I'm not really criticizing it; just offering the general willingness of teams to work relievers in ways they'd never work starters as one of the things that contributes to the perceived inconsistency of relief pitching relative to starting pitching. It would be something a team with a long-term investment in a reliever could control in order to limit the injury risk on the back end of the contract. Unless I'm missing some, almost all of the small number of recent four-year contracts to relievers actually have worked out fairly well (Downs, Affelt, Benoit, Papelbon), which suggests that there might be something to that.
We remember the ones that work out, but do we remember guys like BJ Ryan?
Either teams have figured something out or it's random sampling. Look at this fangraphs article from a few years ago:
More recently, Heath Bell got 3/27 and was a complete bust. Brandon League got 3/22.5. Jason Motte got 2/15. Joe Nathan got 2/20. Grant Balfour got 2/12. Etc.
Relief pitchers are volatile.