This research started as a result of an offhand tweet that I made last week.   Some announcer advocated that, as a result of the balloting issue with the Kansas City Royals, the fans should not be selecting the All Star teams; it should go back to the players.   This is an idea that I very intensely dislike, and, acting out of annoyance rather than reason, I tweeted that this was the worst idea I had heard all year, and "doesn’t anybody remember how horrible they were at that?"

            In response to my tweet other people tweeted that the players were terrible at that, but then so were the fans, and so were the people who voted for the Hall of Fame, the MVP Awards, etc.; they all did a terrible job of it.   Well. . .not really; some of those systems are messed up, but some of them are pretty good.   Some of them usually get it right; some of them get it wrong pretty often.   Condemning all of them with a broad brush doesn’t really help. 

            I realized that, by speaking on impulse rather than speaking thoughtfully, I had contributed to the general miasma that colors the discussion:  All of these people suck, it’s hopeless, and there is no good way to do it.   We pick at one another, we find fault with one another, and it creates an impression that none of us know what we’re doing, a perception of chaos and hopelessness where reason and understanding can create progress.  

            OK, let us find the question that can contribute to understanding.    Were the All-Star teams chosen by the players (1958-1969) actually worse than the All-Star teams chosen by the fans?   Were they better?   Were they actually different?    How do we study that?

            My first effort to study this involved looking at the entire All Star roster, rather than the starting lineups; it seemed to me, intuitively, that this was more about choosing All Star teams than about choosing starting lineups.   I spent two days studying All Star rosters, and concluded that I was burrowing my way down a blind alley.    Why?   It’s hard to explain; it has to do with roster size, and expansion.    I was comparing the "value" of the players chosen for the All Star team to the value of the best players NOT chosen, but this changes enormously with the size of the All Star roster, and changes very significantly when there is an expansion.     Since the All Star rosters started at 18 players per league and are now 34 and have been changed many, many times, there is massive instability in the data, having nothing to do with the players who are selected, but with the selection procedure.   This made the whole-roster approach unworkable; I’m not sure that makes sense, but it would take me two pages to explain it any better than that.     The standards from the 1950s don’t work at all in the 1960s. 

            I then switched to studying All Star starting lineups.   There are also many, many problems with studying the issue in THAT way, and we’ll deal with those as we go; it is far from a perfect approach, but the alternative is—or seems to me—totally unworkable.  

            Let me jump ahead to a sort-of conclusion, and then I’ll explain how I got there; it will make more sense once you understand where we’re heading.  

            One of the worst-selected All Star teams of all time was the National League team in 1985.    Gary Carter in 1985 hit .281 with 32 homers, 100 RBI, and threw out 34% of opposing base stealers, which was better than league at that time—but Carter did not start the All Star game; the position went to Terry Kennedy, who hit .261 with 10 homers.    According to Baseball Reference Carter had 6.9 WAR that year; Kennedy had 2.6.   Keith Hernandez, also of the Mets, hit .309 that year with 91 RBI and brilliant defense at first base, 5.0 WAR, but Hernandez didn’t start, either; the start went to 36-year-old Steve Garvey, also of San Diego, who had 1.7 WAR.    Mike Schmidt hit .277 that year with 33 homers, 93 RBI at third base, 6.0 WAR, but Schmidt wasn’t part of the team, either; the third baseman was 39-year-old Graig Nettles, who hit .228 and registered at 3.3 WAR.    The three best outfielders in the National League in 1985 were Tim Raines (7.6 WAR), Willie McGee (8.1) and Pedro Guerrero (7.9), but none of the three started the All Star game; instead the starters were Tony Gwynn (5.2), Dale Murphy (7.2) and Darryl Strawberry (4.8).  

            Capping it all off and putting a cherry on top, the starting pitcher for the National League in 1985 was not Dwight Gooden, who was 24-4 that season with a 1.53 ERA, but Lamarr Hoyt, who struck out 83 batters in 210 innings, going 16-8 with a 3.47 ERA.   I should stress that I am not (at this point) blaming the voters for this or blaming the players who opted out due to injury or anything like that; I am simply making what seems like an indisputable point:   when you have Doc Gooden in your league, having the greatest pitchers’ season of the 1980s, but you don’t start Doc Gooden, you start Lamarr Hoyt, you are not putting your best foot forward.    I am not saying that it was anyone’s fault, but the National League All Star team in 1985 was exceptionally badly selected.   They got the wrong man, so to speak, at 7 out of 9 positions, getting it right only at second base (Tommie Herr) and shortstop (Ozzie Smith).   

