MLB.com

MLB opened in Tokyo

When the Major League Baseball season opened this year in Tokyo with a game between the New York Yankees and the Tampa Bay Devil Rays, Americans wanting to watch had a choice. They could either wake up in the dark to see it live on television, or go to MLB.com, subscribe to its video service, and watch the game at their leisure on their computer. And as the season progressed, they could watch a lot more: some 300 games a month?available live, as well as archived, with archived games shown in their entirety, as well as compressed down to less than 20 minutes of highlighted play. (Not every game is available to everyone. Due to licensing restrictions, fans cannot watch their local teams, either at home or away. Blackout restrictions are even more severe in Japan.)

This streaming video service, called MLB.TV, is the premium offering of a website that attempts to put everything you ever wanted to know about the baseball season in one place. "We want to exploit the assets of baseball on behalf of the baseball fans," says Joe Choti, chief technology officer for MLB Advanced Media, which operates the website. “⁠It's the stats, the video, the audio, the player information, the fantasy baseball information, the ability to purchase tickets, to buy official baseball memorabilia⁠”⁠ all in one place.

MLB.com's popular Gameday window shows the play-by-play action of every game played, including the position of every pitch as it sails over home plate and every batter with his associated statistics. The feature is especially popular with people at work. You can order tickets to any game, get the schedule for any team, view the standings for each league and its divisions, as well as the wildcard race. You can even read the official rules of baseball, which seem as intricate and arcane as tax law.

5 billion page views

All of this has attracted a huge and growing amount of traffic. Midway through the 2004 season, Major League Baseball Advanced Media claimed it had registered a record 5 billion page views, versus just 1.85 billion for the same time period in 2003, and 4 billion page views for all of last year. The MLBAM claims it attracted nearly 500 million visitors during the first half of the season, versus 650 million visitors to the site for all of 2003. In a given baseball season, the site delivers some 44 million minutes of streaming media and 2,500 full-length games. With all this growth, MLB.com's objectives remain the same: take a sport with 30 teams and 30 owners whose ideas about how baseball should be marketed and make it look like a single cohesive entity on the Internet.

MLB Advanced Media

Major League Baseball Advanced Media is based in Chelsea, New York, 50 miles north of Manhattan. Of MLBAM's 120 staffers, 30 actually run the site?"the same number that have been running it since day one,"says Choti. The hardware/software infrastructure is primarily Sun. Last March, MLBAM and Sun announced a two-year alliance to build new data centers for the site. The centers will include Sun Fire servers, Sun storage systems, Java Enterprise System software, and Sun's Digital Asset Management Reference Architecture. Statistical data that is collected and transferred from stadiums across the country via a TIBCO network is crunched by the Oracle database. The entire site consumes several terabytes of data storage. MLB.com has about 600,000 building block components, defined as anything from a scoreboard entry to a statistic. About 25 to 30 components comprise a given page, making the site unusually dynamic.

Streaming media and Gameday

MLBAM has taken advantage of the increasing number of broadband connections in both the U.S. and overseas to add more video streaming. At the beginning of the 2004 season, the company announced a deal with Akamai to move streaming content closer to MLB.com's international audience in 65 countries. "We acquire over 300 baseball games a month," Choti says. "After providing them as live streams, we make them available in our archive about 90 minutes after the game is over. We also metatag the events of the game?home run, doubles, double play, pitcher changes, etc.?so that people can see, for example, a specific Barry Bonds hit home run."

Metatags

MLBAM's programming team also links the game box scores with specific events within the video stream, so that users can look both at the numbers and the plays behind those numbers.

"The other things the metatags do is let us condense games," Choti says. "We can boil a full game that runs 3 to 3-1/2 hours to 18 to 20 minutes. You don't see a player stepping out of the box, spitting on the plate, arguing with the umpire. You only see everything that meant something to the game action. For the fantasy player, we also have a product called Highlights Direct. A plays can pepper his roster with players that are on his or her fantasy team, and on a daily basis, we email the results. He or she can then look at the video that's associated with those players, seeing what they did the night before."

Gameday

If video streaming is MLB.com's premium offering, Gameday is the free service for the masses?available and fully functional on any computer with at least a dialup connection. "Gameday has taken quantum leaps since we introduced it in 2001," Choti says. "We have re-architected it from the ground up," During play, the Gameday Window shows every pitch, hit, error, walk, stolen base and out. All of this and more is recorded by an MLB.com stringer sitting in the press box. Using a laptop, the stringer notes who advanced, who got caught stealing, who scored, as well as the location of every pitch from the vantage point of the umpire. All of this is displayed as it happens, via Macromedia Flash, using batter diagrams, a box score, and brief play-by-play written descriptions. "That stringer at the ballpark is the most fundamental piece of our business because that is where all the data starts," says Choti.

