I am really into some of what this article is suggesting. I think that he doesnt succeed particularly well in looking for wider solutions to speed up play-time, but the concept of averaging damage roles in play is interesting.
For minion creatures, or for large numbers of creatures, the averaging of damage rolls make much more sense. Against 5 goblins making 2 attacks per round their damage over 5 rounds, for example, will in principal average itself anyway. Standardising this process seems like very sensible behaviour.
For single powerful creatures, however, the variation between a string of 1’s and a critical threat is what makes the encounter super-edgey. In addition, powerful foes rarely last for 10’s or rounds anyway so that averaging the damage is completely artificial – they are unlikely reach average scores in their own short time so forcing them is misrepresentative.
I think you make a good point about the difference between a horde of goblins and a big bad solo monster. Just from a practical standpoint, taking an average for multiple goblin enemies over several rounds saves time. That does take away some of the drama and randomness. Like anything, a DM can pick and choose how to use the idea in his or her campaign.
Thanks for checking out the article.