            Despite putting one of the worst-selected All Star teams of all time on the field, the National League won the All Star game that year, 6 to 1.    The team the American League put on the field that year was well selected (Carlton Fisk, Eddie Murray, Lou Whitaker, George Brett, Cal Ripken, Jim Rice, Rickey Henderson, Dave Winfield and Jack Morris), and was a much better team than the National League team.   The American League had seven Hall of Famers in their lineup, all seven of them having good years; the National League had only two Hall of Famers.    The National League COULD have put a better team on the field than the American League, if they had sent Schmidt, Carter, Raines, Gooden, etc., but they didn’t.    But they won anyway.

            OK, if you read that through, you can see that I have made about fifteen assertions of fact there—that the National League team was poorly selected, that the American League team was a better team, that the National League COULD have sent a better team than the American League, etc.   How do we know that any of these things is true? 

            I studied the issue not by WAR but by Win Shares; let’s say it is six of one, half a dozen of the other.   I have the Win Shares data organized so that I can get to it for a study like this; I don’t have WAR stored in a spreadsheet in a way I can get to it.    Using Win Shares, I created four teams for each league that I studied:

A1, which is the actual starting lineup (in the All Star Game) for the American League,

A2, which is the best players in the league at each position who were NOT in the starting lineup for the American League,

N1, which is the actual lineup for the National League, and

N2, which is the best players in the National League who were not in the starting lineup.  

            In 1985, the Win Shares for the National League All Star Starters (P-Hoyt, C-Kennedy, 1B-Garvey, 2B-Herr, 3B-Nettles, SS-Ozzie, OF—Gwynn, Murphy and Strawberry) total up to 190.   The Win Shares for an All Star team of those NOT selected to start the game (P-Gooden, 1B-Hernandez, 2B-Sandberg, 3B-Schmidt, SS-Templeton, OF-Raines, McGee and Guerrero) totals up to 275.   The team NOT selected is 85 points better than the team that took the field.    (N1, 190; N2, 275)

            In the American League, the Win Shares for the starting All Star team total up to 230.   The best players NOT selected that year (P-Bret Saberhagen, C—Lance Parrish, 1B—Mattingly, 2B-Willie Randolph,  3B--Boggs, SS—Tony Fernandez,  OF—Harold Baines, Phil Bradley and Jesse Barfield) totals up to 225 Win Shares.    The team actually selected was better than the team left out, so it’s a well-selected All-Star team.   (A1, 230; A2, 225).  

            Usually, most leagues, the team of players NOT selected is better than the team of players who start the All Star Game, but we’ll get back to that point.    By now, most of you have noticed that we are using full-season stats to evaluate a team chosen in the middle of the season.    Does this cause a problem for the study?

            Of course it does, but what are you going to do?    It’s an All STAR team; it’s not an All-guys-who-have-had-an-early-season-hot-streak team.   In 1982 Toby Harrah, playing third base for an American League team, was hitting .332 at the All Start break, with a .548 slugging percentage.   George Brett was hitting .301, .493 slugging percentage.    That doesn’t make Toby Harrah a better player than George Brett; it just means that the stats haven’t evened out yet.  

            Going back to the issue, which is "who should choose the All Star teams?", what we are talking about is who is really the best player—not who has the best numbers so far, early in the season, but who is really the best player.   If you think about it, the season can be divided into "what has happened so far" and "what will happen the rest of the year."   When you are choosing the All Star team, the All Star GAME is not part of the "what has happened so far package", it is part of the "what will happen the rest of the year" package.

            You can argue that issue one way or the other, but I don’t really have any problem with it; I am comfortable evaluating the All Star team by the full season stats.   But there are many similar issues.   What if a player plays well the first half, then gets hurt just shortly after the All Star break, or in the All Star game like Dizzy Dean and Ray Fosse?   That’s not something that can be known; the All Star selection shows up as a bad selection based on things that could not have been known at the time.   Sure; no study is perfect.    What if one player is NAMED to the All Star team, but decides not to play and is replaced by another player?   

            Sure—but that’s not a problem if you take it one step at a time.   Step One is not "evaluating how well the voters did"; Step One is "evaluating the strength of the All Star team."   When you get past Step One, then you can worry about evaluating the various factors that CAUSED the team to put a good lineup on the field or not to put a good lineup on the field.    This is just Step One.  

            A similar problem, an old admonition.  .. ..statistical analysis is not a perfect guide to the truth, and we must not be arrogant in assuming that OUR answer is necessarily the RIGHT answer.  Sometimes maybe our system shows the Win Shares as Joe 22, Jack 16, but the voters or the manager who makes the selection knows something that the numbers don’t know.   

