Daniel Steinberg: Distributing the Future broadcast February 10 2006, the Mechanical Turk, Voice 3.0, Java podcast and reading pigeons.

Announcement: The future is going to be bigger than the past. There is going to be more happening, connected systems, connected devices, connected data. What can we do in that world that will be interesting, powerful and useful? Changing the world through capturing the knowledge of innovators. The future is here, it's just not evenly distributed yet.

Daniel Steinberg: Welcome to "Distributing the Future" from O'Reilly Media. I'm Daniel Steinberg and each week we bring you the technology and the people behind what you use now and what you'll use next. This week Rael Dornfest and Tim O'Reilly look ahead to next month's emerging technology conference, Yahoo's Jeff Bonforte talks about Voice 3.0 and voice recognition, Chris Adamson surveys a current crop of Java podcasts, and we look forward to the arrival of pitchers and catchers at spring training with a selection from Baseball Hacks.

Daniel Steinberg: Distributing the Future is brought to you by GoToMeeting, host your next meeting online with GoToMeeting, everyone logs onto your meeting during your conference call, so they see your computer desktop on their computer screen. It's like meeting in person but less expensive and time consuming. With GoToMeeting, you can demo products, present or train just like you're in the same room. GoToMeeting is a low flat monthly rate, so meet as often as you want for as long as you need. Try it free today at gotomeeting.com/dtf. That's gotomeeting.com/dtf.

Daniel Steinberg: A lot has been said about harnessing the collective intelligence of the web. Our first two stories take two very different approaches, Yahoo's Jeff Bonforte talks about how the "did you mean" feature that you see in Yahoo and Google searches benefits from mining what people clicked on after search. He asks how they might apply in voice recognition? In other words, you'll benefit from the millions of searches by people you've never met. That's different than what you get on sites like digg.com where popular stories rise to the top, or on amazon.com where people actively review books and products. Artificial intelligence is being rethought. Maybe the machine doesn't do all the work, maybe the machine does the parts that it's good at, but there's still a need for a person inside. And this first part of Rael Dornfest and Tim O'Reilly's emerging technology conference call with various members of the press. They discuss the role of the Mechanical Turk, as Tim points out, when you move from using a shovel to using a bulldozer, the nature of the work performed by the human and by the machine changes. But the human is still integral. As our computers connect us greater amounts of data force in an instant, how can our friends and our colleagues help us? Rael outlines the problem.

Rael Dornfest: People are completely, just overwhelmed by the amount of information between RSS and email and now a lot of these upcoming and eventful feeds of events in your area, events not in your area and so forth. The sheer amount of information that we're dealing with is staggering, and you look at the tools that are being built around this stuff and they're still very much in their infancy.

Tim O'Reilly: And a broader context is not just about client-side applications for filtering and attention, it's also about the way that the internet as a whole acts as a filter. And the applications are going to be most interesting out there on the internet as a whole, are ones that now have leverage collective intelligence to help do some of these filters.

Rael Dornfest: Yeah, absolutely, it's more than that. I mention RSS specifically as the fact is that more and more RSS readers being built and they haven't yet done what digg and various other web based applications, other applications that do harness collective intelligence have done, and that is hone some of these things down. You're still very much reading in a vacuum. You still very much have no concept of on BlogLines for instance, that Tim and I might read the same feeds and we might know each other and it might matter to me if Tim goes ahead and marks something read or unread. It doesn't work in email that way, but in RSS it certainly can.

Tim O'Reilly: I'm actually looking at the program right now and I thought that maybe I'd talk about the few of the talks. The talk by Felipe Cabrera of Amazon about artificial intelligence. "Oh my god", everybody runs screaming, but you know, it is one of the hidden secrets of the web that it has in fact brought a lot of the promises of AI too us, not by being too ambitious, but just by figuring out things like how to harvest reader opinion to get better search results, how to give product recommendations. And there's a lot of really interesting things going on at a place like Amazon and Google, where we're actually bringing this all bug-a-boo out of the closet again and saying, you know what, it actually is working.

Rael Dornfest: The talk is all about artificial artificial intelligence, it's realizing that it's the people in the machine that should make a lot happen. A lot of people, when A9 released their photos of storefronts mapped onto Yellow Pages, were still trying to figure out how they managed to line the map property and so forth. And it was actually one of the first applications of their Mechanical Turk, was actually to line those things out, there were people doing that by hand.

