November 1, 2019

The Sprawl Session Recaps

For those interested in Actual Play for my campaign of The Sprawl, a Powered by the Apocalypse cyberpunk RPG by Hamish Cameron, I've made a landing page for them.  The landing page has setting, PC, and corp summaries to kick this series off.  I plan to aggregate each chapter there.

September 25, 2018

Do Split the Party

Most RPGs can handle "splitting the party" decently well.  The problem with splitting the party is that players get bored when they're not actually playing the game.

It's one thing to wait your turn in combat, where you are part of the action -- especially if the GM is highlighting the stakes and context of the situation when it's not your turn (see the twitter thread below).  It's another thing to wait a long time while the other PCs are off scouting or investigating or negotiating.

You can tell GMs to cut frequently.  You can tell GMs what to avoid, how to try to match cut speed to pace, etc.  But what I've discovered in the last three years is that as a GM, you need to learn how to recognize triggers that cue you to cut.

If you're not reminded to cut back to the other players, you might not realize you've gone on too long.  You might not realize you're boring them.

So here are some triggers to remind you to cut.  If you internalize these eight scene cut-away triggers, you'll get better at running split-party scenes without boring your players to death.

Failing a Skill Check:  When a PC fails a skill check, cut!  Cut to the other PCs immediately.  This has a lot of great benefits!  First of all, depending on your system, failed checks will happen decently often -- especially when the party is split.  In D&D, it might happen every three or four rolls.  That's about the same for PbtA games.  Second, it gives you time to think about the move you want to make. With extra time, you can think of a really good complication that really adds to the tension. Third, the player who failed the check will be waiting with bated breath to hear how bad things went with that roll.

A Player Needs OOC Time: This one is obvious - if you're running for a split party, and a player needs time, it's time to cut.  Players might need time for lots of reasons:  OOC things like getting another piece of pizza, using the restroom, taking a call, fixing a tech issue (playing online), or having a sneezing fit.  Players usually won't take a break in a tense moment, so if they signal a need to break, it's at a lull, and a good time to cut.

A Player Needs Time to Think:  Players might also need time to think of a plan, think of what to say, think of how their character would react, make a tough choice, or figure out a character ability.  If you're going to force them to make a decision under pressure, don't cut away.  Apply that pressure.  Talk them through it.

Aside: When you force a player to make a decision in a split second, you're testing the player, not the character.  For some styles of play, this is great.  Actor stance play, such as LARPs, horror games, or high character immersion play can be enhanced by forcing players to make split second decisions in character.  You're encouraging bleed (see here, here, and here).  For other styles of play, this is bad.  Author and director stance play should test the character, not the player.  You might spend ten minutes thinking about how your character would handle a split-second high-stakes decision. In those styles of play, it's often fun to decide that your character made a bad decision. 

There's a Rules Question:  If the table runs into a rules question, cut.  The players can look up the rule while you run the other scene.  This one should be obvious. Just make sure to associate it with a cut-away trigger in your mind.

Cliffhanger Moment: When something surprising happens, cut away after you see the players' reaction.  Don't drop the surprise and cut immediately.  Why?  Because when you get the players' reaction then cut, you get five or ten extra minutes to think about how to play to it!  If you surprise them and cut away immediately, you don't get their reaction until after you cut back.  And that means you have only seconds to plan how to play to their reaction.  Put another way:  The surprise is big.  The players' reaction to the surprise is even bigger.  Your response to their reaction is the third most important thing that happens at a plot twist, and giving yourself extra time to plan that is gold.

GM: "Jasen, you've put down the third vampire spawn.  You're wounded, exhausted, but victorious.  You walk out of the alley, back into the crowd.  Nobody noticed the battle.  Standing there, staring at you is the vampire who commanded the spawn to attack.  She pulls her hood back, and it's none other than Lucia, your former mentor." 
Jasen's Player:  "Holy crap!  I thought she died in the crusade!  I don't care about the crowd. I don't care that I'm wounded.  I charge.  Do I roll initiative or what?" 
GM: "Hold that thought.  Let's cut away."

A Conversation Milestone: Conversations take longer than you think.  When you're GMing a conversation, you take on the NPC's persona, and start thinking about what the NPC wants, what they're afraid of, what they know, what they're watching for, etc.  Often, that means you stop focusing on a lot of the logistics of gamerunning.  You lose track of time, lose track of players who aren't in the conversation, etc.  So teach yourself to cut away when the conversation reaches a milestone.  That is, cut away when something new comes up; a decision is announced; the mood changes; or parties enter or leave the conversation.  Just teach yourself to watch for conversation milestones, as shorthand for that.

Plan B Doesn't Work:  The PC tries one approach.  They don't get the result they wanted.  The PC tries another approach.  They still don't get what they wanted.  It's OK to run a scene where a PC fumbles around a little.  The dice sometimes force that on us.  Players also sometimes don't know exactly what they're after - they go into a scene and just push buttons (literally or metaphorically) until something happens.  That's fine, sometimes.  But while it can be frustrating to the PC who's flailing, it's extremely frustrating to watch.  So teach yourself to cut away when the player's second approach doesn't go anywhere.  This has two benefits:  The other players don't have to sit through more than two false starts in a row, and the active player gets a few minutes to think up a better strategy.

A Clue is Revealed:  When you reveal a clue that's a "piece of the puzzle," cut away.  Unlike a cliffhanger, you don't need to see the player's reaction to a clue.  Most clues are just useful information, not major changes to the conflict.  (If the clue is a cliffhanger, see above.)  Cutting away right after dropping a clue will save you a ton of table time.  First of all, the players who just got the clue need a few minutes to process it.  They have to think about how it fits into their investigation, what it means, and what follow-up questions they need to ask.  When you reveal a clue, the players often ask a lot of confirmation questions -- stuff they already know, but just want to be sure about.  If you give them a few minutes, when you cut back, they'll have cut that down to only the most important follow-up questions.

Handy Infographic Version

Here's a handy infographic you can share if you're so inclined.

Click for a larger version.

Other Run a Game articles on splitting the party

I've written before about the benefits of splitting the party, which is still pretty good, though the game I used for the example is now an edition out of date!

I also did an article on cutting between scenes before, but I think my skills have evolved since, and I've also figured out how to communicate what I've learned in the last three years, since the last time I wrote on this topic.  For instance, in the older article, I recommend 15 minutes between cuts.  Now I'd say 10 minutes is pushing it, and you should aim to cut every 5 minutes, if you can.

Update:  Use these skills even when you're NOT splitting the party!

See this twitter thread about it. (Click through to see the full thread from here.)