"We use a TIBCO network to carry that data back to our production servers where it is manipulated and presented in Gameday. But we don't update numbers in the stat section during a game because there have been plenty of cases where an official scorekeeper has changed his mind about a particular call?a hit that becomes an error, for example. In one game, there was even an issue as to whether a hit was a home run or a double?it was uncertain what the official score was at the end of the game.

Uniform management

Gameday is popular. More than 500,000 people at a time access Gameday, each one effectively requesting updates every six seconds. That load, says Choti, puts such a demand on resources that, on another website, it would constitute a denial-of-service attack. One challenge in offering Gameday is that the service cannot tolerate latency. "When a user make a request every six seconds, he or she may actually be hitting a different server each time. But I cannot afford to have you hit a server in California that is on pitch three and a server that’s in New York on pitch two. So we serve all of that traffic out of our production data center in New Jersey?so that our battery of servers can serve the same data at the same time."

Load balancing and compression

Choti says that MLB.com, at times, handles some of the highest load of any website. "n a given day we can go from several hundred to 6 million page views an hour,' accelerating in a matter of seconds. The site's technology partners don't quite believe those numbers until they see it. The Akamai deal has been part of MLBAM's answer for handling the volume. The company also uses NetScaler for load balancing, and it does a lot of compression. "Our pages are full of images and data. We need to crunch it up as tight as we can get it."

Delay for accuracy

Looking at the Internet feed while getting live coverage on radio or television, you realize that the information on the website lags a bit. It is real-time, almost, but not quite. Choti says the delay is not due to a bandwidth bottleneck: data can be displayed on the Web within three seconds of input. "We want clean data," Choti says. "We'd prefer the stringer be slow and get the play right, rather than be quicker and get the play wrong.⁠”⁠ The raw data is pushed into a massive Oracle database and made available online. Between the statistics and the mass of biographical data, the diagrams and the descriptions, you could be your own announcer and give your own play-by-play account.

Statistics, served any way you want them

One of MLB.com’s most impressive services is its astonishing breadth of statistics and information about the game. The site features extensive biographical and statistical information about each player: How tall is Ichiro Suzuki? 5 feet, 9 inches. What was Hideki Matsui’s batting average last year? .301. What school did he attend? Seiryo High School. You can find team histories, biography information on baseball commissioners, statistics on the annual All-Star Game, historical information and baseball’s exhaustive official rules. The site is a "stat-head's " dream, answering questions that only the most obsessive fan could care about.

Sidebar:MLB trivia

Q:How did Miguel Tejada do against left-handed batters in July?: A:250 average, with 3 home runs, 4 doubles, and 9 RBIs.
Q:How did Barry Bonds to against Hideo Nomo in 2003?: A:In four at-bats, he hit .300?including one homer and a double, with 2 RBIs.
Q:What was Hideo Nomo’s ERA [earned run average] with the Dodgers?: A:2.54 in 1995. 5.05 in 1998. 3.39 in 2002.
Q:How did Ichiro Suzuki, Carl Crawford, and Derek Jeter compare as of mid-August?: A:.511 versus .493 versus .474, with Jeter getting more than twice as many home runs (15) than the other two, combined. (The site shows a graph of each player’s batting average since the beginning of the season.)

Imposing uniformity

Most websites? whether for a company, a publication, or a service like eBay?are beholden to a single management team. One of MLB.com’s more unusual challenges has been in dealing with 30 team owners. While this consistency might seem an obvious design approach, it took some doing because the game of baseball as it operates in America is still controlled by a group of owners who don't always agree on anything except that the pitcher's mound should be 60.5 feet from home plate. Indeed, the very structure of the game promotes regional distinctions. Unlike a tennis court or soccer field, the outfield at each baseball stadium has different dimensions and ground rules determining what is fair and what is foul. And until MLB.com, each baseball team's website, for better or worse, reflected the personalities of each club.

Centralization

The 30 clubs ran 30 different websites of varying degrees of quality?some of them amounting to no more than an online brochure. The theory behind centralization was to create 30 great websites for the fans, reduce the cost of technology by spreading it over 30 clubs, and augment revenue by selling baseball memorabilia. The proposition seemed simple: if the clubs would give up some control, the quality would rise, the costs would sink, and everyone from fans to the teams would be winners. But for a few teams with established websites, the transition was difficult. Some fans saw the new standardized site and went elsewhere. But over time, they've returned, as MLB.com's rising “⁠hit⁠”⁠ numbers demonstrate.

Future of MNB.com

Those numbers will probably continue to rise?but not without much scrambling at MLBAM. “⁠Baseball fans are pretty demanding, as they should be,⁠”⁠ says Choti. “⁠And every time we think we've provided the functionality and data they are looking for, some fan has a great new idea that sends us back to the drawing board.⁠”⁠

Sidebar: Gameday

Small screen, a lot of information

While MLB.com's streaming media games show the power of the Web, its Gameday feature demonstrates MLBAM's programming acumen. The Window consumes just a third of a 17-inch screen, yet tells viewers everything they could possibly want to know about the game. The information is clear and easy to follow: in many ways, the model a how to design a real-time information site.