Tim O'Reilly: There's also an element that I think is really important here, is the wonderful phrase used by a guy named Uman Tang, as a founder of a company called Boxxet, and he described what they're doing as bionic systems, where they're really trying to get, kind of like this six million dollar man, where's sort of a fusion of human and machine, and that's what the Mechanical Turk is doing, by basically a lot of the process is done by the computer, but there's some little bit where it's just more efficient to get the person to do that bit.

Rael Dornfest: Who is that again Tim? I don't know that.

Tim O'Reilly: Boxxet, b-o-x-x-e-t, they're probably still sufficiently in stealth. I can't tell you what they do, but I will tell you that they're kind of working on the same idea, and I love the way, in their pitch they use that phrase bionic systems, with explicit reference to the six million dollar man, because it really is, I think one of these deep themes. There's actually also a wonderful talk by George Dyson on that subject that he gave at Google, about Google. And it ends with this sort of chilling and very provocative image, that if there were an AI, what would it really look like? It would look like this giant temple with all these happy acolytes, you know fed, you know happy working, serving the beast.

Rael Dornfest: Well Tim, you're kind of making me get this image of people in supermarkets now, who essentially are just organically grown manipulator arms for the scanners, because that was cheaper to use human grown end-effectors and swipers than try to build a robot that did that, but all the good work is being done by the machine.

Tim O'Reilly: What we're really starting to see, is technology is becoming so pervasive, that it's changing the nature of the relationship between human and machine. I mean on the old computer, it was pretty clear, there was this computer and you kind of told it what to do. And now it's something we interact with it, and the whole idea of results to get better the more people use it, I mean that's been one of the key concepts that we've talked about as one of the central principles of Web 2.0.

Rael Dornfest: It's especially interesting, it's almost like back in the day, we had these 16 bit computers running at a certain rate, that were very limited in what they could do. While their understanding and computational intensity hasn't sufficiently moved along, what has happened is their ability to handle mass amounts of data has gone up, and now we're sitting in front of these now 64 bit machines and the Google machine and so forth, and we just can't cope with it all anymore. And what the focus is now is on trying to find ways for us to help each other, and help the machines by giving them hints as to what is it that we'd like to get access to and have flowed to our direction.

Tim O'Reilly: Yeah, but I also think about in terms of the industrial revolution, think about heavy equipment, it's really a fusion of man and machine, in a way that's far more profound than say, the shovel, so here's a guy with a shovel, and they're clearly the human effort is the larger part of the effectiveness. But as you move along the continuum, a man with a bulldozer is providing some direction, the bulldozer's actually doing most of the work. For a long time there was this vision that this would all become completely artificial, and that's kind of like the idea that the bulldozer would have no driver. What we're figuring out is how we hook the people into the machine in ways that make it possible to do things that neither could do alone. A good example, I just wrote a column for the latest issue of Make magazine, which should just be hitting the stands, about my eye surgery. They kept saying, while the laser's in my eye, the surgeon is saying "keep looking at the red dot", with quite some urgency, "keep looking at the red dot". And I say "what happens if I don't keep looking at the red dot, you know I mean, is my eye going to get you know, permanently...", and she said the computer will stop. You know, it'll take longer. Yeah, you can't hold yourself still about this without that amazing feedback loop where there's a scanner looking at my eye and the surgeon does some chunk of it, but without that computerized tool, this is not something that would be possible.

Rael Dornfest: Because its reaction time would be so much shorter than the human surgeon's would be.

Tim O'Reilly: That's right.

Rael Dornfest: So it can stop before it hurts you.

Tim O'Reilly: That's right.

Rael Dornfest: Yeah, okay.

Rael Dornfest: That's the same kind of thing, it's sort of like, we can't filter through all this information, so one of the frontiers is figuring out how do we get the computers to do the part they can do, we to do the part that we can do, and together we manage to get the job done.

Daniel Steinberg: Tim O'Reilly and Rael Dornfest, how people and machines can work better and smarter together. This was part one of their preview, for next month's emerging technology conference in San Diego. You can find links in our show notes at oreillynet.com/future.


[music]

Daniel Steinberg: Voice 1.0 has been with us for a hundred years. It's a telephone that you know and love. Yahoo's Jeff Bonforte talks about Voice 2 and 3. As we've moved farther away from the traditional telephone model, there are many cool opportunities that Jeff discusses. But he also realizes that there's plenty of room for people who aren't ready to move to Voice 3.0 anytime soon. Take a guess, how many people in the US do you think still use rotary phones? Jeff will have that for you as part of his talk from the emerging telephony conference.