September 11, 2018

Ye Olde Magic Item Shoppe

Let's say you're running a fantasy RPG, and you want a more serious fantasy tone.  The first thing you want to eliminate is "Ye Olde Magic Item Shoppe."  It's a silly thing that seems to come from video games, not fantasy literature or historical epics.  But how?

The obvious answer is to tell the players "there aren't magic item shops in this setting."  The problem with that approach is that the players will eventually be dripping with magic items they don't care about, and yearning for magic items that they still haven't found.

They'll look for low-key magic item shops.  "Hey, can we 'donate' these +1 maces and axes to the high temple of the sun god and ask the priests to forge me a magic glaive?"

So the second most obvious answer is to make magic item shops that don't resemble a JRPG or MMO.  You create a red dragon that collects magic items, and will buy them from adventurers for gold (which she extorts from kings and merchants).  You create a shadowy wizard that will sell knowledge (spells and scrolls) for gold to fund his secret experiments.  You create a good-aligned temple that will forge blessed weapons and armor for those who demonstrate their faith (with deeds, yes, but also coin).

But you still haven't eliminated "Ye Olde Magic Item Shoppe."  You've dressed it up to look a little bit like fantasy fiction, but it's still a transaction of magic items for coin and vice versa.  Because the player activity is functionally the same (tallying coins, asking for prices, deciding how much to spend and what to sell), the fictional activity will largely feel the same.

The best solution

The only way to eliminate "Ye Olde Magic Item Shoppe" is to give the PCs the magic items they want, and only the magic items they want.

The only reason a PC would want to sell a magic item is if that magic item isn't useful to them.  That happens when you give out magic items because you wanted to equip your villains with them (but didn't think of the items' utility for the PCs) and when you give out magic items based on the random tables in the DMG.  The only reason they would want to buy magic items is if they've got a lot of gold (happens a lot in 5e) and feel like the items they want should be available.  Combined, the two factors really make players seek out magic item sellers:

"We have two +1 maces, a +2 sickle of evil, and a +1 heavy crossbow that none of us are using, and I still don't have a magic greatsword yet.  Let's sell these useless things and get the +1 greatsword I need. +1 weapons seem common enough that someone must be selling one, or maybe I can get someone to forge me one."

Giving the PCs exactly what they want doesn't mean you have to be generous with magic items.  You can be more stingy than usual with this technique and the players will probably be happier.  Here's how you do it...

First, call for a wish list

Ask the players to submit a "wish list" of five magic items they want.  Let them flip through the books like kids making their Christmas wish list from the toy store catalog.  They can even make up their own magic items.  If your campaign is going to be shorter or longer, ask for shorter or longer wish lists.  A three-year level 1-20 campaign might call for six or seven per PC.  A 10-session short campaign might only call for two per PC.  If you're running an intentionally low-powered, low-level, short campaign, you might also want to limit the players to Uncommon and Rare items.

When you call for the wish list, show your players this article, so it's clear what you're doing.  If you're transparent with the process, they can be more strategic with their choices.  For instance, they might see value in asking for a Rare Flametongue greatsword and a Legendary Vorpal greatsword, so they get a middling-powered magic weapon early, and get an upgrade to a super-powered one much later in the game.  They can always give the Flametongue to a henchman or beloved NPC.

From that, make a magic item treasure table

Combine the lists into one "treasure table" and arrange them from weakest to strongest.  Always consider defense items to be stronger than offense items, because offense items are more fun (they speed up play and provide wow moments).  This makes a list of 15-30 items.

Using the table, decide on some plot items

Some of your magic items shouldn't be random.  Make plots and villains to contain the best items from the combined treasure table. Cross them off as you place them in the world. This probably cuts your list down to 10-20. The Lich King should wield the Staff of the Magi your wizard PC wished for.  The Glabrezu should have the Holy Avenger sealed away in a trapped vault.  The real nice Ioun Stone should be rumored to be at the top of the ruined tower of the mad mage, deep in a troll-haunted swamp. Don't write these adventures ahead of time. Just make sure to tell the PCs the legends of where they can find these items.  It'll motivate them.

Then just use your table

When you give out treasure and a magic item should be in the hoard, choose or roll from the list.  If you roll from the list, only roll 1d6 and count up from the bottom, skipping crossed off items, of course.  When you give an item out, cross it off.  The reason to use a smaller die than the list is that you want to give out more modest items first.  It's only fair:  PCs with more modest wishes get to use less powerful items longer (since they get them earlier).  Plus, you don't want to drop a Staff of the Magi at level 1.  I like to give out fewer, more powerful items, but that's going too far!

Tip:  Since you're giving out fewer permanent magic items, consider giving out more consumable items -- especially consumable restorative magic items, like potions of healing, Keoghtom's ointment, potions of neutralize poison, scrolls of remove curse, and so forth.  When you're giving out gold and mundane items, here's an inspirational, curated list of select interesting mundane items by value for you that will help give you a little inspiration.

Is "Ye Olde Magic Item Shoppe" really a problem?

Not always.

I've claimed that D&D is its own subgenre of fantasy.  It's a goofy power fantasy, and it can be a lot of fun to play up the... D&D-ness of it, even if you're playing Dungeon World or 13th Age or Pathfinder.  So much of the D&D system intrudes into the fiction of the game world that D&D almost has to be its own genre.  Daily refresh abilities, magic item tables, trap mechanics, Vancian casting, and encumbrance (and therefore henchmen) are system artifacts that create in-fiction shadows that have, over time, made their own culture -- their own fictional genre.  And that's fun because it creates a culture and genre that's unique to our hobby.

This concept ties to the idea of "tone" or "mood" in your game.  Setting the game's tone is a very important "session zero" task, and it's important for the whole group to work together to maintain the tone.  The GM's role in setting and maintaining the tone is even more important.  If you create a "Ye Olde Magic Item Shoppe" situation, you're saying something about the tone, and what you're saying is tied to images of moogles and Azeroth, and that can be awesome or jarring, depending on the tone you're going for.

Image Credit: CC0 license from via

So when you're running a very... "D&D" game (like I am right now), play it up!  Make fun magic item shops run by crafty dragons and mad priests and shady wizards and extra-planar entrepreneurs.

But you might want to run a D&D game that's more like fantasy novels, movies, and TV shows.  And that's fine too.  That's when you need to use the techniques above to eliminate magic item shops.

August 14, 2018

Encounter Stakes

Too many GMs hammer the party with encounter after encounter of "kill or be killed" life-or-death fights to survive.  There are several reasons why this is a problem.

First, it's toothless:  Either you kill a PC every other session, or else your "kill or be killed" encounters are mostly harmless.  Even if you kill a PC every other session, the death risk in any given encounter is probably one in five or one in ten.  Not insignificant, sure, but hardly dire.

