DIPS is essentially the notion that pitchers are in control of their strikeout rate, walk rate, fly ball rate, and not much else. So things like clusters of hits and a disproportionate amount of fly balls leaving the yard can be attributed to variance (which shows up in stats like BABIP, HR/FB). Not involved in DIPS but whiff% is also a good indicator of pitcher skill as it isn't affected by framing, umps, etc.
In application, DIPS are stats like xFIP, FIP, SIERA, etc. Defense Independent Pitching Statistics. We've had some very good ones created, notably by an old poster named JFaS (xxFIP, TIPS), and by a current poster, Northof49 (FRIAS). They are usually better predictors of next-year ERA than current year ERA is. There are rare cases (rare enough that Mariano Rivera is one of them) that over a large sample, a pitcher can out-perform his DIPS in his ERA. Buehrle is another guy like that.
However, guys with less-than-massive samples who are assumed to be "DIPS-beaters" usually aren't. Jeremy Hellickson, Ricky Romero, Matt Cain are some good examples of guys that aren't. That's the big argument against Sanchize. Estrada could be, very slightly, but 4/10th's of a run is a MASSIVE difference, and even then, Estrada doesn't look very good. As well, over the course of his career, he hasn't out-performed his DIPS that much, despite being a good candidate for it.
In all, it's not really wise to bet big amounts of money on guys being "DIPS-beaters." Would love to answer any further questions.