Torture Should Not Work in Dungeons & Dragons

First let’s get one thing straight--we shouldn’t torture even if it is successful.  But is it really successful?

In Dungeons and Dragons games, the answer often is yes, very.  When you need information from a group of bad guys, the game develops its own rhythm of torture.  

Track them down.  Leave one alive. Interrogate. Intimidate. Interrogate. Torture. Interrogate.  Congratulations, Jack Bauer wanna-be, you’ve extracted the necessary information to proceed on your quest!  Please move your alignment yet another notch towards neutral evil and proceed until you encounter the next group of bad guys to torture.

But should torture and a successful intimidation check yield useful and actionable information?  Real life research says no.

Why Torture Doesn’t Work

Proposition 1: People say whatever they think will help end their torture.

In an article in Scientific American entitled, We’ve Known for 400 Years That Torture Doesn’t Work, the author provides this excellent example:

During the European witch craze the Duke of Brunswick in Germany invited two Jesuit scholars to oversee the Inquisition's use of torture to extract information from accused witches. “The Inquisitors are doing their duty. They are arresting only people who have been implicated by the confession of other witches,” the Jesuits reported. The duke was skeptical. Suspecting that people will say anything to stop the pain, he invited the Jesuits to join him at the local dungeon to witness a woman being stretched on a rack. “Now, woman, you are a confessed witch,” he began. “I suspect these two men of being warlocks. What do you say? Another turn of the rack, executioners.” The Jesuits couldn't believe what they heard next. “No, no!” the woman groaned. “You are quite right. I have often seen them at the Sabbat. They can turn themselves into goats, wolves and other animals.... Several witches have had children by them. One woman even had eight children whom these men fathered. The children had heads like toads and legs like spiders.” Turning to the flabbergasted Jesuits, the duke inquired, “Shall I put you to the torture until you confess?”

In short, people lie to end their torture.  This begs the question: can’t a PC with a high insight check detect their lies and force them to keep talking until they tell the truth?

Proposition 2: People are terrible at spotting lies.

According to an article on FiveThirtyEight entitled Why Humans Are Bad at Spotting Lies, we spot lies 54% of the time, little better than mere chance.  And those who are paid to spot lies aren’t any better. The article explains:

Take people whose job it is to professionally detect lies — judges, police officers, customs agents. Studies show they believe themselves to be better than chance at spotting liars. But the same studies show they aren’t, Alcock said. And that makes sense, he told me, because the feedback they get misleads them. Customs agents, for instance, correctly pull aside smugglers for searches just often enough to reinforce their sense of their own accuracy. But “they have no idea about the ones they didn’t search who got away,” Alcock said.

Full disclosure: my job falls under the above category of those paid to detect lies. And I don’t mean my job as GM trying to spot my players’ baloney--I mean the one the that pays the bills.  And I’ll add another angle to this issue that I have observed time and time again: the expressions and tells that people exhibit when telling lies are the same ones they exhibit when intimidated by authority.  When in the face of an authority with power over them, people appear nervous and unsure of what to say--not because they are lying--but because they are nervous and unsure of what to say! Here’s what an insight check might yield in a real-life interrogation situation:

Successful insight check: You determine he’s nervous.

Unsuccessful insight check: You determine he’s nervous.

The best way to get reliable information from people who view you as a threat is to earn their trust and analyze what they’re saying in the context of other information you have, such as clues you’ve turned up using an investigation check. Using insight to expose their secret motive likely will only expose your own unconscious prejudices.  

Wait, you say, we’re playing a fantasy game where there’s magic and superhuman abilities.  Sure in real life maybe people are overconfident and only 54% accurate. But my character’s got a +11 to insight, whereas ThePlanarDM must be stuck with a negative bonus to insight since he works for the government.  Plus, if my insight fails I’ll stick the bad guy in a zone of truth. And if he succeeds on that save, I’ll kill him, disguise myself to look like his friend, and use Speak with Dead. Heh, good luck resisting THAT, Unnamed Underling Number 7.  

Proposition 3: If Torturing Underlings Worked, Bosses Wouldn’t Share Sensitive Information

It’s like DND’s Heat Metal problem.  Heat Metal is a low-level spell that utterly debilitates anyone wearing metal armor, without a saving throw.  In a world where this spell exists, everyone should prepare countermeasures, such as not wearing metal armor or wearing armor that can be removed instantly by cutting a single strap.

Similarly, in a world with Zone of Truth and obscenely high insight scores, bosses will be rightfully paranoid about information leaks.  They will share only the minimum information necessary to accomplish their goals, and also spread misinformation to throw would-be interferers off track.  They will gate information so that each underling knows only enough to place their piece of the puzzle but remains ignorant of the other pieces. And they may make threats sufficiently severe and personal -- such as to underlings’ family members -- to buy their silence even if the underling faces death or torture.