Second, it's tiresome:  If every hostile creature you meet turns out to want to kill you or die trying, it gets dull.  More, "kill or be killed" encounters tend to drag.  After about 2 or 3 rounds, it's clear that the PCs have won, and the monsters are just trying to make their deaths as costly as possible.  Once every now and then, that's interesting.  Every time?  Gets boring.

Third, it leads to murder hobos:  If every encounter eventually ends in grim slaughter, whenever a conflict arises, you're going to go straight to grim slaughter as a solution.  The minute anyone cracks wise or threatens your PCs, they're going to go straight for the most efficient kill.

The solution is, luckily, not all that hard.  Just vary the stakes of the encounter.  Here's a big list of encounter stakes that are not "kill or be killed."

Stakes Progression

I've divided these examples into four tiers.  Start with low stakes.  As your adventure progresses, keep raising the stakes.

A lot of stakes come with built in progression:  If the PCs are framed (level 1), they might be at risk for capture (level 4), if the frame-up is successful.  A frame-up is only level 1 because it doesn't lead to the PCs' capture, it leads to a risk they might be captured, if they can't clear their name.

For the lower tier stakes to qualify for their lower level, the PCs have to have a chance to avoid the risk posed by the follow-on stakes:  If the PCs are delayed (level 2), they must still have a chance to prevent their rival from snatching the thing they wanted to get (level 3).  If the delay leads to the snatch without any chance the PCs could have stopped it, then the delay was really a complicated snatch, not a delay.

Level 1 Stakes: Social or Emotional

Sticks and stones may break my bones, but words will never hurt me.  

First level stakes don't cost the PCs much except their reputation or their good mood.  Getting humiliated or spooked or tricked might ruin your day, or let an enemy get away with crimes or escape capture, but they won't cost you anything and they won't hurt.

  • Humiliate:  The foes win if the PCs feel humiliated
  • Embarrass:  The foes win if the PCs do something embarrassing
  • Reputation: The foes win if the PCs' reputation suffers
  • Enrage: The foes win if the PCs get mad at them
  • Censure: The foes win if the PCs suffer a superior's disapproval
  • Framed: The foes win if the PCs are suspected of a crime they did not commit
  • Count Coup:  Each foe wins if they touch the PCs without getting hurt
  • Scare: The foes win if the PCs flinch (take a defensive or restorative action)
  • Spook: The foes win if the PCs regroup, retreat, or begin acting more cautiously
  • Threaten: The foes win if they take the foes and their faction more seriously
  • Bluff:  The foes win if the PCs believe the bluff
  • Reprisal:  The foes win if the PCs are worried of additional reprisals

Level 2 Stakes: Material or Tactical

Stand and deliver!

If the stakes threaten to cost the PCs resources, they rise to the second level.  Second level stakes can also threaten the PCs' tactical position, raising the stakes they might encounter in the future.
  • Steal:  The foes win if they take stuff from the PCs by stealth, threats, or force
  • Break: The foes win if they break or spoil the PCs' stuff
  • Deplete:  The foes win if they get the PCs to use up limited resources
  • Foist:  The foes win if they make the PCs take on stuff they don't want to carry
  • Block:  The foes win if the PCs don't take the guarded path
  • Oust: The foes win if they force the PCs to leave an area
  • Divert:  The foes win if they force the PCs to take the selected path 
  • Feint:  The foes win if the PCs react to the feint
  • Distract:  The foes win if they get the PCs to engage with them for long enough
  • Delay:  The foes win if the PCs take a few rounds, a minute, an hour, or a day longer
  • Alarm: The foes win if they get warning to their allies
  • Pay: The foes win if they make the PCs pay more than they had to through trickery
  • Sell:  The foes win if the PCs buy what they're selling
  • Beg:  The foes win if the PCs give them charity
  • Extort: The foes win if the PCs pay them a bribe or blackmail money
  • Split the Party: The foes win if the PCs become separated

Level 3 Stakes: Goals and Bonds

There are fates worse than death...

These are goals or people or places or things that might mean more to the PCs than their very lives.  Would you die to protect your community?  Your family?  Would you risk your life to pull strangers from a burning building?  These are character defining questions, and level 3 stakes help us get to them in ways that level 4 stakes do not.
  • In Decline:  Harm or take over an organization they care about
  • Lost Friend: Harm, beguile, or alienate a person they care about
  • Special: Harm or take a thing they care about
  • Noise: The foes win if the folks that matter don't know who to believe
  • Homewrecker: Harm, control, or bar entry to a place they care about
  • Escape:  The foes win if they escape justice that the PCs want to mete out
  • Competition: Claim an opportunity that could have helped an organization they care about
  • Rival: Claim an opportunity that a person they care about wanted
  • Snatch: Claim a thing they wanted to get
  • Outbid: Claim an opportunity that would have helped a place they care about
  • Demoted: The foes win if the PCs lose formal status
  • Divide:  The foes win if the PCs become unjustly suspicious of their ally
  • Lost: The foes win if the PCs get lost

Level 4 Stakes: Personal and Physical

Take no prisoners!

Stakes that are direct attacks on the PCs' bodies are the highest of all, but because they're so direct, they're often very blunt, unrevealing situations.  Of course you're going to fight to defend yourself.  Of course you care about being locked in a dungeon.  On the other hand, they're tense, exciting moments that can feel terrifying or exhilarating -- usually both!

  • Hurt: The foes win if they harm one of them one of the PCs in particular
  • Maim: the foes win if they cause a specific injury to one of the PCs in particular
  • Assassinate: The foes win if they kill one of the PCs in particular
  • Guerrilla: The foes will try to kill the PCs, but will retreat to avoid any casualties
  • Surrender: The foes win if the PCs surrender
  • Capture: The foes win if they capture or arrest one or all of them
  • Consume:  The foes win if they successfully eat part of all of one of the PCs
  • Infect: The foes win if they cause one or more PCs to contract a disease
  • Envenom: The foes win if they poison one or more of the PCs

More About Encounter Stakes

Foreshadowed Stakes vs. Surprise Stakes

Foreshadowed stakes are stakes the PCs know about well in advance.  For instance, they might know that Armlor the Brewer is looking for them to chew them out.  That tells them that there's someone wandering around town looking to cause them some reputation or emotional harm (Level 1 stakes, emotional or status).  They know ahead of time, so they're anticipating it.  In effect, you've already levied the stakes at them.  The stakes are real, even if they haven't met Armlor yet.

Foreshadowed stakes are the best because the players experience them for a longer period of time, and their characters can start engaging with them well before the encounter ("well if Armlor comes by here, you can tell him we'll meet him at sunset outside our inn, if he's got the guts").

Surprise stakes are fun because there's an element of the unexpected. Surprise stakes can be...

