"This Time it Counts."
"I play to win."
And while the players may play to win, the managers don't manage to win. The games are managed like a Little League game, where everyone gets at least two innings in the field or one plate appearance, and everyone gets a medal for participating.
There's nothing wrong with an exhibition game - but if MLB is going to insist that the All-Star Game counts, then the managers need to be nudged to manage it that way.
First, the wrong guy is managing the All-Star team. The San Francisco Giants have about a 1% chance of reaching the World Series this year; whether or not the National League gets home-field advantage in the World Series doesn't matter all that much to Bruce Bochy. It would matter significantly more to Mike Matheny, Fredi Gonzalez, or, dare I say it, Clint Hurdle.
So, first change: Instead of the All-Star Game being managed by the manager of the previous season's World Series team, the game should be managed by the manager of the team with the best record in each league.*
Second, as Bill Parcells said, "If they want you to cook the dinner, they oughta let you shop for the groceries." The manager of each league's All-Star team (or, to be more accurate, the management organization to which the manager of each league's All-Star team belongs) should get to choose the roster that he will use to try to win the game for his league. There's still a place for fan voting, but instead of fan voting deciding the 8 position players (and AL DH) that start the game, fan voting should be used to select five "fan favorites" that are placed on the All-Star Game roster. The manager gets to decide if they start, or even if they play.
Third, the manager has too many toys. If you have seven outfielders, you'll have a tendency to use seven outfielders. If you have a roster constructed more like a regular-season roster, you'll have a tendency to use it like a regular-season roster.
Putting these two pieces together leads to the following process to select the rosters for the ASG:
1. Fan voting. Each team nominates up to 8 players - including both position players and pitchers - to be placed on the All-Star Game fan ballot**. Fans get to vote for five players in one league*** with the top five vote-getters being placed on the roster.
2. The total roster size for each league's All-Star team is 24 players and 8 "non-playing representatives". The 24 players consist of 9 pitchers and 15 position players, to give each manager a roster makeup closely approximating the roster makeup they'd have in a regular game, adjusted to take into consideration that the roster is being constructed for only one game, not a 162 game season.
The 9 pitchers are three starters, all of whom need to be able to pitch the day of the All-Star Game on regular rest or longer, and 6 relievers. Starting pitchers who are not able to pitch on regular rest, whether selected in the fan voting or selected by the manager, can be named as "non-playing representatives" - they are still considered All-Stars, but don't get to play in the game.
The team gets three starting pitchers: one will start the game, and two will be in the bullpen, with one serving as a long reliever and one serving as an "emergency reliever" (to be used if the game goes deep into extra innings). The roster has only six relievers, even though most teams carry seven, because one of the starters is serving as the long reliever that is in most bullpens.
The 15 position players are 9 starters (including a DH in each league) and 6 backups. This is two more position players than on most normal rosters, but it allows the manager to carry a third catcher and a "wild-card" type player - a Billy Hamilton type who can be used as a late-inning PR, a masher to use strictly as a PH, a defensive specialist, etc. So the "six-man bench" will typically be two catchers, a fourth outfielder, a middle infielder, a corner player of some sort (1B/3B/corner OF), and whatever 6th bench player the manager wants.
The manager should include the five "fan favorites" on the active roster, but doesn't have to. If any of them are pitchers who won't be available on normal rest, they get put on the "non-playing representative" list. The manager can also choose to put any of the other "fan favorites" on the non-playing rep list****.
Limiting the roster size to something approximating a normal roster will tend to curb the tendency of the manager to try to get everyone in the game, and instead use "situational" replacements. You won't swap in your backup middle infielder in the 5th inning, because you might want to pinch hit for one of your MIs in the seventh. You won't replace your DH in the 4th, because you want to save your strong hitters for a possible high-leverage PH situation later in the game. You won't pull your starting pitcher after one time through the lineup because you don't have 13 pitchers to use.
Alternately, MLB could scrap the whole "WS home-field advantage to the ASG winner" crap and turn it back into a plain old exhibition game, in which case none of this would matter. But then, what would sports writers and bloggers have to write about for a week in the middle of each baseball season?
* Details of timing can be worked out - it could be the best record through 81 games, the best record as of one week before the ASG, or whatever else works. If there are two teams with the same best record, the tiebreaker can be the end of season record in the previous season, with the tie going to the team with the worse record in the previous season (so the tiebreaker is based on the amount of improvement).
** This is an "All-Star" game. Few teams since the '27 Yankees have more than 8 "Stars".
*** If you're an NL fan, you shouldn't get to pick who you're playing against.
**** For example, if the fan favorites include three 1B/DH types who can play nowhere else, the manager would be justified to have only two on the active roster, and put a player who can play 1B/3B or a corner outfield spot on his bench while the third 1B/DH type is on the non-playing rep list.