Miss Gosling's Last Case

With Inform's core code having grown so much that only the smallest scenarios produced by it will still fit within Z-machine, one might wonder whether that venerable format is destined for the dustbin of history. Sure, there are still people actively using Inform 6 with the PunyInform library to make Z-code games for retrocomputing platforms, but that's a niche within a niche. With this work, author Daniel Stelzer proves that the Z-machine is still a vital platform when used with the relatively new but sophisticated Dialog language.

Miss Gosling's Last Case plays very well. Puzzles are meticulously designed and well-suited to appeal to those who would be attracted to the murder mystery genre -- requiring an active imagination that takes careful notice both of what is said and what can be imagined about the scenario being depicted.

Only basic verbs and simple commands are in use, a constraint imposed by the separation of story protagonist from primary actor. By ensuring that there is an in-game reason for preferring simplicity, the player is subconsciously prompted to throw out any ideas for actions that cross a certain low threshold of complexity. It gives something of the feel of a limited parser game without actually being one.

The game's text has an emphasis on providing backstory and characterization, largely eliding physical descriptions of the scene outside of a few key objects. This is done skillfully -- at first I did not notice the style, because room description text provides introductory exposition as the player gets familiar with the situation. Should the lack of detail become noticeable, that is a cue the player should simplify the approach being taken. The object implementation is spare enough that, should imagination fail, even brute force approaches are likely to pay off within a reasonable number of commands.

Quite a lot of work has been put into creating a smooth and seamless play experience. New players will benefit from many "invisible" parts of the system that are designed to support that goal. First and foremost are >FIND and >GO TO verbs that make navigation as simple as can be. Object disambiguation is handled with a numbered selection that makes it very clear how the parser is "thinking," and that in combination with very descriptive error messages will rapidly train a new player in the preferred method of interaction. More subtly, the game design itself ensures a sharp focus on specific goals at all times, even during the middle game when one has a choice of order in which to pursue subgoals. Lastly, the introductory scene offers a tutorial voice that is sure to help total newbies get started with a parser, though it is extraneous to someone familiar with the form.

My initial impulse is to give this game four stars, which translates roughly as "distinctly above average" and/or "highly recommended" in my rating scale, but there are a couple of minor shortcomings that keep it just below that threshold. One of the segments (Spoiler - click to show)(involving identifying a rosebush of blooms with a particular color) does not feel as well-implemented as the others. (Spoiler - click to show)Specifically, although a point is awarded when the correct actions have been taken, the player is not notified about which rosebush is correct and must deduce it from some diagrams. This is not difficult, but neither is it particularly interesting, and stylistically it is out of step with the rest of the work by adding even a speck of unnecessary friction. Also, the multiple locations of the tea garden just seem "deader" than other parts of the house from a writing perspective; they are restricted to repetitive descriptions of largely undifferentiated locations with few objects. Perhaps less important but worthy of adjustment is the pacing in the final scene. (Spoiler - click to show)It took several tries to work out the correct move to trigger a win, and it felt very arbitrary that it should by that move which does so. Repeated barking should be just as effective given the situation, and would be the low-friction option to conclude the game after the real puzzle has been solved. As a final nitpick, it would be nice to be able to turn off the tutorial mode at the beginning. (Note that any or all of these criticisms may have been negated in release 2, which was recently posted.)

I'm going to go ahead and round up a bit for my star rating, though I'll hold off on letting it count toward the average in the hopes of a post-comp release to sand off the handful of remaining rough edges. In the meantime, I do very much recommend this piece to anyone looking for a bit of fun, and I would even suggest it (with reference to the provided hints, if needed) as a first experience with IF for someone who likes the murder mystery genre. My hat is off to Stelzer for creating a first-class introductory work easily on a par with Infocom's best of that type. Bravo!

[Note: It turns out that much of the preceding unintentionally -- but almost exactly -- echoes an off-site review by PB Parjeter, which was written prior to this and to which I've added a link on the game's page here. I guess that's evidence that the observations are well-founded!]