Jeff Bonforte: Yahoo we think of voice in terms of sort of a 1-2-3. Voice one would be something we all recognize, right, it's sort of the boring dial-tone part of the world. What's nice about it, is it has a pretty established business model, right, some people say a trillion dollars. It's a hundred years old, consumers certainly are comfortable with this business model, angry but comfortable. They certainly know it exists. But I lump into Voice 1.0 types like Vonage really, they just want to be this PSTN experience, they're just replacing their last mile, they've sort of changed the economics a bit.


But when we shift from Voice 1.0 to Voice 2.0, we change the sort of medium by which we're carrying the media, and that makes things a little bit interesting, because voice becomes a lot more like software, and Skype and Messenger today have this sort of Voice 2.0 service, right, it kind of behaves like PSTN, there is this sometimes it does interact like with ours and Skype's with the PSTN. But we have the powers of servers and desktops to do something a little bit more interesting, but this itself is not an end, right, we don't all want to just simply throw away our home phones, which actually work pretty well, and move to something simply on the desktop or even wired to the desktop or wireless from the desktop. We really want something that's more interesting and that's where I think Yahoo can really contribute something. Because as we move from Voice 2.0, voice simply as a platform, as a software basis, to Voice 3.0, it starts to get really interesting, because now voice is starting to integrate with the very things that we do on the internet every day, that we are spending hours and hours on our PCs and our laptops doing on the internet and increasingly with our mobile phones.


What I like about Voice 3.0 is we move away from the concept of a call, in fact you may be talking to the mail server, you may be talking to a community of people in groups, you may be talking to a group of buddies in 360, or you may be talking to a game on Yahoo Games. Who, it doesn't really matter, of course you could be talking to your friend around the globe, but maybe you want to do something a little bit more interesting, maybe you want to be in the middle of a call and say "hey what's the score on the Dodgers game again?" and have the system say "Dodgers are leading the Padres at the bottom of the 7th, 3-2".


Now, does that mean the whole world has to follow behind and run as fast as they can to 3.0. Absolutely not, right, Voice 1.0, the dial-tone, it's still a very healthy margin, many of us would say too healthy a margin, business, and Voice 2.0 also healthy, great business. But for someone like Yahoo, our world really exist in a Voice 3.0 world. I don't want a Tintin when I grow up, I want to be Yahoo when I grow up, and I want to integrate voice into the very experience of what people do with Yahoo. So as you see Yahoo, we see sort of voice as an integrated fabric piece or an element of that and what's going on. So, the sort of downside of Voice 3.0 world is, the business model is not as fleshed out in the 3.0 world. But consumers will, should expect to be paying for things sort of on an application basis or transactional basis instead of a call or per minute basis. There's also a lack of consumer awareness. Consumers have been using some of Voice 3.0, these VXML based, and they're getting better and better, these applications, but I don't think they have the sort of awareness or comfort yet to work with those. So basically the PSTN guys don't really want to go to Voice 2.0, and there are some within, let's say Bell South, and they say "oh we want to race them, we want to be this cool software voice". The truth is that the consumers aren't really willing to go there, the phone works very well. In fact, I learned this stat the other day, there are more people with rotary phones than Vonage servers in America. 1.2 million people still have rotary phones, by the way, still paying monthly for those rotary phones.


[Audience laughter]


And I think there's like a million Vonage subscribers now today, at what, 600 dollars per subscriber or whatever it costs to acquire those guys. So, that world is not going to change fundamentally, nor necessarily does it need to change, because it accomplishes what it does very well. Now, if you look at the Vonage guys or the PSTN replacement players, they can do something a little bit interesting, right, because there's a server at the back end of that service, of course many of the PSTN services will do this, but usually it's for efficiency or cost, not necessarily for increasing service. The mobile guys actually have been racing towards more integrated Voice 2.0 services and data type models. Mostly because they just want to keep their average revenue per user up, right, so this ARPU, people are in a constant race, since people offer 300 minutes for 20 bucks a month, where you've got to find another way to get that $60 phone bill, I think it's like 57-65 dollars as the average monthly phone bill for a mobile phone, stayed completely constant for the last five or six years. And the way they're doing that is changing that mix, right. But still, the metaphor of the phone itself is somewhat limited in saying "well you can go as well as the hardwares and the networks". Well, the internet is a great place to be, and if you really want to integrate with the applications and the services and the data and all the content that's out there, all the media relationships that you want to do, well for someone like Yahoo that's where we're most comfortable. Let's talk a little bit about voice recognition. My theory the difference between great voice recognition and good recognition is a little secret sauce that I think Yahoo gets to bring to the mix.