  • New stakes out of nowhere:  On the way to the inn, the PCs are attacked by robbers (Level 2 stakes, extort)
  • Significantly changed stakes: Arriving at sunset to discover that Armlor is there helping the owners try to put out a raging fire in the inn (Level 2 stakes, break their stuff) or arriving to find Armlor's fresh, bleeding corpse (Level 1 stakes, framed)
  • Surprisingly increased stakes -- Armlor shows up at the tavern with a cadre of Duke's soldiers to arrest them (Level 4 stakes, capture)

Surprise stakes are the best because everyone loves a twist.  But you can't make every encounter a twist.  Try to use a twist every couple of scenes, though!

Play to Find Out

To make stakes work, think of them this way:  You're playing out this encounter to see if the foes will win their stakes.  Therefore, all the stakes examples, below, are phrased as "the foes win if..." to remind you that these NPCs are done when they achieve their stakes.

Level of Stakes vs. Probability of Loss (aka Challenge)

Challenge matters.  Consider how likely it is that the PCs lose in the contest for the stakes.  For instance, low stakes (humiliation) with high probability of loss (the PCs will almost certainly be humiliated) can be very powerful.  High stakes (assassinate) with low probability of loss (the PCs can easily defeat the assassin) can be very weak.

Higher challenge raises the stakes, but it almost never raises the stakes up a whole level.  Humiliation can be really painful, but losing a fortune, losing a friend, or losing an arm is a lot more painful.

PC Stakes

Your players are going to set their own stakes, based on what's going on in the fiction.  If an NPC gets in their face with threats and bluster, they PCs might decided to shut the NPC down emotionally or to beat them up, or even to kill them.

You don't get to control the PCs and what they decide to do.  Their stakes are their business.  Your job is to control the NPCs.  The PCs' actions might trigger new stakes, though.  If they kill an NPC who's yelling in their faces, they might be wanted for murder.  The stakes go from humiliation (level 1) to capture (level 3) as the town militia is called up to hunt them down for trial.

What do the Foes do when they Win?

Most of these stakes end long before one side or the other is dead.  You, the GM, get to decide if the NPCs have won their stakes.  Once they've won their stakes, they should act naturally.  Typically, they'll just leave.

Does this mean you're going to re-use the encounter later?  You bet you will!

Isn't that boring?  Heck, no!  Players love to see NPCs they've met before.

Will encounters combine?  That is, if the guard goblins succeed at raising an alarm and run away, will they join with other goblins and make a really Deadly encounter later?  No.  I mean, you could do that, but you're creating a strong incentive for your players to kill everything they meet, in case they have to fight it later.  There is an enormous conceptual difference between "get a chance for revenge when you meet the same NPCs again" and "any NPC you don't kill might join with another encounter and make your life harder."

Printable Infographic Version

Click here to download this at a readable size
Here's a 8.5x11" printable list of these stakes, to put in your adventure prep inspiration kit.

July 19, 2018

How to Run an RPG Campaign in 5 Easy Steps

If you want to run an RPG campaign, you ought to do it right.  Here's a simple five step process to run a game that your players will never forget.

Five Steps to a Memorable RPG Campaign

Step 1: The Pitch
Pitch a campaign idea with enough detail that everyone understands the vision (genre, tone, themes, setting, conflicts, main action). This is a conversation, not a dictum - it's their game, too. Make sure everyone is on the same page, you included. (Here's a really old article from this site on making a campaign pitch.)

Step 2: Character Creation
Accept characters that fit the table's shared vision (see #1), and can work together.  (It's probably best if they already know each other).  Make sure all the characters have things they're intensely passionate about - people, places, things, goals, groups - that fit the campaign vision.  Many RPGs have passions baked into character creation. GUMSHOE games ask you to list your sources of stability. 5th edition D&D asks you to describe your Ideal and a personal Bond, etc.  Work within this structure where you can, but make sure to push players to give you real good passions - not cop outs.

Step 3: World Building
Sketch the world roughly with lots of blanks. In it, create major antagonists that have goals that brutally conflict with the PCs' passions (see #2). Give your antagonists stuff: People (henchmen, goons), places (dungeons, cities, lairs, etc.), things (artifacts, rituals), groups (titles, influence, cults, factions), and knowledge (of the PCs, of the future, of the past, of how things work).

Step 4: Starting Setting
Fill in the space close to the PCs in much more detail. This is your "starting village" -- your Tatooine or Emond's Field. Even though I said "much more detail," you should still leave some blanks to fill in as you go. As you fill in, fill it with the stuff the PCs care about (see #2) and the antagonists' stuff (see #3) - especially at least one henchman.

Step 5: Inciting Event
Decide what the local henchmen are up to that will damage the nearby stuff the PCs care about (see #4) and what happens to tip the PCs off to what's going on in time to do something about it (inciting event). Drop the inciting event, then just respond to their actions.

If you followed these steps, the PCs should care intensely about what's going on, because what's going on directly conflicts with their passions. There's no need for railroad tracks - the game is more of a fox hunt than a railroad. The PCs will drive the story, because they told you what they cared about and you made them a game about it.

As they follow leads from the local henchmen to the other stuff your main antagonists have, you just introduce them to more and more henchmen and more and more locations and villain goals (that continue being toxic to the PCs' passions).  The villain goals might shift, too, and get even more personal.  Where "corrupt the church of Ilmater" was their goal before, "Torment [the PC] Jakiri the Cleric of Ilmater by kidnapping the ones he loves" is even more personal.

Not all RPGs work the same way, though.  Here are some important caveats...

RPGs with Structured Adventures 
Many RPGs have internal structures that get in the way of this basic process.  These are RPGs where the game creates a conflict that the game designer or GM pushes on the PCs, rather than one the PCs investigate on their own initiative.  There's nothing wrong with that -- these are fun games. But because the structure is somewhat set ahead of time, we have to add another step.

For instance, in Monster of the Week, you're creating one-off threats for most sessions.  (It's literally in the title.)  In Night's Black Agents, the PCs are burned spies uncovering a conspiracy of vampires.  In Shadowrun, you're often doing black ops jobs for corporations through cutouts called Mr. Johnsons, rather than deciding what passions to pursue, yourselves. These conflicts come baked into their respective games. 

Here's how you deal with that:

First, be honest with your players in step #1.  Explain that they'll be playing Shadowrun (or whatever), and the structure of the game involves getting hired for covert black ops corporate espionage and sabotage jobs (or whatever). 

After that, make sure that you still get a lot of passions in step #2.  Step #3 and #4 are the same. 

