I'm experimenting with some pen-and-paper role-playing involving characters with hidden agendas. To foster a proper sense of paranoia, I want these hidden agendas to be unknown to not only the other PCs, but the other players, too. A PC's hidden agenda is known only to the GM and their player.

One of ways the secret agenda can manifest in the game is through the PCs intentionally failing tasks that would contradict their hidden agenda. For example, an undercover agent in a terrorist organization could try to aim their shots wide of the intended target, pretending to just have a bad day with their aim. Mistakes happen!

I want a rolling system that allows my players to intentionally make "unfortunate" mistakes without automatically implicating them of such treachery to the others. Here's a list of criteria I'd like the system to fulfill:

  • As little overhead over normal rolling as possible: no pen and paper public key crypto!
  • The system must keep intentional failures indistinguishable from unintentional ones, at least between the non-GM players. Relying on suspicious gestures like note-passing or secret signals should preferably be avoided.
  • In particular, there mustn't be a way for a player to prove that their roll was honest, lest the players decide to require such proof from everyone after each roll.
  • Preferably, the system makes it possible for a player to cheat downwards only - failure can be arranged, but success still requires luck with the dice!
  • Preferably the players roll their own dice - it's not a huge deal, but getting to roll has a certain feel to it!

Does anyone have a system they have used or seen used that works given these criteria?

Note: Methods using cards and other randomization devices are also acceptable in lieu of dice, as long as the method proposed meets the standards outlined above.

  • 72
    \$\begingroup\$ This strikes me as a really good question--situation and criteria explained clearly--that's attracting a lot of really bad answers. "Here's this idea" with no discussion of how it worked in play, repetitions, &c. I've downvoted a lot of unsupported suggestions, and suggest further readers do the same. Back It Up! is a real thing. \$\endgroup\$
    – nitsua60
    Commented Feb 10, 2017 at 1:20
  • 11
    \$\begingroup\$ Yes, we expect answers to cite sources or experience, and encourage "here's an untried idea" answers to be downvoted. Consider rescinding or revamping your answer if it lacks this. \$\endgroup\$
    – mxyzplk
    Commented Feb 10, 2017 at 1:27
  • 7
    \$\begingroup\$ This question still continues to attract "try this thing I just thought up" answers. It would be a shame for it to get closed since there are about 3 good sourced answers. So what we'll start doing is if people flag an answer as Not An Answer because it fails to Back It Up!, we'll delete it. Consider this before adding an answer. \$\endgroup\$
    – mxyzplk
    Commented Feb 10, 2017 at 13:29
  • 6
    \$\begingroup\$ Several "untested idea" answers have been deleted due to multiple delete flags. If this has happened to you, feel free and edit your answer to Back It Up! and flag for undeletion. \$\endgroup\$
    – mxyzplk
    Commented Feb 10, 2017 at 19:27
  • 8
    \$\begingroup\$ Remember, it's up to everyone to ensure we can have nice things. \$\endgroup\$
    – mxyzplk
    Commented Feb 12, 2017 at 23:31

7 Answers 7


Secret Rolls

You could simply ask that the players roll all of their checks in secret, then tell the other players the result. The success of a lie depends on the player's bluffing skills. You can choose between announcing the difficulty of any roll (if your system has such a concept), allowing players to be sure that such-and-such a result will fail, and by how much. Alternatively, you could ask the player to announce their result without revealing the difficulty, forcing them to make an educated guess if they want to fail.

To make the rolls secret, players might:

  • Roll behind their hand, a piece of paper, or a 'player screen';
  • Roll in an opaque box, or rolling cup (such as those that come with perudo, and other dice games);
  • Use a dice-rolling app on their phone without showing other players the screen.

I did this for death saving throws in my last D&D 5e campaign. This allowed players who wanted to save their characters to lie when they failed. This could easily work the other way round. You would just need to have your players' word of honour that they wouldn't try to pass failures off as successes.

Policing Rolls

If players are rolling into a pot, you could ask that a player who has just rolled in the pot should pass the pot to you to check, before shaking the pot to obscure the result from the other players. The success of this depends on:

  • The number of players (more players makes passing the cup to you in secret more difficult for those further away);
  • The number of rolls (more rolls makes this a more tedious, and time consuming exercise);
  • Number of dice used per roll (more five in the cup are more likely to knock against each other);
  • Number of faces on dice used (more faces makes the die more likely to roll around during 'transit').

Essentially, in a game with frequent rolls (particularly d20 rolls, as in D&D, or rolls of many dice, as in Edge of the Empire), or many players, the system is likely to be tedious, and difficult. In a game with infrequent rolls of dice with fewer faces, and fewer players, the system could work effectively.

If players are using a dice-rolling app, the success of the system depends on the number of players, the number of rolls, and the players' willingness to hand their phone to you frequently.

If players are rolling behind something, you have to be willing to get up and come round to them for every roll. That would be very tedious in a game with frequent rolls.

Regardless, I still do not recommend the checking system, as I have found that it is always better to trust one's players. For example, as far as I can tell, no-one has yet cheated my death saving throw system, despite the possibility being there, probably because of the guilt involved in doing so.


