First off, I think it reflects the point that if you are signing a 30 year old to a five year contract, present performance is not likely to be sustained for all five years.  In the past, I think the assumption was that the value of a win had significant helium, so inflation makes sure you don't end up over paying.  Since revenue growth over the next few years is uncertain, teams are concerned about tying up too much money over too many years. 
Specific to the M's, I think that Jack values roster flexibility so that he can avoid locking in on one player versus another.  This is why Chone Figgins is the primary target.  If Figgins unexpectedly sustains his 2009 performance, he's a steal.  If he stabilizes at his 2007-2009 mean, he's a good deal, and if he underperforms, he's an overpriced Mark McLemore.  The chance he's a liability is absolutely minimal.
Contrast this with Adam Dunn.  Dunn is currently a 3.5 win player at DH, and likely shouldn't own a glove.  I'd say his value matches Figgins value, but he is the antithesis of flexibility.  Why does this matter?  Because a declining bat could easily make him utterly useless -- see Vidro, Sexson, ...  He is not at all flexible.  If Poythress or Raben surprise, but have no fielding skill, Dunn blocks them and lead gloved prospects have limited value in trade.  Among Tui, Triunfel, Lopez, and Saunders who's going to develop further?  Who are you going to want in your lineup in 2011 and 2012?  With Figgins you can play the odds and expect someone(s) to develop, but you don't have to guess who. 
The point is with a flexible roster you don't need any particular player, so you don't need to pay a premium to acquire anyone.
 

Traditionally, great players are under paid, even using a linear scale for value.  I think this is because the collective bargaining agreement and risk averse GMs have lead to the over valuation of league average veterans.  If someone is over paid and the dollars are fixed, someone must be underpaid.  While this is primarily young players, it has also been stars.
I believe the reason is quite simple.  Pujols has been worth $110M over the next three years ignoring the non-linear nature of value, but it doesn't really matter.  Why? Because only a very limited number of teams could even consider paying him that much.  That would be 40% of the M's payroll.  What teams could pay that amount?  Yankees, Mets, Red Sox... Dodgers, Cubs?   

I completely agree with your points.
I think Doc is right that the more W you are above RP the more the Ws should be valued. We saw this last season where the stars still got nearly '07 value, but civics were very much underpaid.
Last year it went a little too far though and kind of created situation where civics are where the value is. A few years ago, like '04, it was the stars like Guerrero and Escobar that were underpaid in the market.
 

Incidentally, Tango responded to that question a few days ago. Anyway, the reason stars aren't paid an exponentially greater salary than average players is because they are guaranteed many more years. Average players don't get 6 or 7 year deals, only the best players do. So while the stars may be paid in a linear fashion on a yearly basis, they are still guaranteed an exponentially larger amount of money than lesser guys. That's why long term contracts so often overpay players near the end; a lot of the money the player deserves early in the contract gets pushed out into future years when the player is in a decline. Therefor, if you think players should be paid in an exponential fashion on a yearly basis but want to maintain the current risk/reward then you should guarantee fewer years for stars (or guarantee more years for average players).