number of cases you know about. Because otherwise, any time some new situation comes up, you have to write a whole new program. So I find it hard to believe that our brain has wired these specific six moral foundations into our brains, and only these six.

So Poole has a good point. We really don’t know enough about the brain yet to be drawing any grand conclusions from the information with a lot of confidence.

* * *

“Essentially, all models are wrong, but some are useful.”
George Box

But at the same time, if we don’t use what little knowledge of the brain we have, we’d still be asking and trying to answer the same questions about ourselves. Only we’d be doing it without this added scientific information. What we had before this explosion in brain research in fields like aesthetics was not really a science at all. It was mostly just academic jargony humbug.

It’s like condemning the entirety of the Internet because Webvan.com was a disaster. Yes, there were a lot of crap businesses at the beginning of the Internet, and there are a lot of crap theories at the beginnings of neuroscience. But that’s part of the process. Until we can exactly replicate a human brain from scratch, everything is just an imperfect model.

Some of these models will be more useful than others. Today’s models may be deeply flawed, but they’ll be less flawed than yesterday’s. And upon a few of these models, the Googles and Facebooks and Twitters of neuroscience will be born, the models of the human mind that we find truly useful. I see no reason to give up on that vision.

A Baseball Learning Matrix
by Ken Arneson
2012-09-24 17:19

Russell Carleton has an interesting article today on Baseball Prospectus today about the “Search for an 80 Brain“. He explores whether the difference between prospects who make it and those who fail lies in their ability to learn, and wonders if there’s a way to test those learning skills.

For one thing, it’s hard to observe a player’s learning skills, even with a really fancy stopwatch. But if the ability to learn is key to turning raw talent into actual performance, why not spend some time figuring out if the player has a 20 learning tool or an 80? Many players are drafted based on their physical tools, but what about the guy who doesn’t have blow-you-away stuff now, but can develop quickly because he can learn? In general, the closest thing that I hear to this is when scouts talk about “makeup.”

Can this learning ability be measured? My answer is “Yes… I think…”

I think so, too. But off the top of my head, I’d think there wouldn’t be one measure of learning ability, but four.

Here’s why: in order to explore how to measure learning, we need to be clear exactly what kind of learning we are talking about. Learning is about creating memories in the brain, and making those memories accessible when needed. It would be useful here to point out the two main types of memory: declarative and nondeclarative. I’ll quote from a book by Larry Squire and Eric Kandel called “Memory: From Mind to Molecules”:

Declarative memory is memory for facts, ideas, and events — for information that can be brought to conscious recollection as a verbal proposition or visual image. This is the kind of memory one ordinarily means when using the term “memory”: it is conscious memory for the name of a friend, last summer’s vacation, this morning’s conversation. Declarative memory can be studied in humans as well as other animals.

Nondeclarative memory also results from experience, but is expressed as a change in behavior, not as a recollection. Unlike declarative memory, nondeclarative memory is unconscious. Often, some recollective ability can accompany nondeclarative learning. We might learn a motor skill and then be able to remember some things about it. We might be able to picture ourselves performing it, for example. However, the ability to perform the skill itself seems to be independent of any conscious recollection. That ability is nondeclarative.

In other words, declarative memory holds conscious thought, while nondeclarative memory holds motor skills.

So let’s say we have a hitter, like Carleton’s example of Wil Myers, who is a bit too passive, and doesn’t quite swing at enough pitches. We want to make him a somewhat more aggressive hitter. How do we do that?

So it’s not a matter of merely telling Myers to “be more aggressive”. The idea of being more aggressive is a declarative memory, a conscious thought. And that declarative memory, that idea, is independent of the skill itself, of the nondeclarative memory, the motor skill required to output the desired behavior. That conscious thought needs to be translated into a motor skill. A declarative memory needs to be translated into a nondeclarative memory.

As Carleton points out in his article, this much easier said than done. The reason is that while declarative memories are under our conscious control, nondeclarative memories are not. They are created subconsciously, involuntarily and automatically. These memories are often context and emotion dependent. If you want to manipulate the nondeclarative memory system into creating the muscle memory you want, you basically have to trick it. You can trick it by repetition and practice, and/or by manipulating whatever emotions are needed, whether anger or calmness or excitement or determination.

* * *

So a scouting report for learning might look something like this:

Joe Prospect, Learning Scout Report

declarative input nondeclarative input
declarative output 30 40
nondeclarative output 50 80

Upper Left: declarative input, declarative output.
This would represent the player’s ability to repeat an instruction in his own words.

Coach: “When I say, ‘cut down on your swing’, what does that mean?”
Player at level 20: “I dunno.”
Player at level 80: “It means I shorten my stride, and bring my bat to this position here…”

This square really measures a player’s ability to coach more than it measures his ability to play. Perhaps it might also measure a player’s ability to be a catcher who can take a game plan and execute it, and to handle and communicate with a pitching staff. It can also help pitchers, not so much in the physical act of throwing a ball, but with setting up hitters and sequencing.

In general, though, this is the least important square in the matrix. Because what we’re aiming at in regards to players is the nondeclarative output, the muscle memory needed to perform at a high level. And nondeclarative input — the sensory and pattern-recognition feedback the brain gets from actually playing — is more important than the theoretical, declarative input in this square.