You might want to take some inspiration from the game Paranoia.* In particular, instead of trying to avoid hidden notes, make note-passing between the GM and players a common and ubiquitous mechanic in your game, so that no single note stands out as unusually suspicious in any particular situation.

Combined with a mechanic where the GM is the final arbiter of rolls (i.e. players don't know the exact targets they're rolling against) you can easily support intentional failure simply by having the player pass a note to you in advance saying e.g. "please fail my next Electronics roll" or "please fail my next roll to shoot an NPC." Such a note doesn't necessarily need to be passed immediately before the roll it affects, reducing the potential for correlating the note with the outcome.

To make this work, you will need three things:

  1. First, as stated above, you need a mechanic that lets you, as the GM, apply hidden modifiers to a roll (or to do the roll itself secretly). If the players get to roll the dice and also know exactly what they're rolling against, then fudging a roll becomes all but impossible.

  2. Second, you need to let your players know in advance that they can do this. Since you will presumably be communicating privately with each of your players before the game, that's probably a good time to also explain the note-passing mechanic, and to suggest some examples of how their character might want to use it to advance and conceal their mission.

  3. Finally, you need to ensure that note-passing will indeed occur frequently enough at the table. One way to encourage this is to occasionally pass "dummy" notes to your players, and to suggest that they do the same. For example, if you notice that there aren't enough notes being passed around, you could hand a random player a note saying "please ignore this note," or even "write something on this note and return it before your next roll." With all players constantly writing and receiving notes, there's no way for anyone to know whether any particular note really means something, and if so, what.

Of course, you also need a setting where a certain amount of mistrust between players is to be expected, so that having such a note-passing mechanic in play won't by itself seem out of place to any players. That said, the scenario you describe, of a terrorist organization with a suspected infiltrator, would certainly seem more than suspicious enough to warrant some secret communication between you and your players.

*) Disclaimer: I haven't actually GM'ed Paranoia, only played it a couple of times. I may, however, just possibly have picked up a thing or two above my security lev&f(77DT687%&57/&% <SUBJECT TERMINATED. A NEW CLONE HAS BEEN DISPATCHED. THE COMPUTER IS YOUR FRIEND.>


In the past we have tried a few of those methods in roleplaying games. It always sucked. We never got it right. It always felt artificial and cumbersome.

Recently, my group (re)started playing boardgames, too. Some of them have very interesting mechanics so that you can have "traitors" in a cooperative game, although there is not even a neutral GM.

Battlestar Galactica was one of the games that really impressed me in that regard. It has a spaceships crew fighting for survival, where one or more of the players in reality are traitors that try to destroy the ship.

How this works is: Every turn, an "event" happens. This event has a goal (think a DC in D&D) and it requires means to reach said goal. Every player has cards of different colors and each card has a numerical value. Red for combat, blue for logistics, yellow for politics and some more (I think I missed exploration and another). So every event requires certain colors. In terms of roleplaying, an event might be a bandit ambush DC 17 that requires combat and exploration to beat. So all players put in cards face down to beat the DC 17 with combat or exploration cards. The "event" gets two random color and value cards added to the pot. Then the pot is shuffled (so you don't know who put in which card) and turned up. Card values with the right color are counted. Cards of other colors are counted against them. So every yellow card would subtract it's value from the total, because it's neither combat nor exploration. If the total beats the DC, the event is a success. If not, it's a failure.

The players are not allowed to talk about card values in absolutes. They are allowed general chat, like saying "I can do most of it, just need a little help", or "I cannot give cards, I don't have that color". And it's obvious how many cards someone gave, just the color and value is unknown.

There is a tremendous amount of ways how you can be a hidden traitor and look all innocent. You could put in a single card and it could be a high-value failure card. Or you could add 5 cards of correct color... but all are just low value. Or you could spend all your cards on a stupid unimportant event and then when it would count just go "well, I spend all my cards to help you last time, now I have no cards left in my hand, see?"

The point is that people get different colored cards each round. For RPG, the fighter might get combat and exploration, the bard might get social, magic and exploration, the cleric might get magic and social, the mage might get magic and combat. Now when you need combat and find a lot of social cards (that fail) you know it was either the bard or cleric. So you need to be careful not to show who you are by failing more than plausible by the two random cards.

All in all, it's a great system that you can easily use without a board for roleplaying games. The only drawback it has is that it does not feature those "lone" decisions or dicerolls. It's always the whole group that succeeds or fails a task. So you cannot say "Bob missed the bad guy by a mile", but it will always be "your party fails to capture the bad guy" and everybody will look to Bob as the main source of combat cards... and Bob will say "what, somebody dropped a high-value social, I cannot do it all alone guys!". And you will never know if that was the random event card, or the bard... or maybe Bob itself got hold of a social card and used it to fail.