So let's say for example, I'm on a search engine and I type and rekognize and I admit it's misspelt. I type in rekognize, what does Yahoo do right, immediately comes back and says "did you really mean rekognize?". Now we don't spell check every search result, we simply know that nine times out of ten someone that typed in the word rekognize with a k really clicked on a link that was relevant to recognize without the k, and so the search engine gets smart. What's interesting about that is that voice recognition, today the way voice recognition gets accurate is to limit the number of things the user might be saying, so it's called dynamic grammars right, we create dynamic grammars on the fly, we reduce the number of things the user might be saying, so we ask the user, we say "hey, pick us sports category", they say NFL, we go oh great, NFL, that's a finite number of words we're going to hear from this person, they might want to hear about games or plays or players.


Say that what you want to hear more about, "players", great, the NFL has 28 teams, probably a 100 players by team, 2800 guys, we are going to start to get pretty accurate here. We want to get really accurate, name the team that your player's on that you want to hear about. They say the Broncos, then they want to hear about Jake Plumber. Now we get really accurate, right, because we said the Broncos, we got down to a team name, possible list of about a 100 people. Now voice recognition gets very accurate.


Yahoo on the other hand has search, and this huge asset, one of the top two search engines in the world. And what's great about this asset, is it's constantly learning, and it doesn't learn through this sort of fuzzy logic, it just learns because of massive amounts of user input. So if a Polish quarterback whose name I can't pronounce, I've never heard of, throws the winning touchdown on last second of Superbowl, a million people immediately walk over to their computer and type in the various thousand ways that they can spell those Polish quarterback's name right. And the system within about two seconds knows exactly how to spell this guy's name, even though the search, there may have been one search on this guy's name six hours before that. Well, six minutes after that throw, the system has a million searches on this guy's name, and humans being as errant as we are, have misspelt it a thousand times. Well, voice recognition can kind of work the same way, right, we can dump what the voice recognition engine hears into the search engine, and the search engine who learns so quickly because of the amount of input it's getting every single day, gets very smart. "Did you mean Polanski?", I did mean Polanski, I didn't mean Polonski.

Daniel Steinberg: Jeff Bonforte of Yahoo at emerging telephony. The last time we heard from Chris Adamson, he was telling us how to get started with Dance Dance Revolution. Chris is a Java developer who edits our java.net and ONJava.com sites, and this time his piece is a little closer to his day job. He's been listening to some of the Java podcasts and he recently ran a two part series on some of them, on ONJava.com. Here's an audio sampling that gives you a taste of the wide variety available in Java podcasts.

Chris Adamson: Java means code, according to its critics, lots and lots of code. As you might expect, the articles we run on ONJava.com are heavy on code examples, sometimes they run a couple hundred lines long. So it was a bit surprising to hear that there are podcasts about Java programming. This media initially seems orthogonal to the subject, like say painting, about music. I interviewed several of the people behind these podcasts, for a two part ONJava article called the Java podcasters. In looking around, I found that not only are there a half dozen or so well known Java podcasts, but that they're very different from one another. Probably the best known of the bunch is the Java Posse, around table discussion with four Java developers discussing the week's Java news. The round table format allows them to cover interesting news and opinion in a format you can listen to anytime. Moreover, the Skype based conversation makes it easy to bring in interview subjects.

Java Posse: Hi! Welcome to the Java Posse episode 28. This week an interview with Howard Lewis Ship of the Tapestry and HiveMind projects. Java Posse talker: Tell us a bit about yourself and how you first got started with Java programming, go all the way back, reach reach back.

Howard Lewis Ship: Let me just jump in, how did I get involved with Java programming? Well, it's kind of fun, I came up through the ranks as it were, and did a lot of work in PL1 for the DOS operating system. So that's about as far as you can get from object-oriented programming, it was like one step past COBOL.

Chris Adamson: Another approach to podcasting about programming is simply to talk about what's on your mind. In effect, to do an audio blog. That's how Tim Shadel describes his Zdot podcast, in which he says he focuses on what he's actually done and then comments on it. This is an approach that allow us for depth, even without code examples, because it speaks from experience and expertise.