 

Upper Right: nondeclarative input, declarative output.
This would represent the player’s ability to articulate his own experiences.

Coach: “Why didn’t you swing at that pitch?”
Player at level 20: “I just froze.”
Player at level 80; “I was expecting a breaking ball away, and instead he threw me a fastball on the inside corner, and because my body was leaning out, I couldn’t adjust my balance quick enough to pull my hands in and start the swing.”

An 80-level player in this square of the matrix would be a reporter’s best friend. High skill in this area can also help a player to understand what he needs to work on, and create systematic workout procedures for improving those self-understood weaknesses. But being able to articulate what you physically experienced won’t really help you unless you also possess a high score in the lower left square.

 

Lower Left: declarative input, nondeclarative output.
This represents coachability: a player’s ability to take verbal or conscious ideas, and translate them into muscle memory.

A player at level 20 probably can’t even do this at all. If he learns anything, it’s only “the hard way”– by failing or succeeding himself in real situations.
A player at level 50 is someone who may need to be told something over and over until it finally sinks in. Or needs to be told something in 1,000 different ways until he finds that one mental cue which triggers the correct behavior.
A player at level 80 probably only needs to be told something once, and can immediately make the physical adjustment.

 

Lower Right: nondeclarative input, nondeclarative output.
This represents a player’s ability to learn from his own senses and body, from the immediate success or failure of his efforts.

A player at level 20 probably isn’t affected much by his own failures and successes. He probably repeats the same mistakes over and over again, and can’t adjust.
A player at level 50 can learn from his own failures and successes, but it takes a long time and many repetitions for those adjustments manifest themselves.
A player at level 80 probably never seems to make the same mistake or get fooled by the same pitch twice.

* * *

A single, Wonderlic-like test wouldn’t work to fill out such a matrix. You’d probably need to develop separate tests for each of the squares in the matrix. And then you’d need to collect that data for a number of years to figure out whether there is actually any sort of correlation between any of that data and the eventual success and/or failure of prospects. Sounds like a lot of work for an uncertain payoff, but it would certainly be interesting to see if there’s something there of value. The sad part is, since baseball teams keep information like this proprietary, we baseball fans will probably never know.

MLB’s Customer Alignment Problem
by Ken Arneson
2012-09-12 15:13

As Apple announces the iPhone 5 today, I want to make a confession. It’s a bit embarrassing for someone like me who has spent his career in high tech, but here goes: I don’t have a cell phone.

At first, my reason was this: I lived and worked and played on a very flat island with no tall buildings, so coverage was awful. I had reception in only one room of my house, not at all in my office three blocks away, and spotty reception where I play indoor soccer. I spent 95% of my time at those three places, and I wouldn’t be getting what I was paying for. When I went out somewhere that coverage was better, I’d borrow my wife’s cheap pay-as-you-go phone, and my needs were met.

The cell phone service providers made a mistake. They gave me an opportunity to learn that I could live without them.

* * *

This is the image of an industry suddenly collapsing.

Why did the newspaper industry suddenly collapse like that? Because it got hit from two sides at once by the Internet. Craigslist and eBay took away their classified ad business, while blogs and online news sources directed their readership elsewhere for the same information. Newspapers might have been able to handle a one-front battle, but a two-front battle was catastrophic.

But there’s something else that hurt the newspaper industry: the indirect nature of their feedback loop. It’s a business model that provides a service to one group of people, while taking money from a different group of people.

The best kind of feedback for a business is revenue. If your revenue increases or decreases, you’re going to notice. But when the users of your product aren’t providing the revenue for your product, your feedback loop has a natural delay to it. The people who give you revenue might lead you to innovate (or not) in a direction your users won’t like, and you won’t notice that you’re making a mistake because it takes a while for that problem to reach your bottom line. So you react too slowly, and that slowness can be fatal.

* * *

In fact, all businesses that rely on advertising have this problem. Their users want one thing, and the revenue generators want something different. So if a company like Facebook starts alienating their customers in an effort to maximize their revenues, they may find themselves not just the subject of an Onion parody, but ruining their business before the bottom line has time to let them know they’ve made a mistake.

* * *

Amazon CEO Jeff Bezos is a very smart man. He understands this problem. He knows that the key to keeping ahead of the rapid pace of high-tech change is to master the feedback loop between what his customers want next, and what his company makes next. That’s why in his Kindle press conference last week, he laid out this doctrine:

Think about the major players in high tech right now: Microsoft, Google, Facebook, Twitter, Amazon, Apple. Which of these has their revenues most directly aligned with what their customers want? It’s probably in roughly this order:
Amazon/Apple
(gap)
Microsoft
(gap)
Google
(gap)
Twitter/Facebook.

Amazon and Apple sell most of their products directly to their users. When their customers buy something they make, they know the product is good; when they don’t buy, they know immediately they made a mistake. Microsoft doesn’t sell directly to users– they sell to distributors and OEM manufacturers, so there’s noise injected into their feedback loop, and they land just a little lower on this spectrum. Google sells ads, but their ads are often directly related to what the customer wants; if someone is searching for jeans, they get an ad for jeans. Sometimes the ad happens to be exactly what the user wants.