This system is a combination of open rolls and hidden cards which I tried once and which worked quite well:

  • Each player has a deck of cards. The composition of the deck represents the character's competence. So players who play characters with specific strengths get more and better cards which apply to these strengths. An easy convention could be:
    • Club: Combat-related skills
    • Spade: Knowledge-related skills
    • Hearth: Social-related skills
    • Diamond: Dexterity-related skills
  • The player draws a hand from that deck. The rules explicitly forbid players to talk about what cards they have on their hand (because it's not character-knowledge).
  • When one makes a skill check, roll a D10. At the GMs discretion, add or subtract from it depending on difficulty of the task. For the task to be completed successfully, the player must then play an appropriate card from their hand which is lower than the result. When they can't or don't want to, the task is failed and they have to discard an unrelated card. In either case, the card is passed to the GM face-down, so the other players don't know if it was successful. The player then draws a new card from their deck.

These game mechanics had the following effects on the game which I found quite interesting:

  • Players can justify losing rolls by saying that they either didn't have a low enough card at that moment or had one but wanted to safe it for a more important situation. But that doesn't work without limits. If one rolled very high and still failed claiming not to have an appropriate card, it was quite likely that they failed intentionally.
  • It created suspicion: "Why did you say you want to try defusing that alarm when you don't have any good diamond cards on your hand right now?" "Why did you refuse to do it even though you have the dex-focused character? Do you really just have a bad hand or do you want to sabotage us intentionally?"
  • Pure psychological effect: Each player having a hidden hand of cards is a constant reminder that everyone has a hidden agenda and knows things the others don't.

Note that Ladifas's answer covers hidden rolls behind a player screen. This answer focuses on its implementation and potential issues.

We ran a few games like that. The solution was to have a screen, similar to the GM's one, for each player. That way all the dice rolls are hidden. This allows the player to roll dices and either take the "true" value or fudge it so it fails. No one is the wiser.

The added bonus is that the character's sheet is hidden as well. This also works for referring to the rule book(s) as no one knows who is spending way too much time looking at the "evil cleric spell lists" or whatnot.

A slight problem with this is that it covers the gaming table in screens leaving very little room for maps, miniatures, or snacks/drinks. If you have no gaming table, this is easier: rolls are done on the player's note pad and no one is allowed to look!

Preferably, the system makes it possible for a player to cheat downwards only - failure can be arranged, but success still requires luck with the dice!

This is where this method fails. We always role play with honest people.


Depends a bit on what system you're using, but for D&D and similar it works reasonably well to discuss it privately with each player ahead of time and set up a covert signal. A particular luck invocation, using a particular color of die, whether they roll overhand or underhand. Whatever they think they can keep the other players from noticing and you think you can remember to watch for.

Mechanics-wise, when they give their signal, swap the DC for performing the action with the DC you'd estimate for a "convincing" failure. To this end, it's helpful to keep a list of potentially relevant skills the player has such as bluff, acrobatics, perform, etc. The worse the roll, the easier it would be for someone to notice that the failure was intentional, with a bad enough roll making it obvious to anyone who's watching.

Edit: I'm not sure what all the downvotes are for. It's a simple enough way to handle it that doesn't interrupt the flow of the game and is hard for the other players to spot. I've used it in one "evil party" D&D campaign, and in several Paranoia games (Where stabbing other party members in the back is a major part of the fun and a simple "I want to fail this" signal saves a lot of passing suspicious notes.)

What works best for a signal will be highly table and DM specific. Different people are better at paying attention to different things.

I don't expect this will work nearly as well with "collaborative storytelling" type systems, In lots of those an abject failure ends up being suspicious in and of itself.

  • 1
    \$\begingroup\$ At my table, using a particular set of dice as a signal could work. Most players use multiple sets in order to swap if one becomes "unlucky", so changing dice is unremarkable. \$\endgroup\$
    – neontapir
    Commented May 16, 2018 at 15:33

Baked spindown d20s

In 5e we had a similar situation, the DM wanted to make an "advantage die" (there are many advantage die online but they always swap the 1 and 2 for a 19 and 20, which is lame and obvious). They bought a spindown d20 and loaded it by baking it in the oven at 90c for about 10 minutes. They bought another one, and didn't bake it. If they really wanted a roll to succeed, they could roll the cheated d20 out in the open and we'd be none the wiser.

This is a simple solution and elegant. Each player gets 2 spindown d20s. One has been baked with the 1 up and the 20 down, so it will roll lower than normal. Whenever they have to make a dice roll, they can pick which dice to use and make the roll.

  • Picking up the normal dice instead of the cheat dice requires milliseconds of overhead
  • Intentional failure cannot be deduced during play
  • There is no signalling required
  • The only way to prove the roll was honest is to make dozens of rolls to prove the dice is statistically fair or not
  • You can only cheat downwards
  • Players still roll their own dice

Be aware that this solution is best used on spindown die, since the high and low numbers are clustered. If you try this on a normal d20 then you will increase your chances of rolling a 20 most, but a 2 will also have an increased chance since it is right next to the 20. People have been cheating d6s with this method since forever, but I think spindown d20s are the best.

A similar solution would be possible with any other tools. For example if you want to draw cards from a deck, let the player have 2 decks, one of which is loaded with failures.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .