Here’s an interesting review of Half-Life 2 episodes 1 and 2.  I think it does a pretty good job of capturing what works and what doesn’t in the two episodes, and explaining why I never had the impulse to replay ep 1.

I also agree that the climax to ep 2 sucks.  I could never finish it, in fact– it depends entirely on being able to aim those stupid stickybombs onto striders, then shooting them, then racing to the next strider.  Even with the ability to save, it was just impossible.

On the other hand, exploring the antlion den is weird and interesting– it gives the sense of alienness that ep 1 didn’t really deliver.  And the section where you work your way through an enormous factory is an interesting set of challenges, and I think he’s right that they tie it together spatially in a very clever way.

The one hole in his theory of progress at Valve, I think, is that HL2 is more fun than either episode.  I’ve played HL2 multiple times… there’s really no puzzle to it anymore, it’s just relaxing sometimes to play a good known game.  It’s also a master class in how to make a shooter just varied enough.  The shooting-Combine-enemies sections are nicely broken up by cutscenes, vehicle sections, and the very different feel of Ravenholm and Nova Prospekt.  The second City 17 section is a bit tedious, but the Citadel bit ends on a high note.

I wonder if the real reason for the delay is not the difficulty of continuing the narrative, or modernizing it, but Valve’s unusual internal structure.  People basically choose what project they want to work on, an idea which undoubtedly generates excitement and commitment, but probably isn’t as good for steady progress.  If everybody prefers to work on TF2, L4D2, and Dota, then there just won’t be the manpower for episode 3.