            In the early history of the All Star game, the rosters and starting lineups were entirely selected by the manager, without any input from fans OR players.   It was a one-man operation.  

            My analysis shows that many of those teams were very poorly selected.   Very often, the managers just chose three or four superstars, and otherwise chose players that they liked and ignored players that they didn’t like, seemingly with little regard for talent.   In 1936, for example, our stats show that the American League All Star team, chosen by Joe McCarthy because of a serious injury to Mickey Cochrane. . .the American League All-Star team was very well selected, while the National League All Star team, chosen by Charlie Grimm, was very poorly selected, and the American League team was a far better team.   It’s like 1985; the National League COULD have put a stronger team on the field than the American League could have, but they just didn’t do it; they chose the wrong players.   (A1, 229; A2, 211.  N1, 194; N2, 257).     Nonetheless, the National League won the game, 4-3, the National League’s first All Star victory.  

            That’s the way my system sees it, and that’s the way I see it—BUT.   We have to remember that we do not have perfect knowledge of the players involved.   Charlie Grimm was on field with the players he chose; he played against them, he talked to them, he spent time with them off the field.  He knew vastly more about these players, in the sense of having a truck load of information, than I could ever possibly know about them.   What I have is a different PERSPECTIVE on their accomplishments, that’s all.   Charlie thought about it one way; I think about it a different way—and he did win the game.  

            This is particularly an issue with regard to the early catchers.    In the first 15 years of the All Star game, the All Star managers almost always chose a different catcher than we would choose—and very often chose some catcher who appears, in my analysis, to be not remotely qualified.    In 1933 the American League catcher was Rick Ferrell (17 Win Shares) rather than Mickey Cochrane (26 Win Shares).   The National League catcher was Jimmie Wilson (8 Win Shares), rather than Gabby Hartnett (21).    In 1935 Gabby Hartnett had 26 Win Shares, Jimmie Wilson had 7—but Wilson started the All Star game again.    Do I believe that Jimmie Wilson might actually have been a better player than Gabby Hartnett?  No, of course not—but we can’t be arrogant in assuming that our view of the facts is the only way to look at it; it isn’t.  

            But our way of looking at the facts enables us to get answers to these questions.   Charlie Grimm can’t answer these questions for us; he’s dead.   Our way of looking at these questions creates answers.   Score one for sabermetrics. 

            OK, now to the issues we are trying to get to, for example:

            What was/were the strongest All Star team(s) ever?

            What was/were the weakest team(s) ever?

            What was the best-selected All Star team?

            What was the worst-selected team?

            Does the better team usually win?

            How do the fans do selecting the starting lineups, as opposed to the players?

            I will get to those questions later—I won’t give you completely satisfactory answers to any of them, but I will get to them—but first I wanted to discuss with you the four things I actually came to understand while doing this study, the four things I didn’t understand before but I do now.

            1)  The All Star rosters started small, and just gradually grew.     The Wikipedia page about the baseball All Star game contains this rather gross passage:

============================================================

Selection of players

Until 2009, each league's All-Star team consisted of 32 players. On July 1, 2009, the MLB added a 33rd player to each league's team roster.

On April 28, 2010, MLB announced several rules changes for future All-Star games, effective with the 2010 edi^[3]

• Rosters were expanded by one extra position player, to a total of 34.

======================================================================

            This is kind of starting the alphabet at "X".   In reality, the first All Star rosters (1933) were 18 players each.   In 1934 they changed it to 20. In 1935 to 21, etc; one year the American League had 22 players and the National League 21.  

            The All Star roster has expanded, and expanded, and expanded, generally one slot at a time, from 1933 to the present, and this practice continues to the present time.     I’m not denouncing this or decrying it or anything; in 1933 there were 8 teams in each league and 18 players on the All Star team, now there are 15 teams in each league and 34 players on the All Star team, so it’s about the same ratio. …it’s fine with me.    My point is that I never understood this, before doing this study.   And neither does Wikipedia, apparently.  

            2)  The story line about the National League domination of the All Star game as a result of the National League having black players and the American League not having them doesn’t actually hold up when you look at the timeline.  

            There are two things here

            a)  there is a general point that the National League got to be a stronger league than the American League because the American League was slow to integrate, which I believe to be true, and

            b)  there is a specific story that you hear sometimes now, that the National League dominated the American League in the All Star games because the black stars from the National League regarded the American League as a racist league, and took a particular pleasure in beating the crap out of them in the All Star games.   