Next, step #5 is a little different.  In step #5, you follow the game's baked in structure for an adventure.  You have a werewolf attack the suburban high school; have Mr. Johnson hire the team to steal a briefcase from some corporate scientist; or have the agents investigate a spy that was murdered outside a Bucharest blood bank. Whatever.  You do the thing that the game wants you to do.  But make sure the bad guys know who hurt them.  That's crucial for step #6.

Step 6 (for Structured Adventure RPGs): Now it's Personal!
The first time the PCs win a victory against the antagonists, the antagonists strike back.  They take their revenge on the PCs' passions.  The werewolf moot burns down the Chosen's family's home.  The corp that lost their briefcase sends security goons to "question" the Street Samurai's favorite bartender (he didn't talk, but it cost him three teeth and an eye). The vampires frame the agents for the murder of one of their own beloved contacts.

As you're running the game, continue to use the PCs' passions as stakes whenever you can.  Offer them opportunities to achieve or protect or improve things they care about, and set threats against them.  Make everything as personal as you can.

Character Death and New PCs
If a PC dies in your campaign, their passions die with them.  When the player makes a new character, they come in with all new passions.  How should you handle that?

First of all, reserve character death for only the most extreme circumstances. Because you know the PCs all have strong passions, there are literally fates worse than death in your game.  Use those before you get to character death. 

But even if nobody dies, there are still times new characters appear in your campaigns.  What if someone new joins the group halfway in? 

When you get a new PC, treat it like they're playing a module -- see below.  Tell them all the conflicts going on already and ask them to make a character that feels passionately about one or more of the things at stake in the existing conflicts.  The new PC can have other passions as well, of course.  Work up a new villain plan and new villain stuff (henchmen, prophecies, etc.) that targets those.

An Additional Note on Modules
When you're running a campaign from a module, step #1 is very important.  You need to "all but spoil" a lot of the campaign for your players, so that they can make characters that care about things in it. 

If you're running Curse of Strahd, you need to read the whole thing and help the players make characters that care about the themes and goals they'll eventually have in there.  One should be a vampire hunter.  One should have a sister or wife who looks like the twin to Ireena.  Another should be a priest of Lathander, the Morninglord (in a setting where the sun never shines).  And so forth.

(Here's another really old article from this site on a technique for sowing plot hooks among the PCs.)

October 20, 2017

Players Self-Assigning Rolls

Get your players to stop making die rolls you didn't call for.  

It does more harm than you think.

I get most of my inspiration for blog posts from RPG forums, where I read and help answer questions from DMs.  When a problem comes up a lot, I see a lot of good solutions from talented DMs.  I also get to refine my answer by writing it out for different people with the same problem.  Eventually, I write a post about it here so I can link back to it for them.  In this article, I'm using 5th edition D&D for my examples, because the most recent iteration of this question I saw was in a 5e D&D group on Facebook, but the problem happens in just about every RPG out there.

The problem I see all the time on that Facebook group is players self-assigning die rolls.  Self-assigning die rolls is trouble for reasons I'll explain below.  It happens when you describe the situation, and then the players describe their actions, assign themselves die rolls, and then roll them, all without the GM's involvement.  It looks like this:

"The mural depicts sorcerers of ancient Netheril forging some powerful magic rod, consisting of seven segments each four to six inches long."
"Arcana 19.  What do I know about the sorcerers of Netheril?"

"The hallway ends in a twelve foot square room with candelabras mounted on the walls, a carpet in the center, and no other features.
"Perception 24 - are there any traps or secret doors in this room?"

"Pleased to meet you, master Underhill. Rooms here are one gold piece per night each, double occupancy, and that includes all your meals. Ale, too. Wine and spirits are extra."
"Insight - I got a 17 - is the innkeeper planning to betray us?"

The player jumps in and rolls the dice, then you, as the GM, simply react as if you had framed the challenge and assigned the die roll.  Only, you didn't.

It may seem like the players are doing you a favor and handling some of the system for you, quickly getting it out of the way so you can get on with the roleplaying.  It's not.  Self-assigning causes a lot of problems.

First Problem: Bias
Players will consciously or unconsciously choose to frame the die roll in the most favorable way for their character.  Even players who have no intention of gaming the system are more likely to remember the skills and sub-systems they invested in for their character before ones that they didn't.  The Warlock knows all about the Arcana skill, so she rolls Arcana.  Netheril is ancient history, and the mural depicts an historic event; so this should have been a History check.  Similarly, the Perception check maybe should be Investigation.

The GM calls for the roll, not the player.  However, players might suggest a skill.  For instance, if the GM calls for a History check to understand the mural and the player fails, the player might say "I failed the History check.  Can I discern anything about the magic rod using Arcana?"  The GM still gets to decide, and the GM should still frame the check.

Second Problem: Framing
Players don't frame the die roll stakes -- you do.  Take a look at the Perception example, above.  This has to be the #1 most common way players self-assign die rolls.  They take it upon themselves to roll Perception checks without the DM telling them to.  The reason they feel so comfortable throwing Perception checks around is that even DMs rarely assign failure stakes to Perception checks, and when there are no failure stakes, the rest of the party can try the check if the first player fails.  But remember GM 101:  Every die roll should have interesting stakes for both success and failure!

The Perception check in the example above is poorly framed - the player needs to say what they're doing -- how they're looking for traps and secret doors.  If that action happens to come across a trap or secret door, the DM can call for the appropriate skill check at that point. You can't just stand outside the door and look around the room and see all the traps and secret doors.  They're better hidden than that!  Failing to find a trap should always cause a problem (usually the problem is triggering the trap).  More on trap framing, below.

Every roll should have both success and failure stakes.  It's the GM's job to make the world react to the PCs' actions.  When a GM says, "nothing happens," the GM is derelict in their duty, in my opinion.

Third Problem:  The Inevitable Success Shuffle
When there are no failure stakes, or nothing happens on failure, the players do the inevitable success shuffle.  That's what happens when player 1 fails the check, and player 2 says "OK, I try too."  If player 2 fails, player 3 tries.  Repeat ad nauseum.  It's a huge waste of table time.  Player-assigned die rolls almost never have failure stakes.  That means there's no cost or risk to them.  That's why players are so quick to throw the die and call for their own checks.  They can't fail!  Literally!  Rolling a 2 just preserves the status quo.  To actually fail, something bad would have to happen.

But it's worse than that.  The inevitable success shuffle makes success almost inevitable.  See, even if every PC has a 0 in their skill, odds are one of the die rolls will come up lucky.  I'm dropping some science here:  If five people roll 1d20 each, there's a 98% chance one of them will roll a 10+, there's an 83% chance one of them will roll a 15+, and there's a whopping 23% chance one of them will roll a natural 20.  Put another way, if the chances are only 30% that any individual party member will pass the check, if all five roll, the chances are 83% that one of them will succeed.  The players aren't trying to cheat, and none of this is illegal according to the rules; but it's an "exploit" to use a video game term. 

