Deliberating on deduction mechanics

The Deduction Problem

I’ve been thinking recently about detective games, and in particular, the “deduction mechanic” — the gameplay system of linking “facts” together to produce a new and interesting conclusions that forward the story. It feels like a chestnut of game design right now, even though I’m not sure I’ve ever seen an example of it that really works.

I’ve also been playing the new Hercule Poirot game, and it’s a great example. I don’t want to pick on the game unduly since it’s also one of the best versions of this mechanic that I’ve seen, and the cases I’ve seen so far are smartly written with their clues nicely laid and paced.

But… the deduction mechanic is here, and it’s central, and my experience is that it blocks me as a player more often than it delights me.

Take the tutorial case, for example (with as few spoilers as I can). There is a crime presented, but it’s not quite as it appears. The first step in the solve is to discard the obvious idea as impossible, so you can start to consider different ways the crime could have happened. But there’s a literary allusion sprinkled into that early stage which, if you see it, gives you the a-ha! moment of how it must have taken place.

This is good writing! It’s very Poirot, too. But it leaves me, as a player, a little cold. Firstly, I still don’t actually know whodunnit, only how it might have been done — so I’m not finished, and the game couldn’t let me do a Cluedo style “I know!” even if it wanted to. But all the same, I have no way to tell the game what I’ve deduced. I know it, I’m certain of it, but I can’t express it.

(Meanwhile, to progress, I need to work through a series of fails that I’m now quite certain are fails, but I need to satisfy the machine that it’s okay for us to move on. In practice, this means linking together two obviously-linked facts, to generate the deduction that “it didn’t happen this way”. This is solid police-work but it doesn’t delight.)

The goal of the design is to give me that moment where the detective “pulls together” two seemingly disconnected things. The disappearing body in the castle tower + the design of a cafetière (Jonathan Creek); the fire in Mrs Inglethorpe’s room + the temperature of the day (Christie’s The Mysterious Affair at Styles, and Poirot’s debut).

But when Creek or Poirot does this pulling together, they’re having a moment of inspiration — the external voice, speaking to them! — not a moment of Holmesian “I tried the other combinations and this is what’s left” deduction. In the game, the interface can’t produce this moment because either the player’s got no clue what’s going on and has tried every thing, or else the inspiration has already happened inside the player’s head.

So what feeling are we hoping for? For the result of the deduction step in-game to be surprising and delightful would mean the player must have had not been expecting it, but in that case, the player must have found their way to it by brute-force so they’re likely to feel that they’ve cheated.

If, on the other, the player has figured something out because it’s obvious, then the deduction step is not exciting, because we already know it’s correct.

Both of these are lacking the core component of excitement which is tension. To generate tension, the player needs to be sure but not sure. They need to have had a realisation but still be wanting to test it, and the interface should provide them with a way to test it. For maximum excitement that test must be easy and fast to execute.

A Proposed Solution

Obra Dinn achieves this by the simplest mechanism, similar to that of Cluedo: you can only expression your conclusion in one format; it’s not interested in having you express the complexities of how and why you got there; the mid-points of your solution are irrelevant. Whether you feel clever or not is very much left up to you!

The ideal deduction game attempts to do something more detailed than this; something that can express mid-points; something that can generate multiple deductions per case rather than requiring Obra Dinn’s 50-simultaneous-games-of-Cluedo bloodbath.

Perhaps it’s about being more open to non-linearity. Playing the Poirot game, I had my moment of insight and immediately opened the interface to look for something I could connect together to tell the game, I see you! I didn’t find one but perhaps something could have been there; a super-deduction shortcut hidden within the main-path deductions. Her Story achieves something like this: when you have a leap of logic, it hopefully comes with a good search phrase for its otherwise stubbornly unhelpful detective interface.

Of course, while a broader interface opens up the possibilities for creativity and expression (the “ceiling”), with it the counter-problem of making the game harder for players who just don’t get it (the “floor”). Having multiple successful solution paths is only helpful for player who have an idea of what’s going on anyway. But this is exactly the tension our previous examples were lacking; and it’s a balancing problem. All games deal with balancing problems, employing hints, or grind, or cheats, or no-help-at-all, depending on their flavour.

For a deduction game, however, there seems a good solution to the floor problem already available: Holmes’ axiom that “once you have eliminated the impossible, whatever remains, however improbable, must be the truth.” This has to be the harder path, though: being dull isn’t enough to put players off. It has to be slow, costly, impossible to rinse and repeat at any kind of scale, while not being so punishing and miserable that players simply give up.

And to be truly expressive, the whole has to feel natural, not convoluted. No wonder no one’s cracked it yet!