Tim Shadel: That object, if you put that in your application context, it lets you specify a properties file, and anything in the dollar signed squiggly brackets that it finds in your string file, it will actually go resolve against your properties file. So this is great, that means that we can actually have a shared context file, our string context file does not have to change then, when we move from development to test to production, it's just a simple properties file, that's easy to leave outside the classpath, to put in a symlink to a directory.

Chris Adamson: Another way to get expertise is to seek out known experts. Michael Levin's Swampcast is essentially all interviews. Levin told me "Swampcast is for both you and for me. I get a lot out of interviews and the people I talk with keep me hopping." Levin had a particularly interesting interview with Thinking in Java author Bruce Eckel, talking about the CD that comes with the book, Levin had a bit of an admission to make.

Michael Levin: I say I'm embarrassed to say that, because even though I've...

Bruce Eckel: You haven't listened to the CD yourself.

Michael Levin: Well, I didn't buy the book, I mean, old habits die hard and you know, the books are available on the website, so I downloaded the book and I printed it out. It all weighs about 15 pounds.

Bruce Eckel: You know, well, people, you know, it's... That's been an interesting experiment and people have done that...

Chris Adamson: One of the things that's so likable about the podcast format is its informality. For a language notorious for dotting its eyes and crossing its teeth, it's nice to just shoot the breeze once in a while. That's certainly the attitude of drunkandretired.com, the programming podcast hosted by Charles Lowell and Michael Cote;. They told us their show is "just talking about whatever it is we happen to be thinking about, just like we would if we were rocking on our front porch swing, sipping bourbon and watching the sun go down". In practice, that means their show can be about anything from Ruby On Rails to zombie movies, or for that matter, the unforeseen hazards of on-site consulting.

Drunkandretired1: You know, I was thinking this morning, like I was lying in bed as I often do, about Star Wars, yeah, I was thinking that there probably had to be some pretty intensive software to operate that Death Star array, so I think it would be interesting to know how they do all these complete little side vignettes about parts of the empire that you don't see very often. It would be interesting to see what was behind the programmers of the Death Star.

Drunkandretired2: And you know the tragic thing thing? They probably all got killed when the Death Star blew up.

Drunkandretired1: Yeah, I know!

Drunkandretired2: And you know they were just programmers.

Drunkandretired1: Exactly, they were programmers with those little round black helmets. [laughter]

Chris Adamson: One sign that the Java podcast is truly arrived, is the emergence of the single topic podcast. It suggests a certain evolution, you start with general text shows, then programming shows emerge, then shows about just language get started. Taking this progression a step further, Roumen Strobl hosts a podcast focused on a single podcast, Netbeans, the old Java IDE from Sun. Well this is the news to the Java podcast, and it's still finding its way, the single topic format is interesting, because it allows for significant depth. Instead of talking about the high level features of the IDE, Strobl can discuss specific modules like Metisse, integration with other products, even documentation and training.

Roumen Strobl: Tellem Wollegonk, who is one of our technical writers, asked me several months ago about recording of flash demos. Basically it's a normal tutorial about the contains that will demo icons, and if you click on one of those, a new window opens and you get flash demo of them all of the part of the tutorial you're reading. For somebody who wants to learn to use the IDE, this is just awesome. You can learn all kinds of tricks visually, and listen totally, as he explains how to use Matisse.

Chris Adamson: The various podcasts operate on different schedules. Some, like the Netbeans podcast and Swampcast release on an ad-hoc basis, when they have something to say. Java Posse on the other hand tends to have a more predictable weekly schedule. The five podcast sample for this article show that there's both breath and depth in podcasting choices for Java developers, and all you need is headphones. In this case, curly braces are not required.

Daniel Steinberg: Chris Adamson with his Look at Java podcast, will have links to his related articles in the show notes. If you have an idea for a story that you'd like to do for us, send email to future@oreilly.com.

Daniel Steinberg: There's snow on the ground here in Cleveland, and yet you ask fans of major league baseball now that pitchers and catchers will be reporting to spring training this coming week. Baseball season here usually starts as the snow melt, and ends as the first squalls blow back in. Sure, they're early in late season games when the snow is falling, but the weeks before opening day still feel like the official sign that spring is really on the way. Joseph Adler's Baseball Hacks has just been released, containing 75 tips and tools for analyzing and winning with statistics. Here's part of Hack number 4, follow pitches during the game.


