GM Toolbox: Cementing, Shifting, Delegation

From Rickard Elimää

Roleplaying Tips Newsletter #0628

A Brief Word From Johnn

Final Hours – Last Chance

Would you kill PCs for food? Could you really get that hungry?

Well, it’s important for your players to think so!

The Kill PCs For Food t-shirt campaign has met its threshold. So your order is guaranteed to ship. Thanks to everyone who has ordered so far!

Will Kill PCs For Food

There are just a few hours left in the campaign, so grab your limited edition shirt now:

Add to Cart » Will Kill PCs For Food

Graphic of section divider

Casualties Afflict Phandelver Session #5

We played last Friday and had a blast. A disintegrating blast as you shall soon see.

The session starts with the PCs resting and preparing to assault the orc shaman’s cave. It is a full table. PCs missing last session jumped in immediately on the premise they were getting worried about their friends and went out searching for them. With the party reunited five strong, they venture forth into the mists to clean up the orc raider tribe.

As they approach the cave mouth guarded by a pair of sentries, screaming erupts from the dark hole. Then a clawed-up and bleeding orc body flies out the cave mouth and sails through the air over a cliff. More screams of pain and terror perforate the morning air.

The PCs advance and quickly take care of the guards. With trepidation, they peer into the cave.

They see mass murder. Blood covers the walls and floor. Claw marks scrape the stone walls. And a terrified huddle of surviving orcs with backs turned to the characters peer into a tunnel leading deeper into the mountain.

The PCs charge. And that’s when the fighter perishes. Just as the orcs and their ogre leader are falling to the chop-chop-chopping of the PCs’ hatred, another wave pours from the tunnel, fleeing from the unknown monstrosity. The party’s main murder hobo gets hobbled then murdered as fresh pork overwhelms him.

Finally, the cave goes quiet as the last orc falls. Two mysteries lay before the characters as they retreat outside for a short rest.

First, what was ripping the orcs from limb to limb? Second, what was causing the strange mutations they saw amongst the orc foes they just fought? One orc was orange, one had a two foot wide head, another had 10 pound testicles.

Just as the PCs are girding up for another foray into the cave, they hear weak cries for help coming from that tunnel at the back. It doesn’t sound orcish, so they advance. Passing through the cave piled with orc bodies, they discover the dark tunnel is filled with even more orc carcasses, plus many goblin slave carcasses.

Hopscotching their way through, the party enters a smaller cave that must have been the shaman’s room. The walls are decorated with strange runes, bone mobiles, and wolf pelts. In the corner is a bound man. He’s been tortured by his captors. He’s naked.

This is the new PC for Chris, the dead fighter’s player. He tells the party he saw terrifying tentacled creatures burst through a hole in the wall at the back of the cave. That’s what’s been making pork pie of the PCs’ enemies. Based on this news, the party piles bodies up to block the hole. Unfortunately, this rouses one of the creatures and it begins clawing its way through the fleshy barrier. The creature is fearsome, with two massive claw-tipped limbs, a body of a giant lobster, and a tentacled mouth.

The creature latches onto the druid and claws then grapples him. The stalwart dwarf fends off the poison now swimming through his body and counterattacks. It’s a tough battle, but the PCs manage to kill the creature by pretending to flee and retreating to a more defensible position, where they whittle the monster down.

Flush with victory, the party ventures back into the shaman cave and begins looting. Roscoe the rogue discovers a secret door. He opens it and finds a short passage with a solid oak door blocking the end.

Curious, the party opens it. This is when the paladin goes down.

The room beyond the door contains a bubbling fountain filled with fetid water. The walls are covered in more shaman runes and fetishes. The newest party member jumps into the fountain and frolics. The druid joins him, as does the paladin’s goblin servant. Meantime, the mage casts detect magic and realizes there are two magic signatures in the room. One is the fountain. One is something big and invisible in the corner.

The rogue runs up to the invisible thing and touches it. It’s cold, damp, and spongy. The paladin comes over to investigate, as does the druid. This forces the beholder(!) to attack, making him visible. A ray shoots out of one eyestalk. And faster than you can say “Pile of gray dust,” the paladin is hit and turned into a pile of gray dust. This, just after acquiring some fine full plate armour from the orc leader minutes ago, the party’s best treasure since campaign start.