Twitter and Facebook, on the other hand, need to inject their ads into an environment where the users wouldn’t really want to see ads at all if they didn’t have to. This leads to customer dissatisfaction, expressed not just in the Onion parody above, but also in a real-life alternative social networks like App.net who are trying to sell directly to users.

* * *

My favorite product of all is probably Major League Baseball. I consume a lot of Major League Baseball. But MLB is going in a very tempting, but dangerous direction. When MLB began, a vast majority of their revenues came from the product their users consume. They sold tickets to the games, and that’s how they made their money. But more and more of MLB’s revenues are coming from indirect sources.

At a local level, if you’re a team like the Oakland A’s, who play in an antiquated stadium that doesn’t generate a lot of revenues, a big proportion of your money comes from TV and league-wide revenue sharing. So you can do things over a number of years, like threaten to move away and trade away favorite players, that damage your brand but aren’t directly noticeable in your bottom line. Ownership may not even know how much their fans hate them, because their loyalty to their local team keeps them around despite their dissatisfaction. But eventually, there may be a straw that breaks the camel’s back. Los Angeles Dodger fans may have hated owner Frank McCourt for years, but it was only in the last year of his tenure that the dissatisfaction actually became truly noticeable in attendance figures. The feedback loop in baseball has quite a long delay.

In recent years, baseball’s misalignment problem has accelerated almost exponentially. Like the newspaper industry, the source of this change is new technology. Unlike the newspaper industry, however, the change has caused MLB’s revenues to increase, not decrease, dramatically.

The reason for this change is twofold:

  1. The DVR allows people to skip through commercials on TV. This makes live events, which people can’t skip through, a much more valuable delivery mechanism for advertising.
  2. Internet video allows people to watch television shows and movies without subscribing to any sort of cable or satellite TV service. Cable/satellite may or may not be shedding customers at the moment, but it’s certainly not growing much, and without sports would almost certainly be shrinking. This makes sports networks extremely important to cable and satellite providers: without them, most people will eventually learn to do without cable TV, and just get all their content from the Internet.

So this sets up a strange mismatch between what MLB customers want, and what their revenues tell them to do. MLB fans want to watch their favorite team on whatever device they prefer. But MLB’s revenue stream is depending more and more on their customers NOT being able to watch their team over the internet, forcing them to watch on TV.

MLB’s revenues come less and less directly from baseball fans, and more and more indirectly, from TV networks and cable/satellite providers.

This causes MLB to lose control over the customer experience of their fans. If a TV network and a cable provider can’t come to an agreement on price, for example, Padres fans can go a whole season without being able to watch their team on TV. And if not all TV networks are available on all cable/satellite services, fans have to scramble around to watch the games they want.

Or, fans just go without. And what does that do? It ends up teaching them that they can learn to live without your product. That too, may not be noticeable right away, until it snowballs too late for you to do anything about it.

* * *

But you can’t really blame MLB, can you? If you’re the Los Angeles Angels and someone wants to pay you $3 billion dollars over 20 years for your TV rights, do you turn that down? No, probably not.

In expectation of an even larger payday than the Angels got, the Los Angeles Dodgers recently sold for $2.1 billion. Given that the estimated value of the Dodgers just a year before that was about $800 million, over 60% of the value of the franchise lies in that regional TV deal.

But think about that for a second. The new owners of the Dodgers spent $1.3 billion dollars on a business model that:
(a) depends almost entirely on another industry that isn’t growing, and would be in steep, steep decline without your industry. If either one of your industries catches a cold, the other one will necessarily start sneezing.
(b) puts your industry in a complete misalignment with your customers, requiring you to prevent your customers from consuming your product in the way they’d prefer, distancing yourself from the revenue feedback loop, making it more difficult to know if you’re doing any long-term damage to your product, and where you need to improve.

The new Dodger owners obviously didn’t care. Maybe they didn’t think about these risks. Or maybe they did, and thought the math worked anyway. Or maybe they considered it, but they thought they could sell the team to someone else before the whole house of cards fell in, like an investor in a Ponzi scheme who doesn’t think he’ll be the one who ends up being the victim.

Who knows. But me, I wouldn’t touch a business model like that with a 10-mile pole.

Both Neuroaesthetics and its Critics are Off Track
by Ken Arneson
2011-12-07 13:20

Alva Noë has an op-ed piece in the New York Times in which he rips into the young science of neuroaesthetics:

What is striking about neuroaesthetics is not so much the fact that it has failed to produce interesting or surprising results about art, but rather the fact that no one — not the scientists, and not the artists and art historians — seem to have minded, or even noticed.

Well, I minded and I noticed, but I’m also no one. I’m not a scientist or an artist or an art historian. I did, however, attend the a few of the initial international conferences on neuroaesthetics. But even though I am deeply fascinated by the idea of understanding art through understanding the brain, I stopped going to these conferences. I felt like the neuroaesthetics community was going down the path that wasn’t going to lead anywhere that would lead to any answers I had about art (What is art? How does it work?) anytime soon.