            (a) is true, but (b) is not true and cannot be true.   You realize that it cannot be true if you look at the time line.   The National League broke the color line in 1947; the period when there were a good many black stars in the National League but the American League was dragging its feet was 1950-1959, more or less.   The American League was winning half the All Star games in that era; the American League held its own in All Star competition until 1962.     By the mid-1960s the "integration disparity" was melting away. 

            You hear a lot of stupid shit about race, frankly; people pour "race" over every issue like some people pour catsup or salt over whatever they’re eating.    Henry Aaron once explained the fact that he won only one Most Valuable Player Award by racism in the BBWAA voters—completely ignoring the fact that, in Aaron’s first nine years in the National League, the Most Valuable Player was black eight years in nine (Willie Mays, Roy Campanella, Don Newcombe, Aaron, Ernie Banks, Ernie Banks, Frank Robinson, Maury Wills.)   Billy Williams once said that when he came into the league, each team could have two black players, the star and his roommate.   It is complete and total nonsense; there never was any such pattern.  

            I never heard this story about the black players being motivated to beat up on that bad, racist American League until the 1990s, and it is just bullcrap, frankly.    It IS true that the National League became the better league, in terms of talent, because the Yankees, the Red Sox and several other American League teams were slow to catch up to the times in regard to black players, and it is probably true in a general way that the National League dominated the All Star games because it became a stronger league.    But the specific narrative. . ..it’s just a story.  

            3)  The "player rights" era, beginning in the mid-1970, profoundly changed the All Star game, in ways that were definitely not for the better, and in ways that I don’t think anyone understands.   I certainly never understood what had happened, until I did this study.

            Founded in 19^th century labor practices, baseball players until the 1970s were basically expected to put up with whatever they had to put up with.   The manager is a gigantic jackass who says insulting things about you in the newspaper?   Get used to it; that’s just the way it is in baseball.   You are suddenly re-assigned to play for some other team in some city that you don’t like, taking your kids away from all of their friends and your wife away from the home she has spent four years re-decorating?   Get used to it; that’s just what happens in baseball.   

            Until about 1980, it was common for a baseball player to be named to the All Star game, but to not actually appear in the game.    There was no general idea that every player had a "right" to get into the game; there was no idea that if a player made the effort to go to Cincinnati, he should get into the game.   If you were named to the team but you didn’t play, that was just the way it was; get used to it. 

            In 1974 there were 14 players who were named to the All Star team and who were there at the game in uniform, but who did not play.   In 1978 there were 11 such players.   In 1983 there were 8 such players.  By 1987 there were only 2. 

            Although 1987 was atypical (there were usually 5 or 6 such players, mostly pitchers).. ..although 1987 was atypical, somewhere in that time period, 1974-1987, the idea took root that a player who was named to the All Star TEAM should actually get to play in the All Star GAME.   The effect that this had on the All Star game, frankly, was absolutely horrible.   The game, rather than being any kind of an effort to win, became just a show, an exhibition, players strolling up to the plate to get their face on National television.    John Kruk, a left-handed batter, batted right-handed to dramatize his fear of Randy Johnson, and it was supposed to be a big joke, nyuk, nyuk nyuk.   They were making a farce of the game.  That understates it; they HAD made a farce of the game.   They had taken this great tradition, beginning from a desire to have the greatest players in the world all gathered together, and reduced it to a stupid, interminable parade of pinch hitters making no apparent effort to win the game.   

            Look, I hope nobody thinks I am against the players having rights; hell, Marvin Miller was a friend of mine, and I’ve got witnesses to the fact.    I’m all in favor of players having rights.

            But you have to balance things.   You have to be careful.   Actors who are making a movie have rights, too—but in making the movie, you don’t do what makes the actors happy; you do what makes a good movie.   You have to.   Letting each player get one at bat in an All Star game is exactly like making a movie and making sure that every actor has at least 10 lines.   An actor shows up to make a movie, he has a right to have some lines to speak, don’t you think?   Makes sense to me.  It’s better for the actors that way, but it would make a dreadful movie.    The 1980s, 1990s All Star games were just dreadful.  

            There was a second effect of the Player Rights era on the All Star game.    When players began to get bonus clauses added to their contracts, you could not—and still cannot—write a bonus clause saying that the player gets extra money if he hits .300, or that he gets extra money if he drives in 100 runs, or he gets extra money if he hits 30 home runs; nothing like that.     Those clauses are prohibited, have long been prohibited, and should be prohibited because they create distorted incentives on the field.  

            However, you COULD write a provision giving the player a bonus if he made the All Star team; initially you would see $5,000 All Star bonuses, and then $10,000. . .eventually players would make a $100,000 bonus or more if they made the All Star team.  