In short, torture doesn’t work in the real world, and probably shouldn't be very effective in our fantasy worlds either.  

Unfortunately, DND’s Mechanics Encourage Torture

The ability check system In 5th Edition Dungeons & Dragons encourages torture.  Intimidation (Cha/Str) and Insight (Wis) are defined skills with objective results.  Sure these rolls are opposed by enemy deception or wisdom checks, but with common abilities such as Expertise, Bardic Inspiration, and Guidance, player bonuses quickly begin to dwarf underlings’ bonuses.  The game allows multiple PCs to roll checks, or to assist one individual performing a check, making it even more difficult for one enemy to resist a group of PCs.

One way DMs avoid this problem is by asking players to provide their ability score and skill bonuses, then performing the rolls in secret behind the screen.  The DM provides accurate or inaccurate information depending on the roll, and because the players cannot be sure which one they are receiving, they aren’t able to trust the reliability of the information they receive..

The problem with this solution is that it deprives players of two of their most important resources for interacting with the game.  First, the player made an investment by choosing certain skills and abilities (e.g., intimidation) at the expense of not choosing other useful mechanical advantages.  And second, the dice roll is the greatest currency of the DND session and whenever possible should be kept in the players’ hands. Furthermore, when they roll dice, players should be able to instantly and intuitively recognize and celebrate success or rue failure, as this adds the most excitement to the act of dice rolling.  If the DM rolls secret checks, doesn’t allow a player to make a check, or frequently doesn’t reward a high result with useful information, this is a slap in the face to players invested in those skills and in the power of the dice.

Unfortunately there’s no perfect solution.  But here are a few tips for handling torture that minimize these problems while forcing players to re-evaluate their modus operandi for gathering information.

Tips for Discouraging Torture

• • All insight checks made against creatures that have been intimidated or tortured are rolled with disadvantage.  People are likely to say anything they think their torturers want to hear, and people are bad at recognizing lies.  These facts should be reflected by the dice rolls.
  • Most underlings possess incorrect or misleading information.  If a group of underlings has no need to know sensitive information, none of them will know it.  If there’s a need for one member of a group to have sensitive information, randomly choose which one knows the information, then assign false or misleading information to all other underlings.  

• • Low level underlings say whatever they think will help them survive, and do so as soon as possible, preferably before being tortured.  If they have information, they will share it almost immediately. If they don’t, they will concoct stories to match whatever they think their torturer wants to hear. The point here is to prevent situations where, through escalating intimidation checks and torture, PCs gradually elicit more and more useful information, thus encouraging them to continue to torture prisoners.

• • Roll behind the screen for higher ranked subordinates.  If the PCs intimidate or torture an underling who clearly possesses a rank sufficiently high to grant him access to sensitive information, roll any opposed ability checks in secret behind the DM screen.  Do not tell the PCs if they have succeeded or failed. If the PCs lose the contest, the bad guy actively tries to trick or endanger the PCs. If the PCs win the contest, the bad guy is too cowed to try to trick the PCs and says nothing, or offers up the information.  This method suffers from the problems listed above, but by limiting it to encounters with more senior subordinates, I think it strikes a better balance of empowering the players and maintaining a degree of uncertainty.
  • Torture earns PCs a bad reputation.  Perhaps there’s a code of conduct among adventurers or questgivers, or the townspeople themselves have been subjected to torture or maltreatment, causing them to frown upon its use.  If the party gains a reputation for torturing prisoners, it becomes more difficult for them to find work in the future, and they may even be ridiculed or lectured by patrons or townsfolk for committing such barbaric and ineffective acts.
  • The PCs should encounter innocent but nervous NPCs.  If the only NPCs in your game who act nervous and suspicious around the PCs end up being nasty characters hiding dark secrets, PCs no longer will fear the risk of accidentally torturing the innocent.  Add some NPCs who are nervous around authority, and if the PCs end up torturing one, bring in other NPCs or evidence to prove their innocence. The PCs may choose to be more cautious moving forward.

Conclusion

Torture doesn’t work in real life, but maintaining verisimilitude is not the only reason we should try to keep torture out of our games.  It’s also morally repulsive. And, frankly, it’s often boring. If the players are driven to adopt a standard mode of operations that they apply to all situations, that’s bad design.  And that falls on us as the DMs.

If DMs consistently reward the same strategies, players, like the Pavlov’s Rats that they are, become conditioned to press the torture lever and expect to receive more cheese. But what if instead of cheese arriving on a little mousey platter they get it all over their face?  Or if the trail of cheese instead leads right into the cat’s clutches?

This analogy has become hopelessly tortured, pun intended, but the point is that we can encourage players to adopt more creative and productive ways of gathering information by making slight changes to gameplay mechanics and how the world around them reacts to the PCs’ behavior.

If you like this article, please share it with your friends or let me know on twitter @ThePlanarDM.