Mr. Noë seems to have the same frustration I did with the path this science is taking. But he reaches a different conclusion from me: he basically throws up his hands and suggests it’s probably hopeless:

For these reasons, neuroscience, which looks at events in the brains of individual people and can do no more than describe and analyze them, may just be the wrong kind of empirical science for understanding art.

I think that’s mistaken. I’ll try to explain why.

An example: one conference I decided to skip looked to be about examining brain scans of people in love. I’m not sure how and if love and art are related, and I’m skeptical of the usefulness of brain scans. I failed to see how that is going to tell us anything about the mechanisms of art, so I decided not to waste my time.

I’m a computer engineer. The computer analogy to using brain scans for understanding art would be trying to reverse engineer a piece of software by looking at which disk sectors are being accessed on a hard drive when that software is running. That information is almost useless. If you want to reverse engineer anything–a brain, a computer, a piece of software, a transistor, whatever–you need to know exactly two things: the inputs, and the outputs.

If you know what the inputs are (in this case, works of art) and the outputs are (human reactions to works of art), then you can try to reverse engineer the rest. If your inputs and outputs match the original, even if your new machine works in a completely different way from the original, congratulations, you’ve reverse engineered the product.

A reverse engineering of art must begin not with a cataloging of the mechanisms of art, but of the inputs and outputs. That’s where I think the neuroaesthetics community has gone astray.

That’s not to say that an cataloging of the mechanisms of the brain isn’t useful–it is. Usually–but not always–knowing some of the mechanisms of the original product can help you figure out how to complete your reverse engineering. It can help you better categorize your inputs and outputs. But if you’re focused exclusively on understanding the mechanisms and and not on understanding the inputs and outputs, you’re not going to get anywhere.

As I said, in this field, I’m a nobody. I’m not an academic or an artist or a neuroscientist. I’m just a guy. I’m no one. But I’ve done a lot of thinking about it over these past seven years, and I’m convinced that the key to reverse engineering how art works in the brain lies in the difference between the two types of memory in the brain: declarative memories and non-declarative (or associative or procedural) memories. Understand that mechanism, and reverse engineering the rest will fall right into place.

It all seems clear and obvious to me. A neuroaesthetics community that uses an effective approach to the problem of how art works can probably give us lots of useful and interesting information. I’ve blogged about this for seven years now, I still feel as if I’m the only one who gets it. I’m not getting the idea across. I’m a lonely community of one. I was hoping that reading Daniel Kahneman’s recent book “Thinking, Fast and Slow” would shed a little light on the issue for others, but I don’t think it does. Art doesn’t really come up in his book. But the declarative/non-declarative dichotomy I’ve been talking about is pretty much the same System 1/System 2 dichotomy that Kahneman talks about–I think.

Perhaps if the professionals aren’t going to figure it out, that maybe I’ll just have to write a book myself, where I lay out the whole thing, my understanding of how it all works.

That idea scares me, though. What if I spend all that time to write that book and still nobody gets it? Or worse, I’m dead wrong about it? Hmm…

Notes on Thinking, Fast and Slow: Chili Davis
by Ken Arneson
2011-11-26 19:11

I got a Kindle today, and one of the first books I bought was Thinking, Fast and Slow by Daniel Kahneman. I don’t think I’ve ever looked forward to a book more than this one, a summary of how the human mind works by the leading scientist in the field. I plan on making some notes on this book as I read it.

Here’s the first one. Yesterday, I wrote this on Twitter, about the Oakland A’s hiring of Chili Davis as their new hitting coach.

I approve of Chili Davis as A’s hitting coach, since I liked him as a player & I have no other way to know what makes a good hitting coach.

Right in the very first chapter, Kahneman discusses this kind of mental error. He describes an executive who decides to buy Ford stock because he likes Ford cars. He doesn’t take into consideration at all whether the stock is currently priced correctly.

The executive’s decision would today be described as an example of the affect heuristic, where judgments and decisions are guided directly by feelings of liking and disliking, with little deliberation or reasoning.

Kahneman goes on to explain that when we lack the skills to answer a question, what we often do is answer a different question instead. I don’t have the skills or knowledge to know whether Chili Davis would be a good hitting coach. But I know the answers to some other similar questions. Was Chili Davis a good hitter? Yes! Was Chili Davis a likable player? Yes!

So my mind naturally decides to substitute the answers for the questions I can answer for the question I can’t. And the odd thing is, we often don’t notice ourselves that we’ve performed this question substitution, so we often feel very confident in our answers, without just cause.

I feel very happy about the choice of Chili Davis as the A’s hitting coach. I feel quite confident that he will do a great job. Logically, I know that this is just a kind of cognitive illusion. But knowing how the trick works doesn’t seem to make the trick stop working. I still feel happy and confident about Chili Davis as the A’s hitting coach.

Why a no Chicken?
by Ken Arneson
2011-09-07 11:36

In a recent episode of Louie, Louis CK tells a joke that he admits he doesn’t know how to finish. It involves a duck who thinks he’s special because he has a green head.

This blog entry — heck, this blog — is like that. I’m not sure where I’m going with it, I don’t know how it will end, I just have a feeling that I’ve got something here that can come together in the end.