            Well, by 1990 a very large number of players had contracts that had an all star bonus clause in them.   Almost EVERY player had All Star bonuses of some kind or another, except for the young (pre-arbitration) players on the teams that didn’t like the practice. 

            The thing was, you didn’t really have to GO to the All Star game to collect the bonus; you merely had to be named to the team.    So the point wasn’t to be IN the All Star game anymore; it was merely to be named to the team.    Once you had been named to the team, you had earned your bonus, and you were done.  

            So, by 1990 and accelerating after 1990, more and more players were named to the All Star team but were "injured" and replaced due to injury.   The union—and, more importantly, the player’s community—fully and heartily approved of this tradition.    The more players who were named to the team but were replaced due to injury, the more bonuses were paid.   Everybody wins.    The veteran star who gave up his position on the All Star team to enable some young player to have the honor of being on the team was not only not criticized for cutting out on the fans, he was applauded for it.   He was being generous, allowing the younger player to have his moment.   But as to actually playing in the game. . .well, who needs that?   It’s a hassle, you have travel Monday, put up with 10,000 reporters asking mindless questions to justify their All Star travel budget, hang around to get your at bat, fly back on Wednesday—and it’s just a dumb exhibition game anyway.   Who needs it?   I need the break.   I’d rather spend the time with my family.

            Of course, this further weakened the quality of the All Star games—along with the parade of pinch hitters effect, which was going on at the same time.   By the late 1990s, the All Star game had become 35% less fun than spending some quality time with your dentist.

            This culminated, of course, with the disaster in Milwaukee, when the game was called off after 9 innings because the parade of pinch hitters had run through all of the available players.  The crowd booed angrily, and people were shocked, shocked  to discover that the fans actually wanted to see a baseball game, and not a stupid exhibition in which the main purpose was for as many young players as possible to be able to claim they had once played in an All Star game.   The commissioner stepped in and fixed that, and God bless Bud Selig for doing that, and it’s a lot better now.    But until researching this article, I had never understood how the degeneration of the All Star game in the 1980s and the 1990s had resulted in part from the player rights movement of the 1970s.  

4)   The Cincinnati/KC vote stuffing phenomenon.

            One little thing that I didn’t know, and most of you probably don’t know, either, was that until 1946 the manager for the league picked the players that he wanted with no input from anybody else, or with only such input as he chose to accept.    From 1947 to 1957 the fans voted on the All Star teams, but in 1957 a Cincinnati newspaper printed "ballots" in the paper that fans could just cut out and mail in, and the team as elected by the fans was seven Cincinnati Reds players and Stan Musial.  

            Baseball’s commissioner at that time was the regrettable Ford Frick.    Somebody should write a biography of this man, called "The Man Who Almost Killed Baseball".   Mishandling this crisis as he mishandled every other challenge of his tenure, Ford Frick took the vote away from the fans, because of what had happened in Cincinnati, and gave the vote to the players.  

            Look, I am all in favor of the players having rights, but. . .we don’t play the game for the benefit of the players, OK?   Ford Frick should have been a school principal.    That’s how almost every school principal would have handled this situation, isn’t it?   No, not every school principal; half of school principals are sensible people who have some idea what they are doing.   The other half are nitwits like Ford Frick whose first instinct is to punish the whole class because somebody misbehaves.   That’s the way that people in power think, half of them.  

            Well, of course you can’t have an All Star team with seven Cincinnati Reds and Stan Musial, and, modern version, you can’t have an All Star team with seven Kansas City Royals and Mike Trout.   Of course you have to do something to prevent that from happening.  

            As to what you can do, let me propose a rule, actually two rules:

            1)  An All Star team may have no more than five players from any one team, including those who are named to the team but replaced due to injury, and

            2)  No more than two players from any one team may be in the starting lineup.   

            If a city goes overboard and "wins’ the vote at seven positions, only the top two count.   That makes sense, doesn’t it?   We now have fifteen teams in each league.    Two starters from one team is enough, I think.   I don’t believe there is any team in baseball now or any time in the last ten years which has had three players who ought to be starting the All Star game.   

            There is a simple solution to the problem; there is that one, and then, if you don’t like that one, there are 87 other very simple ways to solve the problem, rather than fulminating about it and talking recklessly about turning the vote over to the players so they can make it a popularity contest among the players.

