Fall mystery games

Thursday, October 10, 2024

Comments: 1   (latest straightaway)

Tagged: reviews, no case should remain unsolved, thalassa: edge of the abyss

I guess I'm going to have to change my mystery game tagline. I'm playing more of these and getting better at them. Or maybe more designers are picking up the Obra-Dinn-ish static deduction model, which is the format I'm most simpatico with.

(Speaking of that model, the next Golden Idol game will be November 12th. And the revamped Roottrees is coming on January 15th! With new material and new puzzles. I probably won't write a full review post, since I wrote up the original last March, but I'll definitely play it.)

But right now...

  • No Case Should Remain Unsolved
  • Thalassa: Edge of the Abyss

No Case Should Remain Unsolved

A twisty suburban drama about a missing girl, told in flashback to/by a retired police detective. As the intro notes, everybody is lying about something -- but what?

The classic fill-in-the-blanks static deduction model; but instead of "who/where/how" as the blanks, it's "who said it / in what order?" You have a disorganized heap of police transcript fragments. You have to put each fragment in the correct column (which suspect is being questioned?) and then get them into the correct chronological sequence. When two fragments are correctly adjacent in the correct column, they fuse. Link enough fragments and you get a key to unlock further fragments.

There's also some puzzle-based gating. You may need to select a piece of evidence to support (or contradict) a particular claim, or fill in a date based on available evidence.

The UI is very polished; it feels great; it's completely opaque. I'm sorry, but even after I finished the game I didn't understand what I had been doing. It's all organized by hashtags in the transcripts. You click on a hashtag, and the game either shows you a new piece of text or it locks up the script and blinks the map at you. At random. The more you play, the worse your odds.

Okay, not really random. Here's my best understanding: every piece of text is gated by one or more hashtags. To unlock it, you have to select a matching hashtag in an already-open piece of text. But each hashtag can only be used once; then it's crossed out. Thus, a constantly expanding "frontier" of available tags and unlocks. When you select a locked fragment, the game blinks all the open fragments that have available matching tags.

This is a pretty neat model. It combines the Her Story association web with a limited-resource mechanic. It lets the author gate some parts of the story tightly (a key hashtag becomes available in one spot) while leaving other sections as free exploration (generous common tags). I can absolutely imagine a Portal remake using this system.

But the tutorial fails to communicate what's going on. To me at least. I don't even know if I've described the thing correctly! If I've hallucinated this model, dibs on the idea.

Okay. UI aside, did I like the game? Yep! It's character drama and the characters are all immediately readable through the lens of their police transcripts. The story does a great job of subverting and then yanking out your assumptions as you move through the space of deduction and revelation.

(Make sure your brain is set for Korean family names, as that's where the story is set.)

I might quibble about the double ending. You can trigger the finale any time after you have all the critical clues, but fully completing the transcript grid gives you access to the "best" ending. But: the "best" ending and the "okay" ending are different and contradictory revelations. This is always a difficult landing to stick; even more so in a deduction game where you're trying to collect everything into a coherent whole.

But, as I said, a quibble. Unsolved is deft, clever, and -- best of all -- short. Set aside an evening and dig in.

Thalassa: Edge of the Abyss

You're a deep-sea diver in 1905. The liner-turned-research-ship Thalassa is out searching the sea floor for a lost 16th-century galleon. But you lose first a crewmate, and then your entire ship. Next thing you know, you're diving for the wreck of the Thalassa itself, hoping for answers.

The fill-in-the-blanks setup here is a tree of individual mini-mysteries. Who was the captain talking to in this (wax cylinder) recording? Supply name and evidence tidbit. Who is the captain upset with? Name, evidence, supporting document. And so on. You scour the ship for evidence bits, collect, and fill in the chart. Each mystery unlocks further mysteries, and when you've resolved everything, that's the game.

But wait, doesn't this mode sound familiar? It's exactly the "mind-palace" interface of Frogwares' recent Sherlock Holmes titles. (Crimes and Punishments, Devil's Daughter, and Chapter One.) Has the Obra Dinn model has wrapped around to become identical to the old-style detective game?

Not entirely identical. You're still investigating purely static evidence; no live interactions with anybody. (The "suspects" are all drowned -- that's the whole point.) Everything you find is perfectly preserved at the bottom of the sea. It's not Obra Dinn's magic stopwatch but it might as well be.

But this does point out the similarities. Detective games love interrogating suspects, but most recent ones send you to the evidence board (or mind-map, or crazy-string-wall) at the end of the day. Obra Dinn was unique only in discarding everything but the evidence board.

Thalassa also takes the mind palace to its logical conclusion. The Holmes games put four or six questions on each chapter wrap-up board. Thalassa has almost sixty; the grid spans the entire game, from initial nibbles to the final grand conclusion.

It does its best to not overwhelm you, though. The game works hard to keep you on track. Large areas are gated on solving key puzzles, either by letting you find important tools (keys, a bolt-cutter, etc) or by outright telling you to solve the damn puzzle before you move on. The mind-map highlights puzzles you have sufficient evidence to solve; the terrain-map highlights rooms where there's evidence still to collect. And, as in all Obra-Dinners, brute-forcing the puzzle blanks is always an option. So you get a pretty smooth course through the game. There's basically no way to get seriously stuck.

I'm describing the dry (ha ha) bones of the mystery, which is unfair, really. This mystery is a psychological drama. (Not "psychological horror", but you go through some intense hallucinatory sequences.) The story is about grief, error, and trauma. As in Unsolved, everybody is concealing something -- and you know that it ended badly.

I thought the game missed a bet in not giving you a voice. You are "Cam", gender not specified, portrayed only in a blank-faced diving helmet. It makes sense that they didn't want to pin that down with a voice actor. But all the other characters are fully voiced (in recording, flashback, or your live buddy on the surface). So it feels strange to Gordon Freeman you.

Cam really is the protagonist of the story. It's their specific grief that inflects everything. To make them literally faceless and voiceless undercuts that. They wind up the ghost at the drowned banquet.

But then, this is a first-person game (from inside the diving helmet!) and that does a lot to ground things again.

I found myself regretting the last shreds of evidence because they meant I'd have to leave Thalassa. I rooted for everybody, even knowing that they would not survive. And I rooted for Cam to make it through to the end.

Good show; definitely worth playing.


Comments from Mastodon