The Colorado Rockies have gone to a four-man rotation; details on how and why they did this can be found in this Baseball Nation StoryStream™. To sum up, they demoted Jeremy Guthrie and didn't replace him, and with Colorado's season in free-fall (they've lost 12 of their last 13 and now have the third-worst record in baseball), manager Jim Tracy decided to experiment.
It remains to be seen how this experiment will work, but this raises an interesting question. Teams used four-man starting rotations for decades during the 20th Century; this began to change only in the 1970s and 1980s. Or, they'd have a rotation but would save their better pitchers to face the better teams, or they'd skip starters against teams they didn't fare well against. For example, Hall of Famer Warren Spahn had far fewer career starts against the Dodgers, a team that beat him regularly (24-37 record in 77 games, and that in an era where starting pitcher wins actually meant something), than against the Cubs, against whom he went 48-19 in 87 appearances.
This practice fell out of favor over the years, and now managers are almost slavish adherents to a system where each of five starting pitchers gets (approximately) 33 starts. Tthe 2011 Rangers are perhaps the best recent example of this: Last year, the Rangers had 157 of their 162 starts made by five pitchers: C.J. Wilson (34), Colby Lewis (32), Derek Holland (32), Matt Harrison (31) and Alexi Ogando (29).
What that does is occasionally give these starters more than four days' rest. For example, Wilson made 23 starts on four days' rest, eight on five days' rest, and one each on six and seven days' rest.
What if you arranged a starting pitcher schedule giving your top four starters the ball every fifth day on the calendar -- regardless of off days -- and spotted in a fifth "swingman" to take the ball when the top four would have to go on fewer than four days' rest?
That would mean your top four starters could start 35 times each -- a total of 140 games -- and you'd need this swingman for 22 starts, scattered through the season. When Mr. Swingman wasn't starting, he could pitch in long relief.
Teams used to do this quite often during the 1970s and 1980s as they transitioned from four-man to five-man rotations. Some pitchers who did this for a number of years, quite successfully and for multiple teams (started about 60 percent of the time, but also had double-digit relief appearances) were Bob Shirley, Danny Darwin and Larry McWilliams.
Why couldn't a team do this in 2012? You wouldn't be overtaxing your top four starters; theoretically, they'd only be making one or two more starts than they are now. The issue would be trying to identify pitchers who could take on a role that really hasn't existed in the major leagues for the last 20 years; from 1970 to 1990, there were 60 pitcher-seasons as described above, but from 1991 to the present day, just 38 such seasons. Two of them were by the rubber-armed Terry Mulholland, and there have been just seven such pitcher-seasons in the last 10 years.
By doing this, you could conceivably reduce your bullpen from seven full-time guys to six and a swingman; the 11-man pitching staff would allow teams to have an extra bench player for versatility. Jim Tracy isn't specifically doing this -- four starters and an occasional fifth man -- instead, he's trying an outright four-man rotation.
But there's no reason every team couldn't try this -- instead of the current managerial trends, which amount to the reflexive yanking of pitchers after seven innings and an arbitrary pitch count, and the incessant lefty-righty switches. Perhaps Jim Tracy, if he used the four-plus rotation theory described above, could start a welcome trend to creative use of pitchers.