            What happened in Kansas City this year is more analogous to what happened in San Diego in 1985 than to Cincinnati in 1957.   The Padres were a down-and-out organization for years.   In 1984 they had a good year, made the World Series, and the city got really excited about baseball.  That’s a good thing.    Only, in their excitement, they got kind of carried away, and they elected entirely too many San Diego Padres to the 1985 All Star game—exactly the same thing that happened in Kansas City this year.  The commissioner, in commenting on the Royals’ ballot-box "stuffing" phenomenon, should begin by saying that this is a good thing, when a city gets really excited about their team, and we don’t in any way want to condemn that or denounce it. 

            Anyway, this is what I realized when I was doing this study, that I had never understood before.    Cities like Cincinnati, San Diego and Kansas City are not the predators in this story; they’re actually the prey.   

            If you look at undeserving players who get elected to the Hall of Fame, what you see is that it is mostly a big-city phenomenon.   Players from New York, LA and Chicago being elected to the All Star game when they really shouldn’t be there. . .heck, that happens all the time.   Those teams, over a period of years, have WAY more undeserving All Stars than Cincinnati, San Diego or Kansas City. 

            The only thing is, when it happens in Cincinnati, San Diego or Kansas City, they all get elected in the same year.   It happens in big cities simply as a result of the ordinary operation of events; it happens in a small city only when the small city gets all fired up and MAKES it happen.    Which, again, it shouldn’t happen, but it shouldn’t happen on either side.  

            OK, let me get now to the type of questions that I was trying to study.   

            I had hoped to do a comprehensive study of All Star lineups, all 172 lineups from 86 games.   I had also hoped to publish it last Monday, the first day of the All Star break.   Unfortunately, it takes TIME to do these studies, so

            a)  I am three days late, and

            b)  I only completed a little over half of the teams.  

            So this is a selective study or a "sampling" study, rather than a comprehensive study.  

            Of the 90 All Star teams that I studied (45 seasons, two teams each season. . .actually 4 teams each season included in the study). . ..but of the 90 All Star teams that I studied, the best teams were the National League teams in 1951, 1953, 1959, 1965, 1969 and 1996, and the American League teams in 1946, 1969, and 2005.  

            The 1946 American League team, chosen by Steven O’Neill of the defending champion Detroit Tigers, had at its core the four Red Sox from the iconic photograph—Ted Williams, Bobby Doerr, Johnny Pesky and Dominic DiMaggio.    They all had great years in 1946, and there is a statue of the four of them out behind center field at Fenway Park.   

            Joe DiMaggio had a poor year in 1946, and was not named to start the 1946 All Star game.    Three of the four Red Sox teammates were in the top four in the 1946 MVP voting; Dominic was 9^th, and the other three were 1-3-4.    Bob Feller struck out 348 batters that year and won 26 games for a bad team; he was the starting pitcher.   Mickey Vernon hit .353 and won the batting title; he was the starting first baseman.    Charlie Keller was the third outfielder; he had his usual 30 homers, 100 walks, 100 RBI.   The team walloped the National League, 12-0, in a game most remembered for Ted Williams hitting a home run off of Rip Sewell’s blooper pitch. 

            The 1951 National League team was P—Robin Roberts; C—Roy Campanella, 1B—Gil Hodges, 2B—Jackie Robinson, 3B—Bob Elliott, SS-Alvin Dark, OF—Del Ennis, Richie Ashburn and Stan Musial    They beat the American league, 8-3.  

            The 1953 team carried over only three starters (Musial, Campanella and Robin Roberts), but again was very strong, with Hall of Famers Red Schoendienst, Pee Wee Reese, Eddie Mathews and Enos Slaughter added along with Ted Kluszewski and Gus Bell, who were great players but didn’t stay great long enough to make the Hall of Fame.   That team beat the American League 5-1.  

            The 1958 and 1959 NL All Star teams stand out as exceptionally Well Selected teams, but the thing is, when you have Willie Mays and Henry Aaron in your league, it is pretty obvious who should be on the All Star team.   The 1958 team was very well selected, but not unusually good.  The 1959 team was both.   From 1959 to 1962 they played two All Star games a year, in another effort to kill the All Star Game by doing whatever was best for the players.   The 1959 (Game One) team was P—Drysdale, C—Del Crandall, 1B—Orlando Cepeda, 2B—Johnny Temple, 3B—Eddie Mathews, SS—Ernie Banks, OF—Wally Moon, Willie Mays and Henry Aaron; six Hall of Famers.   They beat the American League, 5-4. 

            That team ranks as the best-selected All Star team in my data, the games I have done, and it ranks that way because they chose the best available player at five positions.   The four best players in the National League—Aaron, Mays, Banks and Mathews—were all selected to start the All Star Game.   Del Crandall was in fact the best catcher in the league.    My system disagrees with the selections at the other four positions, but they are all close calls, just a point or two off in each case.  The 1959 NL (Game 2) team had two substitutions to the starting lineup, which weakened the team.