[baseball game audience noise]

Daniel Steinberg: People who don't watch baseball all the time think it's a boring game. They don't think much happens. Every minute or so the pitcher throws a ball and the batter swings and that's it. Over time, fans realize that a lot is going on, there's a lot of strategy involved in pitching to each batter. This hack helps you follow this battle, the subtlest and most elegant part of the game.


In addition to making the game a lot more interesting, understanding and following pitching strategies raises a lot of interesting questions to ponder as you watch. Is the pitcher throwing the ball where he should, or is he throwing it at the wrong spot? Is he having trouble with a certain pitch? Does it seem like the other team always knows a certain pitch is coming? Finally, watching the pitcher and catcher can also tell you who's making the decisions. Usually the catcher selects the pitches, but there are exceptions. You can learn a lot by watching.


"Following the pitching strategy." The pitcher's primary goal is to fool the hitter into swinging in the wrong place at the wrong time. There are a lot of ways to do this. Sometimes the pitcher wants to throw strikes that look like balls, so that the batter doesn't swing at a called strike, or that balls that look like strikes, so the batter swings at the ball and misses. Other times the pitcher wants to throw the ball where the batter can't hit it very well. Often this means a high inside fastball. And still other times the pitcher throws the ball at different speeds to fool the hitter to swinging too early or too late. Pitching strategy can be subtle and complicated. Pitchers often throw one pitch, say a slider on the outside corner, to set up another pitch, say a slider outside.


Good pitchers, usually catchers actually, remember what they threw to a batter in previous innings, games, or even seasons. Professional ballplayers and coaches spend hours studying videos before each game, to try to forecast what a pitcher will throw to them and when. Catchers spend hours reading notes and watching videos to figure out what pitches hitters expect pitchers to throw and at what times, so they can adjust their strategies. And then, ballplayers adjust to hitting strategies to what they think are the pitchers' new strategies. You get the idea. It's really tough to predict what each side will do, unless you have a lot of time and a big video collection.


As a fan this is practically impossible, but if you watch a lot of baseball, you can start to pick up on a few patterns and pitching strategy.


"Set up a pitch outside." A pitcher with good location will throw a couple of strikes on the outside corner. Many batters will assume that they look like balls, maybe they're sliders or curve balls, that move into the strike zone at the last second. After two called strikes at that spot, the batters condition that balls in that spot are strikes. The pitcher will then follow with an unhittable ball just outside the strike zone to strike out the batter.


"Following breaking balls with a fast ball." Many pitchers will throw a few consecutive breaking balls to get the batter used to seeing slow pitches. Then the pitcher will throw a fast ball, trying to get the batter to swing at the ball after it already crosses the plate.


"Follow fastballs with a breaking ball." This is the opposite of the previous strategy. A power pitcher, usually a guy like Eric Gagne, who can throw a 95 mile an hour fastball, folds a few fastballs with a really slow pitch. Often you will see the hitters swing at the pitch long before it crosses the plate.


"Always throw the same impossible to hit pitch." This isn't really much of a strategy, the only pitcher who's done this successfully is Mariano Rivera, the Yankees ace reliever, who throws a cut fastball. His pitch is a high inside fastball, that moves in on the hands of left handed batters and away from right handers. He throws the ball in a way that makes it move at the last minute. Sometimes he gets strikeouts but often a hitter will hit the ball weakly out the inside of the bat, grounding out. His pitches cause more bats to break than almost any other pitcher in baseball. Incidentally, the only team that seems to be able to hit his pitch is the Red Sox, probably because the Yankees and Red Sox play each other 19 times during the regular season.


"Move the player off the plate." Many batters like to stand over the plate to hit balls on the outside with the sweet spot of the bat. If you hit the ball on right spot, it travels a lot farther. A common strategy for a pitcher is to throw an inside fastball and a batter to scare him and to get him to move off the plate.

Daniel Steinberg: An excerpt from Baseball Hacks by Joseph Adler. Thanks also to Rael Dornfest, Tim O'Reilly, Jeff Bonforte and Chris Adamson. Distributing the Future is brought to you by GoToMeeting, online meetings made easy. Try it free at gotomeeting.com/dtf.

Daniel Steinberg: Thanks to all of you who emailed and phoned in with your comments, criticisms and suggestions. If you have ideas to share with us, send email to future@oreilly.com. I'm Daniel Steinberg, Distributing the Future is a product of O'Reilly Media. Special thanks to David Batino for composing and performing this theme. Visit David at Batmosphere.com. This program was produced on Mac OS X Tiger, using Soundtrack Pro, Bias Peak and Audio Hijack Pro.