The most unusual part of The Dreamhold is the serious attention paid to making it suitable for new players. Generally the farthest IF games will go in this regard is including a multi-screen document on how to play that nobody can seriously be expected to read through, and wouldn't remember if they did. This game, on the other hand, is set up to give help adaptively. When you start it tells you to type help, and gives short suggested commands. When you find a door it encourages you to try and unlock it; when you find an object it encourages you to try and pick it up. It has suggestions on examining objects and even the basics of solving puzzles. I don't think this is perfect — I played with the tutorial mode a bit and it doesn't seem like it'd be enough on its own to get someone from zero to being able to solve the game — but it's miles beyond anything other IF games have tried, and it'd be great if this got rolled out into a library for other newbie-friendly games to use.
So, ok, putting aside new players, what does it have for everyone else? Plenty. And in a way that's the problem. For people who like big-machine puzzles, it's got some big machines. For people who don't, well, it still has them. Similarly it caters to both people who like red herrings (a surprisingly large number) and people who like puzzles that require careful exploration and lateral thinking — but if you spend your careful-exploration energy on a red herring, sucks to be you. In fact, the two playing strategies are almost totally opposite. The way to interact with a game with red herrings is to keep moving, not spend too much time on anything, and expect that if stuff is solvable you'll be able to solve it by doing the obvious stuff. The way to interact with a game with deep puzzles, on the other hand, is to expect that there'll be only a few puzzles that you'll have to think about for a while; everything is going to be significant but the significance of a piece might not come until you've tried a lot of different things with it.
"Plenty" is also the problem, or, I should say, a suboptimality, in the game content. While all the areas are individually delightful — it's all hand-crafted and a real pleasure to poke around in, and the player is frequently rewarded with cool scenes happening when they poke at things — there's simply too many, and there's no overriding connection between them. While this is in-game justifiable by setting the game in a wizard's house, it makes the playing experience rough. It's hard for the game to bring together the more than a half-dozen conceptually separate areas into any kind of whole, story-wise or just for the player's navigation, and the situation isn't helped by having such a wide-open map with almost all the areas available from the start of the game. The obvious comparison here is with So Far. That game had many diverse areas as well, but it all held together much better: there was thematic unity between the pieces even if there wasn't setting-unity, and they were presented in a way that gave you time to assimilate each part before moving onto the next. Probably the structure of So Far was a little too restrictive, but The Dreamhold goes too far in the other direction. I'm a little surprised a game for newbies is designed like this, in fact. Even more experienced people will probably find themselves wandering around at one or two points wondering what to do next.
The Dreamhold focuses mainly on exploration and puzzle-solving so I haven't said much about the plot yet. There is some plot here: it's a pretty standard amnesia story where you wander around seeing flashbacks and trying to figure out what happened. The plot is even interesting but, like in many of Plotkin's games, maddeningly elliptical in parts. I expect somebody will post on the newsgroup with a proper reconstruction of the backstory, but personally I haven't been able to work it out entirely. That's not a killer for the game — like I said, the plot isn't the main part — but it's an irritation. The same applies to the endings (yes, plural, and that's definitely a plus): suggestive but not conclusive.
This is the last paragraph, so it's the one where I say "despite what I said in all the previous paragraphs, this is a great game." And that's pretty much the case. Plotkin is extremely good at what he does, and even when I disagree with what he's aiming for I have to admire the quality of his craftsmanship. The game is the right size, too. It's much meatier and more rewarding than a comp game: it should keep you busy for a couple evenings, at least. That size also means that whatever you're looking for, you'll probably find it*. I definitely found enough brilliant bits to keep me happily playing for all but my last session; I'm just sorry that in the end it felt like the game design was working against me and keeping The Dreamhold from being as satisfying as it could be.
*Unless it's an NPC, anyway; we've seen a few stabs at them across Plotkin's games but thus far he's always stuck to trickery and misdirection rather than, I dunno, taking the risk he won't be able to write the NPC well enough or something.