* * *

I recently took one of those online narcissistic personality tests. I scored “normal”. But the only reason I even got as high as normal was because I had an over-the-top score in the “superiority” subsection. I’m not vain or power-mad at all, but dammit, facts are facts. I’m special. I have a green head.

* * *

The Louie show fascinates me. If you put me in a focus group where I was holding one of those dials while watching it, I’d probably flatline at the bottom the whole episode. I squirm, I cringe, I feel uncomfortable the whole time I’m watching it, thinking “I hate this I hate this I hate this.” Based on my real-time reactions, the network execs would probably cancel the show. But when you ask me afterwards how I feel about the episode, I usually love it.

Nobel Prize winning behaviorial economist Daniel Kahneman had demonstrated how humans have two distinct kinds of happiness. There’s a happiness that one experiences in the moment, and there’s a second kind of happiness that one feels in remembering things afterwards. The two kinds of happiness don’t necessarily correlate with each other at all.

The standard sitcom focuses like a laser on the experiential kind of happiness. We’ve all watched these shows–30 minutes of set up, punchline, laugh–but the remembrance of it usually leaves us feeling empty. I think Louie’s uniqueness stems from an indifference to the happiness of experience, if not an outright avoidance of it. The show cares more about afterwards, the happiness of memory.

* * *

Steve Jobs recently retired as CEO of Apple Computers. It’s been a helluva career. In the one and only commencement speech he ever gave, Jobs said:

Again, you can’t connect the dots looking forward; you can only connect them looking backwards. So you have to trust that the dots will somehow connect in your future. You have to trust in something — your gut, destiny, life, karma, whatever. This approach has never let me down, and it has made all the difference in my life.

From most accounts, Jobs could be a mean sonofabitch to work for. The experience at the time of creating all those great Apple products was probably miserable thanks to Jobs’ harsh taskmastery, but after seeing the results, the memory of it afterwards was probably amazing.

* * *

But there’s one nagging question I have about this philosophy: what if you only think you have a green head? What if your self-image is deceptive? What if you’re really something other than what you think you are? Why a duck? Why a no chicken?

* * *

There’s a scene in another episode of Louie where Louis CK has lunch with a Hollywood executive. She asks him for his sitcom ideas, and he starts explaining his idea for a show that avoids experiential pleasure. But he can’t explain how it’s special, how it pays off in the end. He’s envisioning a green-headed duck, trusting that the dots will connect and there will be a green-headed duck in the end, but what he’s describing sounds to the executive like a chicken with some sort of deadly disease.

It’s safer and easier, not just for network executives but for human beings in general, to follow the immediate feedback, to trust the constant data streaming in from our current state of happiness, rather than ignore that short-term data and believe that something larger and more rewarding will emerge.

Postponing pleasure now for a bigger payoff later is very risky. If you’re not special, if you can’t make the dots connect, if there’s no big payoff in the end, no pot of gold at the end of the rainbow, no heaven waiting for you after a virtuous life, if you don’t really have a green head, then you’ve got nothing to show for it but misery. No happiness from experience, and no happiness from memory, either.

That’s why shows like Louie don’t get made very often. That’s why companies like Apple are unique rather than ubiquitous. Because most of the time, none of it was worth it. You were just being an asshole for no good reason.

* * *

I’ve worked in the high tech industry from the infancy of the world wide web, and I’ve seen a lot of companies (including some of mine) start out with the Applest of intentions. But then the feedback starts coming in, from customer service and sales, and it’s nearly impossible for anyone who isn’t a complete narcissist to say “nope, our customers are wrong and our vision is right.” Because usually the customers are right and your vision is wrong. So you follow the feedback. Be the bird that you are, and you usually have a pretty decent gig.

* * *

Modern electronic writing is primarily a pleasure-of-the-moment activity. Today’s blog entry is forgotten tomorrow. Our tweets are out of mind as soon as they scroll off our feed. We’re reacting in the moment to last night’s game, this morning’s article, tonight’s political speech. Which is fine, that’s what these media are meant to do. They’re chickens. Chickens are great, as long as you’re not expecting a duck.

* * *

Lately, I’ve had offers to write for a number baseball outlets out there. I’ve thought about trying a Craig Calcaterra, to see what I could accomplish I left my old, higher-paying career to commit to writing full time.

But so far, I’ve (mostly) resisted that temptation. My gut tells me, “don’t make that commitment.”

It’s partly because I don’t have all my ducks in a row in my personal life to make that practical right now. I quit writing regularly two years ago because I was juggling too many balls in my life, and I ended up doing a half-assed job on all of them. I hate feeling like I’m not living up to expectations, I hate feeling like I need to work 24/7 in order to avoid feeling like I’m not living up to expectations, so I resist making commitments that would create any expectations. Hence, for now, this blog, where I can do what I like, when I like, how I like with maximum flexibility and minimum commitment.

It’s probably also because I’m narcissistic enough to believe I’m unique. I’m not ready to cooped up and commit to a life as a chicken. I’m not ready to accept that this is how I finish this story. I feel, rightly or wrongly, that I’m my own species, who simply has not yet encountered the right variety of poultry to fall in love with.

