Let me start with a couple of principles, one theoretical, the other a standard fantasy drafting approach. First, the modest (so I think) theoretical proposal:
Amount of physical talent needed to score goals > Amount of physical talent needed to prevent goals
Defense is a game of angles, offense is a game of shots. Defense stresses less movement, but ample awareness to use angles to your advantage. This is true for goaltenders who are cautioned against flashy moves, as well as defensemen cautioned against lunging at close plays on the opposing blue line. It's evident whenever a defenseman skates backward to keep the forward in front of him/her, and the forward has to find a way to get past or get off a shot. The value of preventing goals versus scoring them isn't at stake here, but rather the amount of talent that needs to be there to be good at one versus the other. And that's where the draft comes in.
Secondly, I've used fantasy hockey in the past to show some of its inadequacies and suggest that it might perpetuate hockey fans' emphasis on boxcar statistics (goals, assists, sometimes +/-) as well as the curious inclusion of penalty minutes. Despite this, I think there are redeemable things in fantasy hockey, in particular with the drafting process. It's common nowadays to see people going into a fantasy draft organize the available players into "tiers" of talent, with an idea that it will make it easier to draft on the fly. Often, the players are organized by position; these positions are determined by how your league is identifying them (for instance, in hockey some leagues separate right and left wings and centers, other just identify wings and centers). If you do this, you notice two things: a.) some positions have greater population in their high tiers, and b.) the spread of performance between the top tiers and the bottom tiers is greater at some positions than others.
The fantasy drafting principle that emerges from this is that, in positions where the spread of performance between the top tiers and bottom tiers is less, you don't draft early what you likely can get later.
To a degree, NHL entry drafts are also ordered into tiers, particularly in the reports from the Central Scouting Bureau. As with fantasy drafts, the tiers are separated into positions; also like fantasy drafts, there aren't a lot of useful ways to compare across positions. While I would agree it's not particularly easy, I do think that you can look at the results of 1st round picks versus their non-1st round counterparts and examine the spread of talent, and deduce which positions should be addressed early more frequently than those that can be likely addressed later. Take these numbers from players born in 1970 and sooner (roughly covering the drafts from 1990 to where I placed the cap, at 2008):
|Forwards||Per Game||82g Season|
|1st & Non-1st rounders 1970-born and sooner GVT/G||0.075||6.114|
|Non-1st rounders 1970-born and sooner GVT/G||0.060||4.912|
|1st rounders 1970-born and sooner GVT/G||0.104||8.525|
|1st & Non-1st rounders 1970-born and sooner ESPPG||0.372||30.511|
|Non-1st rounders 1970-born and sooner ESPPG||0.357||29.274|
|1st rounders 1970-born and sooner ESPPG||0.402||32.994|
|1st & Non-1st rounders 1970-born and sooner GVT/G||0.072||5.941|
|Non-1st rounders 1970-born and sooner GVT/G||0.069||5.679|
|1st rounders 1970-born and sooner GVT/G||0.079||6.502|
|1st & Non-1st rounders 1970-born and sooner ESPPG||0.219||17.946|
|Non-1st rounders 1970-born and sooner ESPPG||0.232||19.012|
|1st rounders 1970-born and sooner ESPPG||0.191||15.668|
In other words, the difference between the average 1st-round forward versus non-1st round forward that made the NHL was the difference between Andrew Ladd and Blake Wheeler this year GVT-wise, or basically the difference between a good 2nd-line forward (lower-end 1st-line forward) and a 3rd-line forward. The almost-four point difference in even-strength points per game should be tempered with the knowledge that 1st rounders get a bit more time in special teams situations, at least initially in their careers. Either way, that's a pretty strong line between one and the other. In terms of defensemen, it's a much finer line. In terms of GVT, the difference is roughly the difference between Andrew MacDonald (upper-end GVT) and Kevin Shattenkirk (lower-end GVT) this year. Once again, the even-strength points difference should be viewed with the knowledge that 1st rounders will initially get more special teams time. The difference in distribution becomes even more disparate if you consider the ceiling of a 1st round forward versus the ceiling of a 1st round defensemen. Even the 1st round defensemen with the best-case results (players like Drew Doughty, or Chris Pronger) will struggle to get into the same GVT company as the best-case results of 1st round forwards (Sidney Crosby, Alex Ovechkin) and even some with good, not great results (Patrick Marleau, Jeff Carter). In fact, we've seen just as many non-1st round defensemen (Zdeno Chara, Duncan Keith, Nicklas Lidstrom, Kris Letang, Dustin Byfuglien, Christian Ehrhoff) compete at the highest level league-wide as 1st round defensemen in recent years. Yet here are the drafting numbers:
To a degree this is understandable: it is much easier to evaluate offensive ability than defensive ability, and so we see a lot of 1st-round defensemen being projected pretty well offensively or as big hitters, but frequently turning out to have dubious defensive reputations or weaker-than-expected offensive numbers (or both). But part of the reason I used GVT was to move a bit away from a primarily offensive comparison, because it is pretty clear that defensemen will not, on average, match forwards in that capacity (which has always been the case). What is clear is that there's a gap in scouting, but I would argue that with the theory I proposed above, and considering the draft is focused on projected talent, you still shouldn't be seeing more than 3, maybe 4 defensemen being drafted in the 1st round in any season.
I find it no small coincidence that what many view as one of the better drafts in recent history was 2003, where a large number of forwards were taken in the 1st, and the defensemen from the 1st did not pan out particularly well. Maybe we'll look similarly on the 2010 draft. Don't even get me started on goaltenders in the 1st round, though I'd like to point out that that is one area where the market has come around to this fantasy hockey-like understanding. If the difference between the tiers of goaltending are small, the market for average and slightly-above average goaltenders should not be too high...and that was the case in the offseason (though there have been some crazy counter-developments).
If I'm a GM drafting, I'm taking forwards in the 1st and maybe 2nd rounds, and defensemen nearly every round after that. Not doing that in the 1st round is kind of like buying a lottery ticket to win $750,000 when a ticket with similar (maybe even better) odds could win you $1 million.
Oh, and I'd get my goalies via tryouts offered to 25-to-30 year olds from the Elitserien, SM-liiga, KHL, and Czech Extraliga.