PILLAR 1: DEDUCTIONS MUST BE VALIDATED
This requires that, after investigating, the player must be able to input their deduction. This inputting can be explicit, such as typing in an answer and having the computer evaluate it (like Her Story), or implicit, such as walking their character to a location they’ve deduced might hold treasure and seeing if treasure is present there (like Red Dead Redemption 2’s treasure maps).
There’s no need to validate every little deduction possible during an investigation, but generally, most plot-consequential deductions should be validated. Forgoing validation in the case of a consequential deduction is risky, but can be used for artistic effect, i.e. major mysteries that you want the player to never feel certain about, such as Heaven’s Vault’s ancient history.
Validation doesn’t always require success or failure states. It’s enough that the computer acknowledges and responds in some way to the deduction, it doesn’t always need to actually judge it as true or false; ambiguity can be much juicier (like Paradise Killer’s trial).
Most importantly, the game must not validate deductions before the player inputs them. For example, just because a player has discovered a city map (A) and street address (B), that doesn’t mean they’ve paid attention and know which building to go to (C). Giving hints or other subtle support is fine, but the game should never, on the player finding A and B, direct them to C.
PILLAR 2: PLAYERS MUST FEEL OWNERSHIP OF DEDUCTIONS
This requires that the player has come up with their deduction without feeling overly influenced by the game’s content or design. In reality, this can be an illusion and the designer can have sneakily guided them quite substantially, but they must feel independent. Players can be over-influenced by content if, for example, an in-game character often blurts out answers or hints that are too helpful, such as Lara hinting at puzzle solutions in Shadow of the Tomb Raider’s Easy mode. This robs the player of the feeling that they’ve made the deduction themselves.
Players can also be over-influenced by a game’s design. For example, if I have to deduce which of many NPCs is an assassin, and all are generic except for one who has a unique character model, then the game’s design has likely given away the answer, again robbing the player of ownership. Offered the chance, many players will ignore the in-fiction clues and instead try to second-guess the designer, so unless your game breaks the fourth wall like Metal Gear Solid, avoid this at all costs.
On this point about design, the most important factor in a player feeling ownership of a deduction is the size of the possibility space. The possibility space is the sum total breadth of all possible answers the player could input, regardless of correctness. For example, if the player needs to deduce which of four items they must show to a character in order to get a confession, the possibility space is small: just four possibilities (see: interrogating witnesses in Ace Attorney). If a player must deduce which English words to type into a database search in order to find a confession, the possibility space is nearly infinite (see: Her Story). Selecting a specific answer from a large possibility space makes it feel personal.
PILLAR 3: INVESTIGATION IS RESEARCH
This last pillar simply means that investigative deduction, instead of a hot-seat quiz, is the act of exploring a body of information, eliminating answers, and testing hypotheses before finally arriving at a conclusion.
Generally speaking, having a high cost associated with the player making an incorrect deduction is a bad idea. The player having wasted a small amount of time or resources is often enough to dissuade them from attempting to brute-force their way through a deduction puzzle, and often the player needs to input incorrect solutions in order to test hypotheses on the way to their final conclusion. Harsh punishments for incorrect deductions, like Ace Attorney’s ‘three strikes and you’re out’ system, are often a crutch used in games with very small possibility spaces, and are frustrating at best.
If for narrative reasons your player only has one chance at a solution to a particular deduction, then rewarding an incorrect deduction with a slightly worse outcome is vastly preferable to an outright fail-state. For example, if the player accuses the wrong suspect, it’s much better to have that character simply be offended than require the player retry the whole scene.
APPLYING THE PILLARS
These principles are, in my opinion, simple to apply to your own designs. First, players should have a method of inputting deductions for the computer to then respond to. The game should feature this input stage prominently if it wants to promote investigation or deduction, and should never validate deductions before the player has inputted them.
Second, the subtler the hints and the broader the possibility space, the more ownership the player feels of their answer, and the smarter they feel for getting it right. Finally, and simply, if an investigation is ‘testing’ the player’s deductive skills, think of it as a graded open-book challenge rather than a pass-fail exam.
Investigative play encourages the player to really engage with narrative content. They must stop, think, and understand the finer details of what you’ve written. They aren’t begrudgingly forced to slow down and pay greater attention, they’re made to want to. I’ve recently been playing Divinity: Original Sin II, and had begun glazing over text in favour of following map markers. But 60 hours in, I finally taught myself the names and identities of the very plot-relevant gods. Why? Because I wanted to solve an optional riddle involving their statues. Surely any story experience would benefit from increased engagement.