In no particular order:
1 Savoir Faire, Emily Short. The best example of a very powerful design approach: concoct an interaction mechanic that is distinctive and engaging, that can be applied widely and modified into variations and otherwise elaborated upon, then build that mechanic deeply into your world, character and themes. It’s also got some immensely satisfying aesthetics. Its main flaw is that the difficulty curve is a little steep at the lower end; like a lot of people, I bounced off the opening section on my first attempt.
2 Spider and Web, Andrew Plotkin. This is the game that made me realise that text adventures could be literate, clever things. Yeah, That Puzzle is a doozy, but that never impressed me as much as the handling of the frame-story.
3 Varicella, Adam Cadre. What Cadre does best is to take really vicious nastiness and transpose it into a funny, larger-than-life, almost slapstick format without taking the edge off. Mechanically, it’s excruciatingly hard, the basic assumptions of the world make an unsatisfying ending inevitable, and it’s in sore need of multiple solutions.
4 Alabaster, Emily Short et al. If this didn’t exist then Galatea would have this spot. It’s the deepest conversational IF to date, it’s appropriately sinister, but at the same time it has that satisfying, Bujold-like feel of a conversation between two thoroughly reasonable people. Since this is a work of many hands, the tone is a little uneven in parts, but overall it’s excellent.
5 Violet, Jeremy Freese. One of the games that I have a dismissive reflex about, but which becomes a lot more impressive when I sit down and think about it. It has an extremely firm grasp on a lot of the things that IF does better than any medium: a world teeming with enticing and evocative objects, constraints on interaction defining the protagonist, play as a dialogue with the narrator. It doesn’t advance IF design in the slightest, but it employs established techniques to their limits.
6 Photopia, Adam Cadre. Shocking choice, I know. And yeah, a lot of this is because the balls-out cheap emotional manipulation worked. It’s still a brilliantly orchestrated sequence. The heart of it is not really about untimely death; it’s about distance, about the other person being necessarily just out of reach.
7 The Baron, Victor Gijsbers. Moral-question IF is extraordinarily difficult to pull off, and this is the best example that I know of. Takes the standard philosophy approach of interrogating a question by repeatedly reframing it, but does so in a way that works as a dramatic story. Wrenching even if you’re expecting the carpet to be pulled away.
8 Cryptozookeeper, Robb Sherwin. It didn’t strike me as strongly as did Fallacy of Dawn, but a lot of that was because I wasn’t expecting to enjoy Fallacy. It’s a night journey, sprawling and random and rich and dark and cheesy, it makes no sense in a huge number of ways, it has a heart of gold and a mind in the gutter, it’s Americana at its best. The usual things to talk about are Robb’s weirdly excellent prose, vibrant NPCs and off-the-wall imagination.
9 Anchorhead, Michael Gentry. A masterpiece of pacing and atmospherics. When I first played it I hadn’t read anything from the Lovecraft mythos, so my estimation of how it matches up is unreliable; my feeling is that it’s as good as anything in the mythos, although this isn’t an incredibly high bar. As with most Lovecraft, the climax doesn’t live up to the buildup, but it scarcely matters.
10 Worlds Apart, Suzanne Britton. A touch too new-agey for my tastes, and you get the distinct feeling that it’s not a complete story in its own right but rather an excerpt from something much larger. Nonetheless, strong, well-integrated puzzle design, strong setting, and an amnesia device that serves to help explore the character rather than an excuse to avoid doing so.
11 Bad Machine, Dan Shiovitz: fiendishly difficult, probably incomprehensible to anybody without a little programming background. About a decade since I played it, so possibly I’m viewing it through the rosy mists of nostalgia. Huge amounts of learning by death, though it could hardly be otherwise. Still: dark, atmospheric, clever puzzle mechanics.
12 Mentula Macanus: Apocolocyntosis, Adam Thornton. Again, I had a dismissive reflex about this when I first played it, but it grew steadily on me. Much of this is because the process of appreciating IF isn’t a fixed, author-reader experience; much of the fun of Mentula lay in looking up the references, reading the code, talking and thinking about it, reading other people’s reactions and analysis, and (ahem) drawing fanart and writing cheap rip-offs. Mentula invites this sort of thing by being self-indulgent and allusive, publishing its code (bringing the player closer to the authorial process), and using an open, fantastic-voyage structure.
13 Moonlit Tower, Yoon Ha Lee. Overwriting that works, rich symbolism and aesthetics-of-objects, a story more suggested than told.
14 Slouching Towards Bedlam, Daniel Ravipinto & Star Foster. Like Savoir Faire, this starts out slowly: when I played it in-comp I got the sense that something good was waiting in the wings, but didn’t manage to actually get to it. But on the second attempt, somewhat later, it really opened up. Probably the best, most deeply engaged steampunk I’ve seen. By which I mean that it’s genuinely interested in both the artistic tropes and the technology of the era, rather than treating it as a stylish gloss.
15 The Edifice, Lucian Smith. As anthropology it’s honestly pretty crap – nobody seriously thinks that horses went from wild to riding animals at a single bound – but it works as a stirring allegory, it has well-integrated puzzles.
16 Vespers, Jason Devlin. The best 9:05 Twist game I know, and a nasty little subversion of metagaming. And I’m all in favour of medieval fiction being bleak and horrible.
17 City of Secrets, Emily Short. I tend to underrate this one, partly because I tested it and partly because Emily has very mixed feelings about it. But even though the underlying story design has some problems, it’s full of beautiful details, the setting’s well-conceived and there are a lot of individual sequences that are very good indeed.
18 Gun Mute, C.E.J. Pacian. The interaction gimmick is really very effective, but this also does a lot of very strong, economical things with setting, characterisation and NPC interaction. It helps that I grew up on 2000 AD.
19 Blue Lacuna, Aaron Reed; 20 Make It Good, Jon Ingold. Both works that I know damn well deserve a spot on this list, even if I haven’t finished them and am not in a position to make a complete judgement.
Things that occurred to me while compiling these:
a) We really need to get better at endings. (I think this is a general problem with computer games.)
b) It’s really hard to think of genuinely good games that don’t seem totally obvious. Genuinely good IF games pretty much always make it into the canon? Which means that if you want to talk about interesting games outside the conventional greats, you pretty much have to deal with things that are mostly broken.
c) I am really not very good at articulating what I like about games.
d) The ones I was sure about cut off after sixteen entries; after that there were probably a dozen games that I think deserve to be on the list, but I couldn’t really distinguish between.
e) man, we are really not good at doing things that aren’t F/SF.