Brainyball: The Sequel to Statistics
by Ken Arneson
2011-02-09 11:55

Our old friend Moneyball will be making a comeback this year, when the film starring Brad Pitt gets released this September. Let me declare seven months ahead of time that I am sick of hearing about how the movie hype is distracting the 2011 A’s during their pennant run. I am also preemptively tired of the rehashing of old arguments, such as how the A’s philosophy failed because the Moneyball generation never won a ring. Finally, I am, in advance, savoring the irony of the A’s winning the 2011 World Series, in the very year that this antique anti-Moneyball argument reaches its crescendo.

I love me a good irony. I took my daughters Monday to see Sally Ride give a speech for the UC Berkeley Physics Department. I looked around the auditorium and noticed that darn near everyone in the room was skinny. Maybe these people burn all their fat off just by thinking so hard about the universe. Whatever the cause, I found myself tickled by this ironic idea: Physicists have very little gravitational pull.

The irony that lies at the core of the Moneyball book is that A’s GM Billy Beane was trying to find a way to weed out players who were essentially just like himself. Beane is a very intelligent guy with an chiseled athletic body whose intelligence got in the way of his performance. You look at him, and you think he was born to be a star athlete. But he never became one. He’d get so worked up about every little failure that his swing and approach got all screwed up. He couldn’t handle the mental part of the game.

So Beane became a scout, then a GM, and tried to come up with a reliable way to weed out players like himself who can’t handle the mental part of the game, and discover the players who can. They tried to accomplish this by using a deeper understanding of statistics.

Which is odd, if you think about it. It isn’t the players’ statistics that are causing players like Beane to fail. It’s their brains. If you really want to be able to recognize players like Beane in advance, shouldn’t you try to do this with a deeper understanding of brains?

* * *

We are living at the very dawn of neuroscience. In the last ten years or so, our understanding of our own brains has exploded, and we’ve still only scratched the surface. Consider this TED talk by Charles Limb:

Limb explains what happens in the brain when jazz musicians improvise. When improvising, jazz musicians shut off a part of the brain called the lateral prefrontal cortex, which is involved in self-monitoring. They literally turn off the inhibitions in their brains, so they aren’t afraid to make mistakes, and are free to be creative.

Now it would be a big leap to say that Billy Beane’s mental failures were caused by an inability to turn off his lateral prefrontal cortex while batting. But it’s not a big leap to think that this sort of understanding of the brain isn’t just possible for musicians, but for athletes, as well.

Someday, perhaps, draft preparations will include brain scans, so teams can see that a Billy Beane’s brain isn’t focusing properly when batting. They’ll know how often you can take a player with Beane’s brain profile, and train him to overcome those brain issues. They’ll discount or increase his value because of this information.

* * *

In Sports Illustrated this past weekend, Joe Posnanski looked into the question of how drafting teams can predict which quarterbacks will succeed in the NFL, and which will fail. In particular, he wonders what set Aaron Rodgers apart from other first round QBs who flopped. He makes a guess:

What you get from these quotes and just about everything Rodgers says — in addition to steady and pleasant boredom — is a sense of someone who thinks about things constantly, even little things that few others think about. He seems to be someone who simply cannot imagine staying the same, simply cannot imagine that he’s already good enough. There are so many potential distractions at the NFL level, some of them off the field (money, fame, fan fickleness …), some on the field (dealing with pain — Rodgers has a history of concussions — standing up to a heavy rush, the inner workings of a team …). And the most successful quarterbacks, bar none, are the ones who deal with those distractions and never believe the hype and continue to hunger for even the slightest improvement.

To which I ask: how does this separate him from Billy Beane the baseball player? Beane thought about things constantly. He obsessed over every failure, trying to fix every mistake. And this sent him into a downward spiral that made him worse and worse, not better.

* * *

Another player who Posnanski’s paragraph applies to is Barry Zito. I’ve written a lot of words defending Barry Zito from his detractors over the years.

I like Zito. If not for the early Cy Young Award and that ridiculous contract, he’d be the kind of underdog people like to root for. Posnanski’s phrase “continue to hunger for even the slightest improvement”: that’s Zito. He’s a smart guy. Curious. He likes to tinker. To experiment. To find a new way to get better. He tries new pitches. He tries new pitch sequences. He tries new release points. And maybe that constant search for improvement has kept him healthy and pitching in the major leagues for a decade with the mediocre-est of fastballs.

But I’d argue that perhaps as often as it’s helped him, that personality trait has gotten him into trouble. Zito has had three pitching coaches in the majors: Rick Peterson, Curt Young, and Dave Righetti. Pitching coaches tend to live by a sort of Hippocratic Oath: if it ain’t broke, dont’ fix it. Zito doesn’t seem to believe in that. Each time there was a transition between coaches, Zito decided to take advantage of his temporary lack of parental supervision to completely change his pitching motion.

In 2004, Zito decided to try a new motion out of the stretch. He’d always wanted to do this, but Rick Peterson wouldn’t let him. When Curt Young came in as the new pitching coach, he didn’t have the relationship with Zito to say no. Zito had a 4.48 ERA for the year, his worst in an Oakland uniform. The next year, he was back to his old delivery, and his usual sub-4.00 ERAs.

