What is the best way to do search or spot tests (or similar) when knowing the outcome of the roll will colour the players' perception of the result?

I have encountered this problem in a number of different RPG systems that include such mechanics.

For example: a Paladin uses Sense Motive to detect lies from a monster. Knowing he rolled a 1 or a 20 changes their perception of what the referee says.

As a referee, I dislike doing the rolls secretly. Is there a better way?


10 Answers 10


In the case of your example, rolling a 1 or 20 doesn't mean any more than rolling a 2 or a 19 or a 3 or an 18. In 3.5 at least, there are no automatic successes or failures with skill checks. Regardless, in most rolls there is not an automatic success or failure. Sure, if they roll a 20 they're going to be confident that they have succeeded. But that's no different than in real life. Sometimes you're absolutely certain of something (I rolled a 20), and sometimes you have no idea (I rolled a 1). That doesn't mean I'm right. And neither should your players always be right. Yea, your paladin rolls a 20, he's pretty confident. Still doesn't mean he's right. And on those kinds of checks, I absolutely never tell my players if they succeeded or failed for certain.

  • "You're fairly confident that door isn't trapped"
  • "You don't have any idea if that monster intends to eat you"
  • "You're certain that the pit trap has been disabled. Oops, you were wrong."
  • 1
    \$\begingroup\$ This method, which is equivalent to "show the dice, hide the difficulty" is very good enough for variable-difficulty systems, like DnD and the like; but not for fixed-dificulty systems like BRP, GURPS, or Savage Worlds, where the roll is clearly a success or failure and modifiers are very rare. \$\endgroup\$
    – sergut
    Jun 26 '13 at 11:08
  • 1
    \$\begingroup\$ Your suggestion for how the dice roll relates to confidence and success seems to be the reverse of expectations. If I get a high total (including the roll), then I'm correct. If I get a low total (including the roll), them I'm wrong. As far as confidence goes, I would relate the information with high confidence at either extreme. A 1 means I'm really confident in my wrong information. A 20 means I'm really confident in my correct information. A 10 means I'm not sure about my okay information. How do modifiers interact with your plan? \$\endgroup\$
    – DCShannon
    Jun 27 '14 at 23:26
  • 1
    \$\begingroup\$ I can see your point on confidence, but I'd probably keep it the way I envision it on that point. Modifiers just add to or subtract from the level of confidence. \$\endgroup\$
    – BBlake
    Jun 29 '14 at 1:00

I typically introduce a red herring to make the players think they're rolling for something else.

Examples (from Shadowrun):

  • An enemy cyber-ninja wants to sneak up on the party, and plant a tracker on one of the group's vehicles, to follow their whereabouts. I tell the group to roll Perception checks (audio-based). None of them beat the cyber-ninja's Infiltration check. The sniper, Jeremias, rolls the best. I tell them: "Jeremias, you notice that it's unusually quiet in this part of the city. There's not as much ground traffic as you would expect." This makes it seem like they rolled well enough to get a hint, so I don't have to worry too much about them metagaming, thinking they missed something.

  • A malfunctioning machine sprite exists in virtual reality, on the node the hacker is connecting to. He needs to roll his Matrix Perception, to determine whether the sprite is detected. If he fails, I tell him that he doesn't notice any Intrusion Countermeasures, or hidden nodes.

  • 1
    \$\begingroup\$ That's similar to another question I asked, and I thnk the considerations are the same. In a game where the system let's you affect the final roll (Shadowrun let's you use edge for this) the player needs to understand what's at stake when they roll. \$\endgroup\$
    – Chuck Dee
    Oct 22 '11 at 21:30
  • \$\begingroup\$ This method is good enough for variable-difficulty systems, like Shadowrun, DnD and the like; but not for fixed-difficulty systems like BRP, GURPS, or Savage Worlds, where the roll is clearly a success or failure and modifiers are very rare. If your Call of Cthulhu investigator has a 20% in Search and she rolls a 39, you can try to introduce a red herring but the player will know that she failed. \$\endgroup\$
    – sergut
    Jun 26 '13 at 11:10

Get over your problem with making the roll secretly.

I know that's harsh, but here's my reasoning.

  1. You have two choices: either let the players in on information their characters do not have, or keep the players in the dark as their characters are.

  2. If the players are in the loop, you might as well let them roll the dice.

  3. If the players are in the dark, then they must rely on the GM to manage the hidden information.

  4. If the GM is managing the hidden information, then all dice schemes like "have each player write down a list of numbers" or "have the player roll without seeing her own result" boil down to a player trusting the GM anyway. Plus spooky dice superstition (a d20 rolled by a player is just as random as a d20 rolled by the GM, people).

So as GM, either let the player roll the check, roll it yourself in the open, or get comfortable rolling it in secret.


For "uncertain" rolls (ex: you search for hidden microphones in an espionage game, how can you be sure you found all that were hidden?) I use a double roll:

1 roll is made by the active player. Another is done, secretly, by the GM, using absolutely the same type of dice, bonuses etc.

If both roll a success, the GM gives full result (e.g.: lists all the bugs) If both fail, the GM gives completely false info (e.g.: "you don't find anything" while there is at least one working bug in the room) If one succeeds and the other fails, the GM gives partial information (e.g: "you find a bug behind a frame on the wall" - but there are two others and the player won't be informed).

This way, the player may be more or less confident of the result (i.e. if his roll was a success, he knows he is getting at least a partially correct information) but doesn't know "for sure".

