Just because no one's done it yet in one post.
Here are the average results using the proper math as given here under the Open Ended Rolls heading:
http://www.rpg.net/columns/rollthebones/rollthebones2.phtml
Base averages
d4 2.5
d6 3.5
d8 4.5
2d4 5
d10 5.5
d12 6.5
2d6 7
Vorpal average and % increase over base average damage
d4 3.33 33%
d6 4.2 20%
d8 5.14 14%
2d4 5.33 7% rerolling only an 8
2d4 6.66 33% rerolling each 4 individually
d10 6.11 11%
d12 7.09 9%
2d6 8.4 20%
So, interpreting 2d4 as a die and only rerolling on an 8 keeps the 2d4 more in line with d12, but it brings it further from everything else because the d12 has the smallest increase. However, rerolling all the 4s makes 2d4 superior to every other die type.
Let's use the +6 weapon example.
Nonvorpal
4[W] = 26
Rerolling only on an 8
4[W] = 27.33 damage
Rerolling each 4 individually
4[W] = 32.66
Personally, I think an extra 1.33 damage per attack not accounting for the extra crit damage is a pretty weak boost to go from a level 26 weapon to a level 30. The basic +6 gives you an average 21 extra on a crit, the vorpal gives an average 33 extra. That doesn't account for how to adjucate crit damage with the rerolling property. And then there's also the daily power.
One more thing, no one has really commented on the bell curve aspect of the 2d4. People have mentoined it, but not really gone in depth. I'm sure we all know that rolling more dice gives a staggered distribution of possible outcomes with a higher probability of values in the middle than at the ends. Well, let's look at a weapon with exploding dice.
The odds that a d12 exploding weapon will give a result of 21 or higher is 1 in 36. That's 1/12 x 1/3, or 2.78%. Now, looking at 2d4, percentage chance of getting 21 or higher on one roll, when exploding both dice is .83%.
Correct me if I'm wrong please. I don't have a PhD in mathematics or statistics, but I know enough to figure the above out.
This just shows that average damage is only one means of comparing weapons. The likelihood of scoring a lot of damage on a single attack is far greater when rolling a single large die. We've known this for a long time and it still applies here.