In 2007, he signed a huge contract with the Giants, and showed up at spring training with a radically new delivery. Pitching coach Dave Righetti was horrified, and they settled on a compromise semi-radical new delivery. The results were just as bad as the other time he tried to overhaul his delivery: Zito’s worst year in the majors, a 4.53 ERA. (Followed the next year by an even worse 5.15 ERA.) Two years into his Giants tenure, Zito finally tinkered himself back into some decent success, with two consecutive years now of ERAs around 4.10.

* * *

Recently on Fangraphs, Jonah Keri suggested that the Yankees try to trade for Zito.

I don’t think there’s anything particularly wrong the arguments he gives, but it is, like the Moneyball story, missing the psychological element.

Psychology clearly matters in the outcome of sports careers. The question is, understand enough about sports psychology that such data points are useful in evaluating players, or is the information we have so primitive that we should discount such information altogether?

The Yankees are unique in that they also deal with the theory that there are some types of personalities who “can’t handle New York“. This theory may or may not be valid, but I’m willing to consider that it is possible.

I’m not going to come out and say that Barry Zito is another Ed Whitson. But New York media pressure or not, we do have these data points: each time Barry Zito has had a change of scenery, he used the opportunity to make a royal mess of his delivery.

I think if you’re Brian Cashman, and you’re thinking of trading for Barry Zito, you should know these data points. There is a non-zero risk that Barry Zito’s brain is going to get in the way of his performance, because it seems to have happened to him before. And there’s a non-zero risk that the New York media pressure will trigger this effect, because it seems to have happened to other players before. And to the extent you’re willing to believe those risks exist, you have to discount Barry Zito’s value.

* * *

In Billy Beane’s case, the constant striving for improvement was nothing but counterproductive. In Zito’s case, we see some mixed results. So even though it’s a different sport and a different position, I have a hard time believing that the key to Aaron Rodgers’ success is simply a matter of willpower, that same constant striving for improvement.

If I had to guess, a quarterback’s success involves spacial pattern recognition, the ability to quickly recognize types of player movement, to filter out inessential patterns and recognize significant ones, and act on them. Maybe some players filter out too much information, and others not enough. Maybe there are places in the brain that Aaron Rodgers turns on or off in better ways than the quarterbacks who failed. Those places are mostly a mystery to us now.

But they won’t be a mystery forever. A new era is dawning.

Birth of a new age.

Willpower Bias
by Ken Arneson
2010-01-11 17:45

This past weekend, I pulled out some crates so we could put away our Christmas ornaments.  My two-year-old daughter decided she wanted to pretend she was a Christmas present, and climbed into one of the crates.

“Close the lid,” she said.

I tried, but she didn’t fit.  “I can’t close it,” I said, “you’re too big.”

“Please?” she asked.

“You don’t fit,” I explained.  “Your head sticks out.  I can’t make you fit if you’re too big.”

“Please please PLEEEEEEASE?”

Two-year-olds see the world as entirely a function of their parents’ willpower. Anything that happens, or doesn’t happen, is because mommy and daddy want it that way—even whether or not a particular girl can fit into a particular box.

Of course, we get older and learn that the world is more complex than that, but that bias towards assuming the universe runs on willpower doesn’t completely go away. It’s built into our psychology, because of the very nature of human childhood.

And because it’s part of our psychology, this willpower bias also gets built into the very structures of our societies. Many of our religions believe a larger-scale version of the two-year-old’s assumption: that anything that happens is because God wants it that way. We see it in sports. We thank God if we win a sporting event, then say, “we didn’t want it enough” if we lose. We elect Presidents and Governors hoping for them to be parent-like and fix things through the force of their will. Every election cycle, we make them tell us over and over how they’re going to fix the economy, when in reality, they have very minimal influence on the economy. “Create jobs, please please PLEEEEEEASE?”

And even more insidiously, willpower bias is built into our languages. Consider these two sentences, one of the few examples where you can avoid willpower bias in the English language:

My arm was raised.

versus

My arm rose.

Raise, like many other verbs in the English language, assumes some sort of willpower behind it, causing the action. The implicit full sentence is “My arm was raised by somebody.”

Rise, on the other hand, differs from raise in one key way: it does not assume an agent behind the action. There may have been willpower causing the arm rise, or there may not have been. But by choosing the world rise over the word raise, we are deliberately excluding any information on whether an agent caused the action. In some other languages, you can take any transitive verb and render it agentless with a grammatical marker, but this isn’t possible in English.

If you think, “so what?” then imagine how we’d think of the world if the word “rise” did not exist in English. You could not say, “The sun rises every day”. You’d have to say, “The sun is raised every day.” Which naturally leads you to wonder, by whom? Copernicus? Carl Sagan? Apollo? God?

If we are choosing a philosophy, it would be good if that philosophy possessed the equivalent of that grammatical marker which the English language is missing. We want our philosophy to be able to distinguish between the forces that can and should be influenced by willpower, those which operate independently, and the various shades in between. We want to choose a philosophy that is as effective as possible, and doesn’t leave us crying “Please please PLEEEEEEASE” in vain.