            There has never been a perfectly-selected All Star team, at least that I found, and MOST teams have at least one or two real klunkers, choices that just seem inexplicable in retrospect, or, if you prefer, in retrosheet.  

            The 1965 All Star game—although the National League merely edged out a 6-5 victory—is the greatest mismatch in terms of talent in All Star history.   The National League featured Juan Marichal, Joe Torre, Banks, Pete Rose, Dick Allen, Maury Wills, and an outfield of Stargell, Mays and Aaron.   The American League a) just did not have anything remotely approaching the same level of talent, and b) did a very poor of selecting from the players they had.  Starting pitchers, Juan Marichal and Milt Pappas.   Second base, Pete Rose and Felix Mantilla.   Center Field, Willie Mays and Vic Davalillo.

            That was a poorly-selected All Star team, not the worst ever, but one of the worst.   The worst NL All Star team ever, and the most poorly selected, was the Padre-laden outfit that I described earlier, the 1985 team.   But the worst selected All Star team of all time, and also the worst All Star team of all time, was the 1943 American League team.   

            Your first thought now is "war time", but that’s not EXACTLY it.   The Win Shares system awards as many Win Shares in a War Time Season as in any other season; it just figures that a win is a win is a win.    By that standard, there were as many good players in the American League in 1943 as in any other year; they just didn’t play in the All Star team. 

            The best players in the American League in 1943 were, in order, Luke Appling, Charlie Keller, Lou Boudreau, Spud Chandler, Stan Spence, Joe Gordon, Roy Cullenbine and Rudy York.    Unfortunately, none of those men—none of them—started the All Star game.   George Case, Chet Laabs and Dick Wakefield started in the outfield in place of Keller, Spence and Cullenbine.   Dutch Leonard (11-13 with a 3.27 ERA) started on the mound in place of the eventual MVP, Spud Chandler (20-4 with a 1.64 ERA). 

            When you get past those things, the team STILL was not well-selected.   Jake Early, a .258 hitter, started at catcher in place of Bill Dickey, who hit .351; Dickey was old and a half-time player, but he was still clearly more valuable than Jake Early.   The best third baseman in the league was a rookie, Billy Johnson of the Yankees, but he didn’t start, either; the start went to Ken Keltner, who struggled through an injury-plagued season with a .260 average and 4 home runs.

            You may have picked up on the key factor there:  The Yankees didn’t start.   The Yankees, although they were missing DiMaggio and Ruffing and some other guys, were still the best team in the league, and had the legitimate All Stars at several positions—pitcher Spud Chandler, catcher Bill Dickey, second baseman Joe Gordon, third baseman Billy Johnson, and outfielder Charlie Keller—but none of them started, and no Yankee started.   No Yankee even made an appearance in the game.  

            But here’s the thing:  that team was selected by Yankees manager Joe McCarthy!   Huh?   What happened here?  

            Probably somebody somewhere knows, but it’s a mystery to me.    I thought of war-time travel restrictions, but the game was played in Philadelphia.   I think the New York players could probably get over the Philadelphia.   I thought that maybe McCarthy was trying to give his key guys a little midsummer break, but several Yankees were on the team; they just didn’t play.   Was McCarthy angry at his team for some reason?   Was he trying to send a message to his team?   

            Or had McCarthy been criticized around the league for carrying too many Yankees in other years, and was he trying to accommodate the league by playing players from other teams?    I just don’t know.   But Bobby Doerr hit a three-run homer in the second inning, and the American League won the game, 5-3, despite putting a historically weak team on the field.    

            The 1948 American League team, selected by the fans, is another curious one—and they won, too, 5-2.   The 1948 AL team had Ted Williams and Joe DiMaggio—on the bench.   Pat Mullin and Hoot Evers started the game in the outfield, Williams and DiMaggio on the bench.  The starting pitcher was Walt Masterson (8-15, 3.48 ERA); Bob Feller, Bob Lemon and Hal Newhouser—all Hall of Famers, and all 20-game winners that season—were on the bench.    But, you know, they won the game, because you just can’t beat guys like Pat Mullin, Hoot Evers and Walt Masterson.   The manager was nice enough to let Joe DiMaggio and Ted Williams pinch hit, gave them one at bat apiece, out of kindness I suppose.  

            Now, does the team that has the better All Star team usually win the game?   I asked three questions:

            1)  Did the team which was better on paper actually win the game?

            2)  Did the better SELECTED team win the game? And

            3)  Did the league which had more high-quality talent to choose from win the game? 

