My thoughts exactly...the value of a transaction has no bearing on the cost to process it...so valuein has should have no bearing on priority. I think the winning formula is to base priority on a combination of age, fee, size, and time since the last transaction involving one of the accounts in the transaction (to deal with the spam issue). That last one should be carefully tweaked to better deal with the recent delay issues experienced with slush's pool. This priority would govern the forwarding behavior of the clients. In addition, clients should factor this priority calculation into the decision on whether or not to accept blocks. Blocks that seem to be grossly out of alignment with these priority assessments should be treated with great suspicion.
The value of a transaction does have bearing.
People are far less likely to spam using 10000 BTC transactions. Faster confirmations for bigger sums encourages use of bitcoin for larger amounts, which is always a good thing for the economy. And as ArtForz pointed out, it encourages sending larger amounts per tx, which implies less block chain size for the same volume of currency flow.
Less likely, but, I wouldn't completely discount the possibility.
The fee has the same effect though. If adding a fee gets faster processing, and you want faster processing, then you pay the fee right? Well, if you want to jump the queue with smaller values, thats fine but that means the fee will be a larger percentage of your total value.
That said, I doubt either of these effects is going to have much bearing on anyone's behavior in terms of what they send. If I make a transaction, it is either to consolidate btc into another wallet, or, its to pay someone for something. Either way, the amount I send is not going to be influenced here.... but the fee might be.