Everyone flees. Except the mage, who takes a moment to disbelieve. The beholder aims an eyestalk at him. The mage flees.

Back to the safety of the entrance, the mage realizes the beholder was not alive. It was dead. It was an undead zombie beholder. In a cave. The party retreats back to camp.

Counting their losses, the survivors consider themselves lucky and head back to Phandalin. It takes three days to return home. On the road back they meet a mercenary – Jason’s new PC – and he joins up with the party.

Back at the mining town, the group convalesces at the inn. They drink heavily. It’s time to break out my new carousing table (I’ve included it at the end of this issue).

The dwarf wakes up the next morning with no memories, naked, and lying beside the chewed-up body of a humanoid. The dwarf’s face and hands are covered in blood.

I forgot to mention that, back at the fountain, the fighter received a permanent hit point after his swim.

And the dwarf earned an intense fear of orcs. This is unfortunate, because the dwarf’s most hated foe is orcs. This must have really messed up the dwarf, because the body he woke up beside was that of an orc.

Roscoe the rogue wakes up with a new tattoo. Rolling a d6, Jeff gets a 6. Roscoe is now the proud bearer of a unicorn tattoo.

The mercenary wakes up with vague recollections about hearing a terrible secret. But he can’t remember the details.

The mage gave all his money to someone with an incredible money-making scheme. The guy has left town.

The half-orc barbarian, who makes the dwarf very uneasy, did not drink, he just caroused. So he wakes up with purse, skin, and dignity intact.

We end the session there, with the party ready to follow-up a lead back down the Triboar Trail, questing for a ruined watch tower.

I think we won’t play now for another month. In two Fridays is Halloween, so we can’t play that night. November 14 will likely see our next gaming.


GM Toolbox: Cementing, Shifting, Delegation

The following article is based on a discussion I had with Rickard Elimää, where I asked him the following question:

RPGs are interactive, so we can’t get the finely groomed and polished story of a book or movie. So I guess my larger question is, what methods can a GM use to help produce those great emotional moments a writer has the luxury of crafting, especially through re-writes?
Rickard responded with great insight, and I have compiled the content into an article for your gaming edification. Take it away, Rickard!


Sequential Play: Cementing, Shifting, and Delegation

I’m playing with either fish tanks, sequential play, or collaborative storytelling. Each one has got their own advantages, and I’m even mixing them up in the same session. Feng Shui (sequential play) and InSpectres (collaborative storytelling) will sound the same when I’m playing them.

It’s hard to analyze myself about how I game master, but I remember one specific episode when we were playing Svavelvinter (an adventure module; Eng. Winter of Sulfur) when I clearly stated several times to the players that it was going to be a hard railroad. One player said afterwards that he felt their choices mattered. How did I manage to do that?

I can see three techniques that I use:

  • Cementing
  • Shifting
  • Delegation

All three are about how to structure an adventure by how different information is presented, to be able to steer the players in a way so they can get a sense of the plot, setting, or history. Which one depends on what effect the game master wants to achieve. I’m mostly into delivering a surprise through giving the players an insight.

The definition of sensation, as one of the reasons of why to participate in a task:

“To experience something. To escape from reality. To get an insight or to live out a different perspective.”
If you ever watched the movie Man From Earth, you know what I’m saying when I talk about seeing things in another perspective.

Sequential Play

Sequential play is a way of structuring how the game is played, very similar — if not the same — as designing a game system. If I want to present a sensation for my players, I will go with either:

  • Linear Play: Pre-written places of interaction followed in one order.
  • Assembly Play: Pre-written places of interactions that can be executed in any order.
  • Ballistic Play: Pre-written start and finish, with the players doing whatever in between.

Alternatively, I will go with a combination of them. For example, pre-written places of interactions that can be executed in any order with a fixed start and end. Note that the place of interaction can be an event happening, a conversation, or an actual place where the players discover something.

