So this time instead of the randomizer, I went looking for games that have fewer than average number reviews. Considered first “The Master’s Lair” but it is not playable without a download, which removes it from my list for now. This may also be why other reviewers have overlooked this game.
Next tackled “Return to Claymorgue’s Castle” by Claudio Daffra. Right away I was impressed with the amount of effort that went into programming this thing. It appears to be a full blown parser type game written with a Twine engine, a system I’ve always imagined would be possible, but haven’t quite seen until now. Return to Claymorgue castle is a standard Dungeons and Dragons type scenario, with a mixed party of four characters , each with their own skills, exploring the ruins of a medieval castle. This is set in more modern times than a classic D&D romp; one of the characters is a tech whiz who carries her cell phone. Commands are given by picking a subject (from any of the four party members) a verb (from a list as extensive as the standard verbs of Inform), a direct object from the current location, and sometimes an indirect object as well. Included with the author’s notes is a letter from classic text game writer Scott Adams, endorsing this game (which shares its setting with one of Adams’ vintage games). Pretty impressive.
A difficult and impressive coding challenge. But was it done well?
Unfortunately there are a number of features of this game which make it difficult to play, in spite of the apparent passion and time that went into writing it. The mechanism to pull words into a sentence to make commands is lumbering. The words must be pulled in the correct order to make productive sentences, and once a command is executed (whether successful or not), all of the word slots are emptied. It took me a bit of practice to even launch a single productive command. At first I assumed that “myself” was the default player, but it turned out that even the noun “I/me” has to be pulled into EVERY command before it can be processed.
I found no obvious bugs in the command processing. However, there is also a shortage of customized responses to un-productive commands. “I read leaflet” or “I examine leaflet” (the verbs you’d normally perform on a leaflet) give a stock response:
“in the courtyard. I examine the leaflet; inspect : a leaflet”
There were some other issues as well. Initially the motivation for exploring this castle is ambiguous.
You are fascinated by the history and the stories of this place, but also wary of the dangers that might lurk in the shadows.
There is some nice pixelated art work (shout out to Scott Adams) but with the block text layered right over it, the text is sometimes difficult to read.
Without the payoff of more detailed and customized responses, I began to lose interest. I turned to the walk through.
This was an incredibly ambitious project, but I think the puzzles would have been awfully difficult without a walk through. Even “verb-object” puzzles can be difficult, but when you also have to assign the right subject to the command, and perhaps a second object, exponentially more difficult.
Back to the goal of embedding a parser in a point and click game; I think it is a very respectable goal. “Mrs Gosling’s Last Case” is aiming for the same thing from the other direction (a point and click full parser game written with inform) but more successfully because the key words of the command are closer together on the screen (and there are usually fewer words). Or if you’re aiming for a more pared down parser system, Robin Johnson’s “Detectiveland” (which I think was also influenced by Scott Adams) won first place in IFcomp 2016. But these systems are hard to write and design, and I know from reading reviews after my own attempt to do something like this in the 2023 comp, that players get frustrated when there is too much effort or delay required to execute a simple action.