In my mind, in order to be worthy of a high draft pick, a position must meet three criteria.
Criteria #1- the position must be valuable. This one is pretty self evident. There's no point in spending a high draft pick on a position that will contribute very little value to your roster. If your league has a "head coach" position that gets you 1 point for a win and 0 points for a loss, then there's no blessed reason to ever take a head coach in your draft ever, period. They just aren't valuable. The difference between the 2007 Patriots and the 2008 Lions is just 16 points.
Criteria #2- the position must be predictable. This one should also be self evident. How frequently does the player who ranks 1st in preseason ADP finish 1st in points? How often does he finish in the top 5? The top 10? How often do guys taken outside of the top 12 finish inside the top 6? How often do guys taken outside the top 24 finish inside the top 12? There's no sense spending a high draft pick on a position that has been historically unpredictable, because you're just as likely to get a stud with the first player off the board as you are with the 12th.
Criteria #3- the position must be difficult to replace. If you don't spend a high pick on the position, then how easy will it be for you to acquire quality contributors during the season? How often are quality players available on waivers? How successful are by-committee approaches to the position?
By my reckoning, defenses will sometimes meet criteria #1 (they are often valuable- depending on the scoring system, owning the #1 defense can frequently give a large advantage), but they almost always fail MISERABLY at point #2 and point #3. There's so much unpredictability from season to season that we never know for sure which units are the dominant ones until the season is half over. Few positions see so many players regarded as "bad" during the preseason wind up radically outperforming expectations during the season (witness Cincinnati, Denver, and New Orleans last year, just to name a few). Part of that is because defenses rely heavily on 11 moving parts, and all it takes is one of them getting injured (witness: Troy Polamalu) or playing out of his mind (witness: Darren Sharper) to radically alter the defense's fates. In addition, I don't think there's a single other position in fantasy football that is affected by matchups to a greater degree, which means that Defense-By-Committee approaches will frequently score as highly as getting a single stud defense. Plus, there's absolutely no positional scarcity (I have yet to hear of a start-2-defense league), which means impact performers (or bad defenses with fantastic matchups) are frequently available on the waiver wire. Just look at Arizona last year. They were a poor defense all season long, but around week 10 a lot of people started noticing that they played Detroit and St. Louis in weeks 15 and 16, so people quietly added them for the playoff run. Sure enough, in those two weeks the Cardinals totalled 6 sacks and 7 takeaways, giving up only 27 points (plus another 7 points given up by the offense on a 100 yard INT return). The Cardinals outperformed most of the "stud" defenses when it really mattered simply because they had the right matchups.
Given all of that, I would never take a defense high. I'd much rather grab some sleepers with upside, constantly pan the wire, and play matchups to achieve the same result (if not better!) with a fraction of the resource investment.
