I opened my copy of The Hardball Times Season Preview yesterday. It's written in a similar format to the BP annual, except that most team chapters are written by a blogger for that team.
The book includes a set of purely computer-based projected standings, and also asks each author to peg his team's "Most Likely" outcome. The bloggers are smart guys, like Larry Borowsky and Chone Smith, but as a group they're still too optimistic about their favorite teams.
Twelve of the authors gave either an exact expected record (like 84-78) or a range, such as 67-70 wins. Comparing these predictions to the computer projections for the same teams, the average author predicted his team would win 2.6 games more than the computer forecast. (For the ranges, I used the midpoint of the range. For example, if the author said 67-70 wins, I used 68.5 for the calculation.)
2.6 wins may not sound like a lot, but it's actually a huge error. If you project the average team to win 83.6 games instead of 81, then the 30 teams would combine for 156 more wins than losses. You'd need to play and lose almost a full season's worth of games to make up that difference.
I think it's great that baseball fans can find reasons to believe their favorite teams will do great things in the upcoming season. However, a good analyst should be able to view his team objectively. When he doesn't, he ends up looking like an idiot in a major national publication. Without that 2.6 game bias, perhaps Rick Morrissey and his ilk would see that the White Sox aren't contenders as presently constructed.