Put on my Linking Cap
by Ken Arneson
2010-01-07 11:26

I’ve got a blog post that’s about 33% written, and every time I write more, it remains 33% written, because it just keeps growing, and I can’t figure out how to break it up into smaller parts.  So in the meantime, here’s some interesting links that don’t fit into the upcoming monster essay:

Americans are flexible about religious beliefs

Bible literalists are the squeaky wheels of American religion, and so they get a lot of attention.  But a large percentage of Americans personalize their religious beliefs, mixing elements of various philosophies and religions into their own.  Knowing this makes the quest I’m undertaking on this blog seem a little less lonely, if nothing else.

Orchids and Dandelions

Context matters.  Some genes expressed in the brain that may lead to criminal behavior in an abusive environment may also lead to beneficial creative behavior in a rich, loving environment.

Aging Brains

Older brains zoom in on the higher-level main idea, and ignore low-level details.  This has its benefits, but also drawbacks.  Can we train the older brain to pay attention to detail?

Should government take “pursuit of happiness” seriously as one of its primary goals?  Money quote:

Wherever I look, some simple patterns hold: A stable marriage, good health and enough (but not too much) income are good for happiness. Unemployment, divorce and economic instability are terrible for it. On average, happier people are also healthier, with the causal arrows probably pointing in both directions. Finally, age and happiness have a consistent U-shaped relationship, with the turning point in the mid- to late-40s, when happiness begins to increase, as long as health and domestic partnerships stay sound.

Rooting Interests
by Ken Arneson
2009-12-28 0:06

Can you watch a sporting event dispassionately, without rooting for one side or another at all? I’ve tried, but I can’t do it. To some extent, I always end up picking sides. For me, it’s impossible to remain objective.

The curious thing is that I can’t help it. I don’t decide that I need to pick a team. I don’t go through some conscious, analytic process to choose a side. It just happens. Even if I try not to pick a side, I still pick a side. It’s subconscious, outside my willpower, and fully automatic.

Few of us choose our sports allegiances through some rational process. Does anyone believe that there exists some objectively “correct” team to root for? While one could probably invent some formula to calculate the “optimal” team to support, most of us would consider such a process silly and beside the point. The emotions, the pure irrationality of our fandom, is the whole point of the exercise.

On the other hand, philosophy feels different to us. We suspect that there exists, if not a single “correct” philosophy, a scale in which some philosophies are better than others. While we have no objections to letting our subconscious passions decide our rooting interests in sports, there’s a sense that when it comes to religion, politics or other types of philosophy, this same decision-making process is flawed.

And yet, can there be any doubt that for the vast, vast majority of people, the decision-making process for picking sides in both sports and philosophy is exactly the same? A large majority of us end up choosing the same religion as our parents, and the same political party. If we chose them by a purely objective process, you’d probably see a far weaker correlation between the people around us and the philosophies we choose.

Suppose we did want to choose a philosophy using some objective method. We’d need to avoid taking sides in advance, in order to avoid letting our prejudgments cloud our analysis. But when it came to sports, we found we usually can’t really help who we choose to root for. It just happens, subconsciously, automatically.

So here’s the big question: even if we want to avoid prematurely picking a philosophy to root for, can we? Is it humanly possible at all? We’ll explore that question next time.

And So To Fade Away
by Ken Arneson
2009-02-04 2:21

To produce a mighty book, you must choose a mighty theme. No great and enduring volume can ever be written on the flea, though many there be that have tried it.

— Herman Melville

This blog entry is my white whale. It has been my nemesis since the genesis of this blog. I have never been able to tame it or capture it. My goal in starting the Catfish Stew blog was not, like so many other baseball blogs, to second-guess The Management, but to express what it feels like to be an Oakland A’s fan. If I have failed as a blogger, it is because I lacked the willpower to bring myself to tell this story, to confront the core pain of my mission. Would Herman Melville have succeeded if he had tried to write his masterpiece without ever once mentioning Ahab’s peg leg, the scar that drives his obsession? If you face the Truth, it hurts you; but if you look away, it punishes you.

Load the harpoons, gentlemen, it is showdown time. Today, my adventure as a baseball blogger ends. I’m going down, and I’m taking Moby Dick with me.

Continue…

     
This is Ken Arneson's blog about baseball, brains, art, science, technology, philosophy, poetry, politics and whatever else Ken Arneson feels like writing about
Google Search
Web
Toaster
Ken Arneson
Archives
2021
01   

2020
10   09   08   07   06   05   
04   

2019
11   

2017
08   07   

2016
06   01   

2015
12   11   03   02   

2014
12   11   10   09   08   04   
03   01   

2013
12   10   08   07   06   05   
04   01   

2012
12   11   10   09   04   

2011
12   11   10   09   08   07   
04   02   01   

2010
10   09   06   01   

2009
12   02   01   

2008
12   11   10   09   08   07   
06   05   04   03   02   01   

2007
12   11   10   09   08   07   
06   05   04   03   02   01   

2006
12   11   10   09   08   07   
06   05   04   03   02   01   

2005
12   11   10   09   08   07   
06   05   04   03   02   01   

2004
12   11   10   09   08   07   
06   05   04   03   02   01   

2003
12   11   10   09   08   07   
06   05   04   03   02   01   

2002
12   10   09   08   07   05   
04   03   02   01   

1995
05   04   02