Game on your desktop

On an August afternoon, the Oakland A’s are in Baltimore playing the Orioles. It’s the first inning, and a diagram shows that A’s first baseman Scott Hatteberg is on the plate facing Erik Bedard, with Mark Kotsay at first. Both Bedard's and Hatteberg's pictures are also displayed. Beddard has pitched 14 balls, and his first two pitches to Hatteberg are low and outside, as shown in a diagram with a rectangle representing the strike zone. To the right is a box score, constantly updated. A pitch-by-pitch list shows the result of every pitch made, including fouls. Alternatively, you can just view scoring plays or a play-by-play summary.

At the top right-hand corner is a rotating scoreboard of other games, both in progress and scheduled. Click on any game in progress, and Gameday shifts to that contest. At the bottom are displayed headlines of baseball news: The Los Angeles Dodgers beat the Chicago Cubs; Colorado’s Joe Kennedy gets tossed from the game; San Francisco Giants win. Click on any headline and you can read the full report.

Sidebar: A conversation with Bob Bowman

interviewer:Bart Eisenberg

When Bob Bowman was recruited to run Major League Baseball Advanced Media and turn 30 team websites into a cohesive whole, he already had plenty of practice. Bowman previously headed ITT Corporation, once one of America's most sprawling conglomerates. He now oversees a staff of 180 people, 120 in Chelsea, New York, the others based in the field.

Q:MLB.com has a tough balancing act: it represents baseball as a whole and 30 separate teams. Were club owners reluctant to give up their own vision of how they should appear on the Web?: A:We've had clubs that, for understandable reasons, think they would do a better job representing their club to their fans. Other clubs are concerned about anything that's centralized ? they want to generate local revenue and use it locally. So for the first year, people were more questioning and skeptical. We’re now entering the third year, and they've grown more accepting of the idea and are looking at how to leverage the site to their advantage.
Q:Was the site designed by consensus or did you sell a vision?: A:We kept the clubs intimately involved every step of the way. We got input, but we presented what we thought made sense. Clearly we're very different than ESPN [the Disney-owned American sports television station]. We try to present a lot of information in a very uniform way so people can quickly find out what's going on in baseball.
Q:How do you work with each team?: A:They designate people, and we designated people. We have club relation people that manage 10 clubs each, and they work with the clubs every day. It's a very involved and time consuming job.
Q:Do you do anything else besides baseball?: A:We stream more live sports than anybody, and therefore we are talking to a number of sports entities about doing comparable services for them. We've created a subsidiary, called AM/PM Sports. We streamed some football games this fall. We're going to do a lot of basketball games this winter.
Q:You give the journalists who write for MLB.com freedom to criticize teams, players and coaches. How did that come about?: A:Clearly the fans understand the difference between promotion and news. We love our industry and we love our game, and we promote it. But we also try to write the news pieces down the middle. At times, clubs are mad at us because we say something they’d rather we had not, but if we don’t keep coverage balanced, our fans will not come back. If the manager or a player makes a bonehead play, we say so.
Q:What’s the most popular part of the site?: A:"Gameday," which is our live graphical game depiction. It's the penultimate live game application. And it will be even better next year when we start showing the trajectory of the pitch. If you can’t be at the game, it’s going to be the next best thing.
Q:Do most people view Gameday on its own, or in conjunction with radio and television.: A:Most people watch it during the day, at work, when they cannot be near a TV. They probably have it as a minimized window, which they pop up every few minutes to see what's going on in the game. [He jokes:] We believe this helps productivity in America. We think it gives people who are working hard at their desks a chance to relax and now and then enjoy baseball.
Q:The National Football League also has a website. How are you different?: A:Football is appointment viewing. Football grew up as a TV sport and that’s how most people watch it. It's shown once a week for three hours, and most Internet traffic for football relates to spreads and games and fantasy. It doesn't relate to video, audio or information because you are watching the game live.; Baseball is very different. Even the biggest fan can't go to every game or watch every game on TV. Baseball is three hours every night, and people just don't have that time every night. The theory behind our Internet site is in relating to that fan?who loves the game, wants to stay in touch with the game, but doesn’t always have the time. If you have just three minutes, we're going to get you what you need to know. If you have more time, we will give you more depth.
Q:What”s the design strategy to accomplish that?: A:Our philosophy is that baseball fans love the game of baseball. Some days they have a lot of time, some days not much. Our site should be able to track to whichever mood they are in that day. If your time is short, we'll tell you who is pitching, who is playing, and who is hot. We won't make you hunt.
By the same token, if you're a stat nut or a serious fantasy player, our site has some 275,000 pages. Our database is larger than other league's database, and it's dynamic?you can manipulate it. After a game has been finalized and our stringer's score matched up with the official score, it gets entered into our database. At that point, it has to be inviolate?we cannot give Babe Ruth a 61st home run.