Also note that anything can happen in that place. So even if I’m going with linear play, the players can still interact the way they want in that particular scene. I remember one theoretic discussion over at the Swedish roleplaying game community years ago, where one suggested the picture frame:

“The player can paint whatever they want, but not going outside the frame.”

I thought this was a nice analogy for one type of structure for an adventure, but it can also be applied for a place of interaction. An example I hate to use is when you investigate a haunted house. You could run out the house and call the cops at the first sign of weirdness, but that’s not how you should play Call of Cthulhu. That’s painting outside the (often unsaid) frame.

Why do I hate to use that example, even if it shows clearly what I mean? Because I think it’s a flaw in the scenario design if you allow that option.

I will talk about three techniques in three different sections:

  • Cementing
  • Shifting
  • Delegation

I will do that by sharing some play experiences and my thoughts about how I should handle them. Which one I’m going to start with, because I mentioned the CoC example, is now obvious because you need to cement certain things to have a better success at delivering a certain sensation.


An acquaintance contacted me and wanted feedback for a convention scenario. The setup was straightforward:

The players played a family with a cruel and abusive father during the 18th century. The cruelness is portrayed at the beginning of the scenario through several scenes, and it should lead up to one member killing the father. The father awakes as an undead and starting to create an army of zombies by brainwashing the citizens of the village, either by magic or by killing them and raising them as undead. What shall the players do to stop that from happening?

Well, what should I do to make this happen? There are a couple of events bound to happen, but it doesn’t start until the father is dead. So that’s a bit I cemented: I changed the scenario so the father was already dead. I wrote on each character how they killed the father. One pushed him over an edge to fall into rocks into the sea. Another poisoned him. A third … well, you get the picture.

What had happened was he had risen only to be killed by another family member. The only one innocent was the one playing the father’s doctor, and I wrote on the doctor’s character sheet that he should investigate what happened (which, amusingly, led 3 out of 4 times to him being framed and often killed by the others).

We started out playing with all characters sitting in a house. It was raining outside and the dad had died. I then took the liberty to jump backwards in time and frame some scenes to show the behaviour of the abusive father. I cemented that too.

When all this was done, I returned to the house in the rain. The players were thinking the dad was dead. Suddenly, the door is thrown open and the dad is standing in the rain, all wet, walking in like nothing had happened. Imagine the players’ faces. Imagine the unease they suddenly felt.

In the adventure suite Svavelvinter the linear story goes as follows:

Characters discover that a temple is abandoned. They are contacted by a group of local barbarians. They are sent to retrieve magic items. An undead king awakes, wanting his sword back, which is one of the magic items they stole. They go to a castle where an evil cult had their residence to kill them. So why should the characters want to go to the temple? Why should they want to retrieve the magic items? Why should they go to the castle of that evil cult?

The adventure itself gives out a couple of goals and character suggestions. One is a priest that wants to contact the temple – only to discover the priests have been kidnapped (by the cult). Another has a mission to look for a person. A person who the character later finds out joined a cult.

One player, who already played the adventure, came up with the idea of playing a pathfinder guiding the characters. (And why would a player want to replay an adventure? Because of the sensation the adventure’s setting brought combined with my game mastering.)

So what should be cemented to not make the characters run away and contact the police in Call of Cthulhu?

  • Cementing through the use of unavoidable scenes. If something must happen, start with it. It sets the tone of the game.
  • Cementing through the use of background information. People will begin to expect something. Use this expectation to your advantage. The example above shows how you can break the expectation to gain an effect.
  • Cementing through the use of goals. This makes the players believe that it’s what they want when it’s really the adventure itself that controls what they strive for.


Shifting is another structure I use when I do sequential play. Playing out a story in roleplaying games is done through scenes, where a new scene occurs when there is a (significant) change in either place or time. There are two types of scenes:

Resting: These scenes wait until the characters do something, such as entering a place, visiting a person, triggering a situation, or fulfilling any other kind of requirements.

Delayed: These scenes will happen no matter what the players do but will show up later. Something will happen after the second day has passed, or they will be contacted by an organization because of the character’s (cemented) agenda.