            All three questions got answers over 50%.   The first question—did the better team win—the better team on paper won the game 52% of the time, 23.5 wins in 45 tries (one game in the study ended in a tie.)  

            The better SELECTED team won the game 57% of the time.

            But the league which had more front-line talent to choose from won the game 67% of the time.  

            Which makes sense when you think about it.   They key isn’t who starts, because the starters don’t usually play the whole game, anyway.    I added together the "chosen" team and the "not chosen" team to represent the quality of the stars in the league.    What matters is the combined quality of the starters and the non-starters. 

            Look at 1955.   The American League in 1955 did a much better job of choosing the right guys to START the game than the National League did.   Willie Mays, Henry Aaron, Stan Musial and Roy Campanella were in the National League in 1955—but they didn’t start the All Star game.   All four of them SHOULD have started, but they didn’t; the starts at those positions went to Del Crandall, Kluszewski, Duke Snider and Don Mueller.   

            BUT those players were still there.   Del Ennis got one at bat; Stan Musial came off the bench and got four at bats, and hit a home run.    Duke Snider got a couple of at bats, Willie Mays came off the bench, went 2-for-3 and scored 2 runs.   Don Mueller got a couple of at bats; Hank Aaron came off the bench, went 2-for-2 and drove in a run.  

            The starting lineup matters, but what REALLY matters in an All Star game is the depth of the talent.   The fact that Musial, Aaron and Mays didn’t start was less important than the fact that they were on the team.   

            Finally, the question "Who should pick the All Star starters?"   The FANS should pick the All Star starters, and frankly the people who want to give the vote to the players should all be dangled out of 18^th-story windows by their toes until they start to get their minds right, just my opinion.   But as to this STUDY. . .there is no actual evidence, in the study, that one method is better than the other, other than it is clear that the "early" method, in which the managers picked the teams, did not work well.   

            Look, I’m just joking about the 18^th-story windows; a 17^th story window will do just fine, and you’re entitled to your own stupid opinion.   But one of things that REALLY irritated me about the 1980s All Star games was in 1987, when John McNamara, paying off an old personal debt of some kind, picked Jay Howell to the American League All Star team.    Jay Howell had a 5.89 ERA in 1987, and he had been blown up in two of his last three appearances before the All Star break. 

            McNamara obviously never intended to actually USE Howell in the All Star game, but the game stretched into extra innings and Howell had to pitch.   And he lost the game for the American League.   I was rooting for the American League.  I was pissed.  What in the hell was Jay Howell doing on the All Star roster, anyway?   Why should John McNamara have the right to ignore the facts and pick whoever the hell he wants, when there were much better relievers available to him?

            And McNamara’s attitude was, Oh, well, it’s just an exhibition game anyway, what difference does it make?   Jesus Christ, Man, if you don’t care who wins, STAY HOME.   Let somebody manage who DOES care. 

            It irritates me when people complain about the FANS picking the All Star team, when in reality, the fans only pick one-fourth of the All Star team; three-fourths of the team is actually picked by the managers and coaches.   And (b) the fans, in all of their years of voting including 1957 and 2015, have never done anything as egregiously arbitrary as putting Jay Howell on the All Star team with a 5.89 ERA.  

            In the early years the managers chose the All Star teams, and their performance at optimizing the starting lineup was clearly not very good.    On average, an All Star starting lineup in those years was 12 Win Shares weaker than a team made up of the best players at each position who were NOT chosen to start.

            Then the fans chose the All Star teams from 1947 to 1957, and they were distinctly better.   An average starting lineup in that era (in my data, which includes 14 of the 22 All Star teams of those years). . . an average starting lineup was one to two Win Shares BETTER than the best players who were not chosen.  

            Then the players took over the vote, from 1958 to 1969.   In that era, the starting lineups in the All Star games were, on average, 5 Win Shares WORSE than a team of the best players who were not chosen. 

            Then the fans took the starting lineups back in 1970, and, since 1970 the average starting lineup has been 14 Win Shares WORSE than a team of the best players who were not chosen. 

            So the fans, since 1970, have not done great at picking the best players (in comparison to earlier years) but it is not a level playing field, either, because of expansion.   When there are more players in the league, it is mathematically more difficult to identify the very best player at each position.  

            Putting that issue aside, the differences in my study are too small to be statistically significant, and, putting that issue aside, there are all kinds of methodological issues or methodological flaws in this study.    I was trying to figure out how to study the issue, and this is what I came up with, and I understand the history of the All Star game better now than I did before I did the study.    Thanks for reading, and I hope you got something out of the study, as well.