I have used this technique for ages, and I think I read it in some old manual, possibly MegaTraveller, but not sure anymore.

One more thing: in the past, for example using Basic Roleplaying, I toyed a bit with the idea of making a translation table for "secret" rolls. I.e. I wrote a little computer program that "shuffled" an array of numbers from 1 to 100, and printed it out with the position alongside the original value... e.g.:

 01 - 78
 02 - 06
 03 - 18
 100 - 21

so when I needed the player to roll for something with an "unclear" outcome (classically a "spot hidden" roll to find a clue) I had him or her roll, and then used the scrambled result to adjudicate the situation... in the example above, a roll of 01 (which is considered usually a critical success) would be "mapped" to 78, which is pretty bad for an average character.

I understand that most modern systems use dice pools or some other "funky" dice mechanism so this may very well be impossible to do properly.

  • \$\begingroup\$ Nice if a bit complex. \$\endgroup\$ Nov 9 '10 at 11:53

One answer comes from Dogs in the Vineyard, where the GM is expected to make it very clear to the players that, for example, the person they're talking to is lying to them. This philosophy asks the players to help manage the story, and moves their knowledge into helping to construct the game world.

One answer comes from various ways of obfuscating the die roll, from a mapping of dice probabilities to letting another player roll and report the answer secretly.

A third answer comes from training the players away from paying attention to the die roll. By giving them immediate and tangible rewards (I personally like to award confidence points for "that was cool!" or "that was well roleplayed!" in Ars Magica) you can promote rolling with the results of the die roll. Functionally speaking, the benefit needs to outweigh the benefit they would get from being able to guess the accuracy of your statement and must be immediate otherwise there isn't a subconscious link between roll, good role-playing and reward.


This one is very good for settings like Cyberpunk where you may want to instill an atmosphere of paranoia.

Occasionally, ask your players to make dummy perception rolls, without any real cause. If they roll really well, be prepared to give them an interesting but irrelevant detail. (You may even introduce future adventure hooks this way)

This way, your players won't know the difference between plot specific and dummy rolls when they fail. They only know if they succeed.


In case you really dislike hidden rolls, you may also try introducing (after having discussed this with your players, of course) an automatically take ten rule for such situations. This takes away a good deal of the variety and excitement resulting from extreme rolls, ruthlessly smoothing check results to their average, but also quickens and eases (social) gameplay.

Just note down the concerned skills of the players in advance, add ten when you'd call for a roll and incorporate the result into your storytelling. :)

(In case it's not D&D you're playing, take the average of the dice you'd have them roll instead of ten.)

  • 1
    \$\begingroup\$ This is essentially what dnd4e did to solve this problem with passive perception and passive insight. \$\endgroup\$ Nov 4 '10 at 16:33
  • 1
    \$\begingroup\$ I haven't played/read DnD 4e, but I'm kinda glad they arrived at this too - it's quite logical, I think. :D \$\endgroup\$
    – OpaCitiZen
    Nov 4 '10 at 17:48

If you dislike hidden rolls just have each players roll a 100 times and list the results in order. Then make a hidden roll to mark where you start on the list and just cross off each result when used for a normally secret check.

This way it is still the player who made the roll but preserves the secrecy of the result.

It been my experience that the average gamer is on the borderline between able to separate in-game knowledge and out-of-game knowledge. For most if they know the results it will color their response.

The ability to act as a separate persona is a skill that is present in gamers in different degree. I found this to be true whether I am at home, conventions, LARPS, etc. What ever makes it easier to make the difference irrelevant is a win in my book.


Trust your players.

They separate player knowledge from character knowledge all the time. If they fail the check, tell them the character didn't get a sense the monster was lying or tell them something way off base, like the character gets the sense that this monster is the victim of a transforming curse and is trying to give vague hints about it.

If you've got good players, you just got some great opportunities for fun created. I can't help you if you have bad players.


There is the same 5% chance to roll any given number on d20. Technically the rules say that "if 1, then..." or "if 20, then...". So think outside the box: turn it upside-down, i.e. 20 fails, 1 succeeds. If you need proof after-the-fact, turn any die to either high or low, and reveal it after the fact. Use the same trick when the check requires a percentage roll (e.g. rogue skills).

The players can react however they want to a 'near-success' (rolling a 19 when 20 is the target) or near-failure, but that's irrelevant numerology.

Be sure to let the players know that you're using this technique/mechanic, however. That maintains the uncertainty you seek.

(I find this to be much simpler and far easier to implement than the answer already chosen as best, above, involving elaborate rephrasing of results on a case-by-case basis. As GM I want less work, not more, so I can focus on the story and the fun. I disagree with the introduction of 'red herrings' solely due to random rolls, nor do I want to make still more rolls when unnecessary. Dice are my tools, not my masters.)

  • \$\begingroup\$ Just be sure to decide which result is auto-success and which is auto-failure before the die is cast. It's easy to fall into temptation and decide the outcome should change if it doesn't fit with your plans. \$\endgroup\$
    – GMJoe
    Feb 18 '13 at 6:28
  • \$\begingroup\$ The first time I read this, I thought you meant remap the whole set of numbers, like instead of just a d20, do (d20 + 10) mod 20. Now I'm not sure. Is that what you meant? Because that's awesome. \$\endgroup\$
    – DCShannon
    Jun 27 '14 at 23:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.