Not only does this effectively "game the system," it's a huge waste of table time.  After going through the work of evaluating and responding to the first die roll, the second player announces that they're going to try.  And so on.

Fourth Problem: Too Many Rolls
If they're doing them at all, the players are probably making self-assigned skill checks as a sort of "minesweeper" in your game.  They're making Perception checks to avoid landmine traps or missing out on treasure and secrets.  They're making Insight checks to avoid being tricked or set up.  They're making Intelligence checks to avoid missing critical information.  They're making Stealth checks to avoid being caught by sentries.  The problem is, they're doing it preemptively.

When the players get to self-assign minesweeper checks, they'll always self-assign minesweeper checks.  There's no cost to doing so (see the second problem), the best PC always uses their best skill (see the first problem), and they almost can't fail (see the third problem).  If you're not giving them hints when there might be a trap, they only fall in a trap when they fail to make minesweeper checks.  All that adds up to a huge incentive to keep doing it. 

Fifth Problem:  Perverse Incentive
What's more, odds are you've rewarded the players for doing it or punished them for failing to.
"The hallway is 10' wide and 50' long with a thick wooden door at the end."
"I check for traps.  Perception 24."
"You spot a pit trap with spikes five feet into the hallway."
You just rewarded self-assigned minesweeper checks.
"The hallway is 10' wide and 50' long with a thick wooden door at the end."
"I walk up to the door."
"Five feet into the hallway, you fall into a pit trap with spikes.  Make a Dexterity save, DC 20 to avoid falling and taking 21 points of piercing and bludgeoning damage."
You just punished the players for failing to self-assign a minesweeper check by hiding a landmine and blowing them up with it.

Now, let's work on fixing the problem.

Fixing the Problem 1:  Fixing Traps

"But Jon!" You protest, "how do I use landmine traps if I don't conceal them from the players!?"

Landmine traps are the worst.

Don't use landmine traps.  I could go on a big rant about landmine traps in D&D, why we use them, where they came from, and why they suck...  But I already did.  Read it here if you're interested. They're only potentially interesting in combat.

"But realistically monsters would hide their traps!"

Sure!  Monsters lay landmine traps all the time, but they're boring and stupid unless there's also some opportunity to spot and avoid them.  Realistically, monsters aren't perfect.  There's always some clue. 

If you use landmine traps with no hints, a trap is just a random HP tax, and the only mistake the players made was not self-assigning enough Perception checks.  Yuck.  I guarantee if you use landmine traps with no hints, your players will become paranoid minesweepers, and your fun dungeon exploration will become a grinding slog full of mechanical crunch and little else.

Real quick, here's how you do that hallway trap:
"The hallway is 10' wide and 50' long with a thick wooden door at the end.  You spot a few kobold tracks in the dirt and grime."
If the player doesn't use a minesweeper at all:
"I walk up to the door."
"As you walk down the hallway, something catches your eye.  Roll Perception."
"You see lots of concealed kobold tracks on the floor.  Through their obfuscation, you can see that the creatures' tracks veer sharply off to the right side of the hallway for no apparent reason, then stray back to the middle.  What do you do?"
See?  This is how to use Perception well.  It's best to call for Perception checks only when success spots the danger in time and failure triggers the danger.  Only call for a Perception check when the player declares an action that takes them through the relevant area.  And by "through" I mean where the trap would be triggered, or away from the hidden treasure, or past the secret door (not getting treasure or not getting a tactical advantage - those are great failure stakes).

What if the player fails their Perception check?  That's when the landmine goes off.  But you gave them a hint.  They missed it.  Then you gave them a Perception roll.  They failed that.  Now they get a saving throw.  How much more generous can you be?  Right?

Fixing the Problem 2:  Assume Competency, Build Trust

What we're doing here is assuming competency.  You build trust in your players when you show them that you assume that their characters are competent.  That way they don't feel like the base assumption is that they're bumbling idiots who forget to tie their shoes unless a player says so.  When you assume that the PC would open their eyes and look around before stepping on a trap, the player doesn't feel the need to call for their own Perception check once every five feet of hallway.  The player knows you know that their character isn't a rube who's never been in a dangerous situation before.  The player can trust you.
"Oops.  9."
"You see some kobold tracks on the floor.  By the time you realize the tracks veer sharply of to the right, it's too late, and you've already stepped on the false floor.  Make a Dexterity save, DC 20."
What if the player becomes suspicious (f'ing kobolds) and takes a minesweeper action?
"Does the pattern of tracks hint at a trap or something?" (Alternately, "I follow the tracks in case there's a trap they're avoiding.")
"They've made some attempt to conceal their tracks.  Roll Investigation, DC 15."  (or Survival DC 15, for the tracking version)
"18.  Success."
"You deduce that there's a spiked pit trap five feet into the hallway.  There's a safe path to the right, which you deduce from their tracks.  Though they've tried to conceal it, they always avoid that one section of hallway.  Once you know exactly where to look, you have no trouble spotting the plaster-covered wood of the trap door, prying it open, and seeing the nasty spikes 20 feet below." (The Survival version would just take them safely past the trap, without revealing it; but that's usually good enough)
And if the minesweeper action fails?  None of that "nothing happens" crap.
"13.  Failure."
"Though they've tried to hide their tracks, you notice that they avoid one area of the hallway.  There's something there.  You prod at the suspicious section of floor, and there's a loud crack, crash, crunch! as you dislodge the trapdoor, dropping it 20 feet into a pit full of spikes.  Whatever's behind that wood door up ahead heard the sound.  What do you do?"
I used a different failure condition than the obvious one in this example.  See this post for more ideas on advanced failure stakes.  For the obvious "you fall in the trap" failure stakes, see the failed Perception check example, above.

If you run it right, where there is no trap, the players won't feel any need to make a minesweeper check.  Where there is a trap, you drop a hint that they'll probably catch on to.  If the players are smart, and they are, they'll investigate.  In the unlikely event that they miss the hint, see above - they might fall for the trap.  But here's the key:  If they fall for the trap, it's not because of the failed die roll.  It's because they missed your hint (and then failed a die roll).  In other words, it's a fun game, not a landmine.  

Fixing the Problem 3:  Fixing Insight

This "assume competency, drop hints, make it a fun game" strategy applies to every kind of hidden information and hidden danger!  Consider the shifty innkeeper situation. 