As previously mentioned, in Svavelvinter, the characters got hold of an artifact sword. This triggered the awakening of a skeleton king. During any point of the adventure, the king will step in, grab the sword, and go away. (It was a setup for the metaplot later to be revealed.)

I played Svavelvinter because I wanted to playtest my now published game Matiné. The characters got involved in a skirmish, and the game system said a combat should come to an end after 5-6 turns, otherwise the game master had to narrate its ending. Again, a skirmish. The possibility it was going to end in 5-6 turns was slim.

So when it was time for me to end the combat, I remembered the skeleton king wanted his sword back. A huge army of skeletons flooded the battlefield, making all participants flee or be killed. A huge knight with old armor covered with cobwebs went up to the character that had the sword, looked at him, took the sword without a word, and went off. I could had done this scene at any time, but here seemed an appropriate moment.

What I find is interesting is that I tend to use shifting while playing other forms of play structures, like collaborative storytelling or an open-ended fish tank. The fish tank is a model where the relations between factions (an umbrella term I use for events, items, people, and groups) ARE the scenario. It’s used to play out mysteries and intrigues, and the players discover the story by having the game master reveal the relations. I use the revelation as a reward for interacting with other people.

When the players talk to someone, I look down at the relationship map that is typical for the fish tank. I choose one or more relations to be revealed and also come up with a reason why the person knows that relation. Someone can always have heard of another relation in form of a rumor.

I talked about three different styles of structures within sequential play: linear, assembly, and ballistic play. This is one part of assembly play, where you’ve got a lot of scenes but they can occur in any order.

  • Shifting can only occur with the right type of preparations. Placing a band of bandits in one place will make them fixed. Having a band of bandits to throw in at any time, that’s shifting. Having events waiting to happen at the right moment, that’s shifting. The players will still go through the scene, but when you feel it’s appropriate.
  • Shifting can be used to adapt to the players’ plans and actions to reward them. Make them feel smart. Make them feel like they contributed with something. Never take that feeling away from them.
  • Shifting can be used to add twists. Playing out a large battlefield between humans? How about throwing in a third party of undead?


I remember when I was in my teens and the players started out imprisoned. I thought out thirteen different ways of fleeing. The players did the fourteenth. That made me realize some things shouldn’t be prepared.

I noticed that if I ever had a solution to a problem I threw in front of the players, then the possibility for me to block the players’ ideas would increase. Why didn’t I like that the game master should come up with solutions to skill contests in D&D 4E? Yeah, you probably understand my answer to that.

So I only come up with the problems, but never the solutions. When I read about bangs, I got the similar notion. Only throw in bangs at the players. See how they react. React, to take a stand where your character is. It’s one way of helping out the portrayal of the characters.

When I prepare a sequential story, I do that to bring a sensation to the players, but they can also give themselves a sensation by portraying a character. Nothing in my preparations hinders the players, but unlike bangs, it doesn’t help either. On the other hand, you could cement personality traits or relations between the characters if you want to put focus on either reacting through personality or creating relationship drama.

So what does this have to do with sequential play? The immersive experience through the sensation of another person’s perspective might seem to have little to do with it, but the process of not thinking up solutions does. Even if a scene is bound to happen, when I game master, the players can still choose from where to enter that scene and where to end it. I will leave that up to them.

When I play Feng Shui, it’s going to start with me cementing some goals through the fight at the beginning of the session and then letting the players take their own way to the fight in the end. Sometimes I have scenes to throw in, assembly style. Sometimes I do a pure ballistic play where the players have a picture frame to paint whatever they want, as long as they are inside the frame.

This Is Pulp is collaborative storytelling, but it’s also ad hoc linear play. All the participants will make up tropes – People, Rumors, Events, and Places – as P.R.E.P. The players then form together a goal to follow, based on the tropes, and the game master then uses all this to come up with an antagonist. That person then frames scenes using the tropes to create dangers for the characters.

The players can do anything to overcome the danger but only affecting what’s happening in the scene. The game master will still steer the character, and the story, towards ending with a fight between the antagonist and the characters. The game takes 1.5 hours to play, so it’s not like you need diversity in how the frames are set.

