First, James Robertson correctly states that the JVM needs to be unfrozen, so that dynamic language performance could be decent. I agree, but I would be more than slightly surprised if Sun ever did this.

At the other end of the spectrum, Patrick Logan picked up and expanded upon my post yesterday questioning whether the days of the VM approach are coming to an end (this of course will make Microsoft unhappy because they just got here). Patrick points out old (but not necessarily well known) results that allow natively targetted languages to achieve all of the benefits of VM's. At least the VM benefits that I listed.

And earlier in the day, there was a confluence of these two fine gentleman over the question of 64 bits for dynamic languages. Anything that makes it go faster is good. Here's looking at those G5 Powerbooks and Opteron desktops when they arrive.

A leader is best when people barely know that he exists. He is the teacher who succeeds without taking credit. And, because credit is not taken, credit is received.
-Lao Tzu, 6th Century B.C.