Let's use a strategy of assumed competency to correct a player's misinterpretation: 
"Pleased to meet you, master Underhill. Rooms here are one gold piece per night each, double occupancy, and that includes all your meals. Ale, too. Wine and spirits are extra."
"I think the innkeeper might be working for the Zhentarim and plans to betray us. I want to know if he recognizes us or knows we're Harpers."
"Based on his innocent look and jovial greeting, the innkeeper doesn't recognize or suspect you. He's honest."
The player misinterpreted your description, seeing something sinister where you didn't intend it.  No roll is needed. You can treat the player's request as an automatic success. 

But if you want to, you can make them roll. 
"I think the innkeeper might be working for the Zhentarim and plans to betray us. I want to know if he recognizes us or knows we're Harpers."
"OK, let's see how subtle you are in your suspicion.  Make an Insight check, DC 15."
"12.  Shoot."
"The innkeeper looks like he was going to say something else, but when he sees your steel-eyed look, he shuts up and won't make eye contact.  You're pretty sure you have nothing to worry about from the innkeeper, but now the innkeeper is worried about you."
I'm applying stakes other than "you learn nothing" again.  By being suspicious and then mishandling the situation (with the bad roll), the PC has added a little color to the situation.  Maybe nothing will come of it, or maybe we'll improvise this into a real conflict later.  A successful Insight roll represents a much less impulsive approach.
"I got a 22."
"You're real subtle, leading the innkeeper through some seemingly innocent small talk that would make a Zhentarim spy sweat, but he doesn't pick up on any of it, and seems to enjoy shooting the breeze with you.  It's getting close to dinner time.  What do you do?"
But that's about a player being paranoid, seeing a hint of treason where there was none intended. 

What if the innkeeper really is a threat?  What if you really did drop a hint?

If there's no difference between your shady Zhentarim innkeeper and your upright innocent innkeeper, the players have to be paranoid -- especially if you once used a shady innkeeper to betray the PCs once and didn't telegraph it. One bad innkeeper, and they're going to suspect every innkeeper from then on, forever.

Always give them hints when they should be suspicious, and then make the challenge figuring out what they should be suspicious of.  That's a tricky concept, but I can explain better it if you've ever heard of this series of movies called Star Wars

One of the most famous "betrayal by a host" scenes in media history is the Cloud City sequence on Bespin in Empire Strikes Back.  Han, Leia, Chewbacca, and C-3PO flee to Cloud City to stay with Lando Calrissian, Han's old partner in crime.  Lando has betrayed them, and they all get suspicious pretty quickly.  If this were an RPG, what happened is that they got to Cloud City and met Lando, and the GM dropped some hints that Lando had an ulterior motive, but not what the motive is.
Leia: "Something's wrong here - no one has seen or knows anything about 3PO. He's been gone too long to have gotten lost! ...I don't trust Lando."
Han:  "I don't trust him either, but he is my friend." 
Immediately after that bit of expository dialog, Chewbacca finds C-3PO in pieces, waiting to be recycled, brings the parts back to Han, and then they get really worried. The stakes get higher and the tension rises. Lando arrives and invites them for drinks; and unaware exactly what is going on, Leia, Han, and Chewy go with him and walk right into the trap.

That's an example of how it works: "We saw the clues, we got suspicious, but we didn't realize the nature of the trap." Cite that sequence to players who don't understand that alert, suspicious people can still be surprised.  It's the difference between being suspicious and knowing what to be suspicious of.

Here's how you signal to them that they need to use Insight.  If the innkeeper was shady, you'd make that clear with your dialog:
"Pleased to meet you, master... Underhill you said? Is that right... Hmm... Anyway... Rooms here are one gold piece per night each, double occupancy, and that includes all your meals. Ale, too. Wine and spirits are extra. You, uh, look tired. You'll be wanting to get out of that armor. Can I take your things up to your rooms while my daughter gets you something to drink?"
See, there are enough clues in there to make the players suspicious, and even if they miss them, you can point back to them later.  But let's assume your players pick up on your hints.
"Good sir, we'll keep the armor on for a while. Why are you so eager to take our bags?"
"'I... I just saw you were weary and was offering to save you the effort, sir... I meant no harm.' Go ahead and make a group Insight check. DC 12. If more than half of you fail, you'll be caught when the trap springs."
[Most of them succeed.]
"You're sure the innkeeper has something planned, so you watch his eyes while you challenge him. They keep darting to the door - the only exit from the inn that you can see. He's waiting for reinforcements, and they're probably going to be here soon. What do you do?"
Now let's look under the hood of that example.

GM Hints - Player Investigates - GM Frames a Check
The GM calls for the Insight check as soon as the player challenges the NPC or takes any action to investigate. This is just another minesweeper check, but the player had a hint, so it's not just routine.  But it's the GM who frames the check. Not the player!

The Heart of the Problem: It's Just No Fun

The difference between "Hint - Investigation - Check" and routine minesweeper checks has a massive, practical impact on your game.  Review the fun formula for RPGs before we get into it.

If minesweeper checks are routine, then there's no tension.  You make an Insight check with every single NPC you meet, or at least anyone who could legitimately pose a direct or indirect threat.  Sometimes the NPC is a traitor.  Sometimes they're honest.  Sometimes you succeed at the minesweeper check.  Sometimes you fail.  None of the outcomes are caused by your decision as a player.  All of the outcomes are caused by the GM and the dice.

So back to the fun formula.

Story -> Problems -> Tension -> Excitement -> Fun

When you don't hint at a problem in the fiction, there's no story, no story problem, no tension, no excitement, no fun.  It's just a routine check, and it has one of four outcomes (see the diagram below).  Only one of those outcomes introduces a story problem -- the yellow circle.  Everything else is either no problem or no problem detected.*  

The "we failed a check to detect trouble" outcome (marked *) represents dramatic irony:  The players know there might be trouble but the players have to pretend they don't know about it.  With mature players, dramatic irony can create tension (excitement, fun), but because they self-assigned the check, they don't know for sure (OOC) if there is trouble.  That weakens the dramatic irony tension, so it's not as fun.

* dramatic irony
On the other hand, when you drop a hint, you've created a story:  The innkeeper is acting fishy.  When your fishy innkeeper could spell trouble for the PCs, it's a story problem.  When that story problem's outcome is uncertain, there's tension.  Tension is exciting.  Getting excited over an RPG is fun.

Failure Stakes
The example Insight group check is framed with failure stakes. The success stakes are the implied inverse of the failure stakes. "Go ahead and make a group Insight check. DC 12. If more than half of you fail, you'll be caught when the trap springs." See how the player doesn't know exactly what the trap is or what the Insight check will reveal?

The players have to trust that a good roll will have a good outcome and a bad roll will have a bad outcome and they will be treated fairly by the GM. Trust is a big deal in player/GM relationships.  If you don't have that, stop reading and go be more generous to your players until you fix the trust problem.

No Self-Assigning Means no Inevitable Success Shuffle
The GM framed it as a group check instead of a single check.  If the player had self-assigned an Insight check and failed, without any failure stakes, nothing happens, right? Then what? You know what!  Inevitable success shuffle. 

There's a reason group checks exist. In fiction, if your friends all trust the innkeeper, they're going to think you're just being paranoid when you're like "Guys! He's trying to ambush us!" "No, man, he just sees we've been on the road all day. Why would an innkeeper be ambushing us? We've got magic great-swords and he's got a rolling pin. You need a drink, dude."

Tactical Advantage
The roll stakes set up a tactical advantage that's relevant in every kind of RPG from Pathfinder to Fate Accelerated.  If the PCs are prepared for the reinforcements, they can choose where the conflict takes place.  They can take the fight out in public, where the reinforcements are likely to hesitate; they can barricade themselves in; they can run away before the reinforcements arrive; they can lay an ambush; etc.

Failing the check starts the encounter. The players can't even object: They had their chance to spot the ambush, and they failed the roll. Their characters may have been suspicious, but suspicious is not the same as informed and prepared. Like Han and Leia, they knew something was up, but they didn't know what it was.  They missed that telltale glance at the door.

By the way, if your innkeeper's trap is more distant, frame it this way:

"Make a group Insight check, DC 12. Either way, you're suspicious. On a failed check, you are eating dinner in armor, weapons at the ready, when the attack comes. On a success, I'll tell you the nature of the attack before dinner."
See?  I'm setting up an Empire Strikes Back situation where the players might know to be suspicious, but without getting more information, they will still fall for the trap.  Let's assume the group check fails.  Now we have tension coming from two directions:  The hints the GM dropped about the innkeeper add tension, but now the dramatic irony of the players knowing they failed the group check.  They know that there's an attack coming at dinnertime and they don't know its nature!
"OK, so things go fine all afternoon. There's a fire and the innkeeper tells some good jokes, though you keep your armor on and guard up. Dinner is beef stew with bread and ale. You're all worried about an attack, so I assume you're not getting drunk. Where are you in the room while eating dinner?"
"I'm eating by the door, standing up, pacing. I pretend to drink the ale, but just nurse it slowly. Maybe pouring it out in a potted plant or something to look like I'm drinking more than I really am."
"I'm back by the hearth, away from the window. Acting casual.  I have hot tea with my dinner, in case I need to throw it in someone's face."
"I'm eating at the table in the middle where I can see anyone approaching through the window.  Is that right?"
"Right. You've got a good view, but anyone approaching can see you, too."
"That's fine.  I try to act casual." 
"I'm eating at the bar, on my third helping of stew, fourth helping of ale."
"Stay sharp Ragnar!"
"Sure, sure, I can handle my ale! (Nope! My Flaw is I like to drink to excess. So I do!)"
"Cool.  Take Inspiration for acting out your flaw."
"OK, everyone make a DC 20 Con save vs poison or fall unconscious. Ragnar, you have Disadvantage on the check."
"Aw $%i#!"

Other Self-Assigned Rolls

Though information checks like Perception and Insight are the most common rolls players self-assign, there are other situations players self-assign rolls.  This article is already pretty long, but I would be remiss if I didn't at least touch on them.

Sometimes they roll initiative before the GM calls for it, assuming the GM is going to use the tactical combat system to resolve a fight (hint: you don't have to use the combat system for every fight).  That's easy to deal with.  Just say "hold your horses!"

Sometimes they self-assign Stealth checks to hide, which can actually be helpful in a D&D or Pathfinder combat if they're playing characters who need to hide in combat a lot.  Outside of combat, and outside those RPGs, make the player back up and then frame the check before they roll.  Stealth failure stakes are some of the most fun situations in any RPG.

Sometimes players make assumptions about how you want to handle something and just throw some dice, thinking they're saving you time.  They'll walk up to a locked door and make a Dexterity (Thieves' Tools) check or leap over a chasm and roll Athletics before you call for the roll.  These players are often over-eager.  Maybe they've got a little bit of the rules lawyer in them.  Ask them "don't you want to know what the DC and stakes for failing that check are before you tell me your roll?"

You see self-assigning all the time with social skills.  it's a special case of the "players make assumptions" situation, above.  The PC will walk up to an NPC, tell a lie, and then the player will announce "22 Deception!" or whatever system equivalent your game uses.  In my experience, this impertinence causes PCs way more trouble than it's worth, because in social scenes, GMs hate being interrupted by system unless they're the one doing the interruption.  And GMs are happy to push the situation and call for yet another check. Once the Rogue with +10 Deception has committed to a bluff, the NPC will turn to the +0 Deception Fighter and say "the guard looks at Ragnar and says, 'are you here to deliver something, too?'  Your reaction might give you away.  Give me Deception, DC 10 or else the guard gets suspicious."  I think GMs generally have good instincts to push back on players who thrust their stats into conversational scenes before they're called on.

Exception: Player Character Special Abilities
There's one situation where the player knows more about the framing of a die roll than the GM does, and that's exceptions to the rules caused by player character special abilities.  PC special abilities often frame the stakes of the die rolls they require very specifically, to limit how broadly they can be used.  Because the die roll is already framed up, the player can usually use the ability and throw the dice without the GM prompting them without causing any trouble.  The GM isn't really free to frame up the die roll in many cases.  It doesn't really matter, then, if the GM called for the roll or not - the system says the roll is required, the system says what to roll, and the system says the stakes are. 

Player Agency

The last thing I want to talk about is the concern some readers might have that denying players the opportunity to self-assign die rolls takes away their player agency.

Player agency is the ability of the players to control the shared imagined space of the game.  When we deny players the ability to have their characters take an action or have that action be meaningful, we deny their agency.  

Putting a stop to self-assigned checks does not deny the characters' actions, and it only makes their actions more meaningful by giving them more opportunities to make meaningful decisions.  

When a player becomes suspicious about a 5' square of dungeon tile and rolls a Perception check, there is no action - just system access.  The action would have been "I lean over and look closely at the tile, prodding and prying at it with the tip of my dagger.  I'm worried that it's a pressure plate."  The GM isn't stopping the player from taking that action.  The GM is simply insisting on making the system decisions related to that action. 

Further, because a lot of self-assigned checks come up around hidden information, a GM who follows my advice and gives a lot of hints is giving the players more agency by pointing out opportunities where their actions can make a big difference in the events of the game world.  Without hints, hidden information is more likely to be like a landmine (whether it's a treasonous NPC or a hidden trap).  It's almost entirely out of the players' control other than choosing to sweep for mines or not with no reason other than an abundance of caution.  That's not a lot of player agency.