  • Let the players take the responsibility of portraying their character. This will add to the general atmosphere for the session. You do, however, need the players to buy into this way of playing. That’s why I used the word responsibility.
  • Let the players come up with ways to come to the end. This will ease the burden for you as a game master, and lets you have less prepared.
  • Let the players take care of coming up with the solution to the problems they face. It will make the game master more open to the players’ solutions, but also make the players more involved.
  • Let the players take care of coming up with their own goals. In Svavelvinter, I can see the game master saying, “Hey, you’re going to fight an evil cult. I want you to come up with reasons to why you want that cult destroyed”. It’s their decision, but based on your premise.

Sequential Play and Adventure Design

When I wrote these previous sections, I began to realize how close writing an adventure and these techniques were:

  • Sequential play: Following the structure of a combat round.
  • Cementing: Picking skills, setting the tone, choosing class, and race.
  • Shifting: Creating a rule system that adapts to the players’ ideas. The moves in Apocalypse World is an example of this.
  • Delegation: Writing background and personality for their characters. Coming up with relatives.

So I wonder if you can create a whole game based purely on the structure of how you write an adventure, and why isn’t this used more in roleplaying games?

Sequential Play: Uncertainty, Investment, And Reincorporation

This is a really hard question to answer, possibly because it’s a big one. If I paint in broad strokes, I would say uncertainty, investment, and reincorporation. Each one is a topic of it’s own.

Secrets and twists, are good examples of uncertainties. Conflicts is a third, which is why we see so many games having combat systems. You don’t have to have conflicts to weave a good plot though, even if it’s a common belief.

Investment can come in several different ways. Uncertainty is one of them, where it (hopefully) sparks curiosity to find out more. Making an effort to discover builds up the investment. Contributing and taking a part in the game also builds investment. Having your choices made meaningful is a yet another one.

Reincorporation is something that Keith Johnstone talks about in Impro, but you can also find it in Dan “Community” Harmon’s Story Circle. You can add new things all the time, and that will bring uncertainty, but it’s not until you start to reincorporate what you previously experienced that the feeling of the story coming to an end begins.

And then you’ve got narrative techniques, themes that excite the audience, and similar things that are components of a good story. Like I said, it’s a hard question to answer. I began to think of some of my own successful experiences, and I wonder what things I use that I cannot see.


I would say it depends. Sequential play is used to create a sensation for the players, and there you can control a lot through cementing and shifting. I should say a fish tank is instead about making the players’ decisions important by showing the consequences of the their actions. The sandbox is about exploration. And finally, you’ve got collaborative storytelling where you thrive in creativity and the feel of togetherness.

So creating an emotional experience varies depending on what the players get from playing with you. The important part is to be able to mash story and game together, but also consider the players as a component in the game system and take them into account.

Story Structure 101: Super Basic Shit — Story Circle

My thoughts about We know so little of our hobby how Improve can be useful for roleplaying games

The Fish Tank as a Mystery — Fish tank (which you know about by now)

Some more ideas about Calling All Writers – Games Are The New Frontier combining story and game experience. (Journey is a perfect example of this.)

How To Avoid Prediction Railroads

Quick tip. Players like certainty. They try to pin you down about the future with divinations, tarot readings, prophecies, and other means.

You can avoid getting into a fixed path with this simple trick.

Whenever it’s used, just change the word “prediction” to “forecast”.

It works for weathermen every time.

How Would You Bootstrap Your World?

That’s the question I asked in the GM Tips group, and there were several interesting responses.

Here’s the original request I made:

How would you bootstrap your world? I am creating a world for a new RPG campaign with my friends. The game rules have some links to the world, such as gods grant certain powers, cultures offer certain traits, etc.

I don’t have time for weeks and months of detailed world building, as my friends would hang me to wait that long to start playing. So I’m wondering if there are methods or models for creating the basics and then iterating over time.

And here are the tips from the GMs:

Use this completely fun mini-game before the main event. It’s called Dawn of Worlds Game 1 – 0 Final Dawn of the Worlds and is free.

Use a pre-made setting that is light and portable to whatever rules you’re using. For example, Beasts & Barbarians for Savage Worlds, Legend of Steel for Barbarians of Lemuria, or the “Point of light” ones. This saves tons of prep time to get started, but offers more then enough room to insert your own stuff piece by piece when necessary.

Plan out the culture pretty well. Make notes like NPCs for places that surround where the PCs are adventuring. This will tell you basic cultural norms. When the characters meet someone from that culture, you know what his culture is. Then adjust the NPC based on how well he would fit in.

Just define a small area with a single town or similar. Keep the PCs local for a while. Expand on it as they explore and when you get time between sessions. Make stuff up when put on the spot.

Build a story that would reduce the need for travel outside the starting culture, but start exposing them to touches of other parts of the world as you can.

Let the players do some of the defining as you go along. They decide to go to a town, so you ask “Ok, what sort of town?” Let them throw in details and riff off that. If they define things in a way too favorable to them, throw in a “yes but.”

“Everyone in this town considers wizards to be gods.”
“Yes, but they sacrifice their gods at midsummer, and that’s tomorrow.” “There’s a diamond mine!”
“Yes, but the dirt there’s poisonous, which is why no food grows here, and why the zombies have moved in.”

There is an RPG called Microscope. It lets you play out a timeline for a setting and is great as a first session if you want players to influence it. It’s also interesting to play as a game in itself, and a great way to let players shape the tone of the world without them knowing too many details of what they’re going to meet.

After you have the main themes, creation myths, races, geography, and overall history setup, ask players what they are most interested in doing. That helps you focus on a place and time. Then encourage players to help build up information. For example, one might create myths for her character’s people. Another might create the calendar.

Thanks to these bright GMs for their tips: Andrew Knighton, David Tannen, Dirk Collins, Jeffery Ingram, Josh Pearce, Marcus Burggraf, Robert Neaves.

Random Table: So You Get Drunk at the Tavern….

At least one character in my new campaign likes to get hammered. Between adventures or while resting up between events, the PCs like to hit the tavern, chat with the locals to get new rumors, gamble, and drink.

Inspired by the Hubris Carousing Table, I’ve created one of my own for random consequences when PCs choose to drink too much….

  1. One-night stand with a half-orc. (Alternate: rival, enemy, least desirable partner.) 1-2 They are engaged. 3-4 They become parents in nine months. 5 PC acquires disease. 6 Enraged spouse eventually hunts PC down.
  2. Became a drinking buddy with someone who shared an important secret. PC has forgotten a key piece of information though, and just remembers enough fragments to track down the secret if they work at it.
  3. PC revealed a terrible secret to someone. Pick the worst person who could learn that secret. 1-2 PC doesn’t remember who they told. 3-4 they remember who but don’t know what they told. 5-6 they remember everything.
  4. Picks a fight with someone they shouldn’t have. 1-2 PC wins. 3-5 PC loses, wakes up with half hit points. 6 A draw, PC at 1 hit point.
  5. PC sang a terrible song that offended almost everyone, including someone they’ll regret offending.
  6. PC and a drinking buddy broke into a place and stole something valuable. 1-2 Buddy has the loot. 3-4 PC has the loot. 5 Loot was buried in a place PC remembers. 6 Loot was buried in a place PC doesn’t remember.
  7. PC offends a witch, who cursed them.
  8. PC bought the house several rounds. Total cost: 10d10 gp.
  9. PC made a bet with someone they don’t want to cross and lost. They must now perform a humiliating chore or act in public.
  10. PC and 1d4 drinking buddies carried back to the tavern something large, something heavy, and something that belongs to the 1-2 Mayor, 3-4 Villain, 5-6 Arch-Bishop.
  11. PC wakes up in puddle of own vomit. 1-2 They also puked on someone else last night. 3-4 Vomit contains fingers. 5-6 It’s not their vomit.
  12. PC gets a tattoo. 1-2 It’s cool. 3-4 It’s of a secret crush. 5-6 It’s a unicorn.
  13. PC gets rowdy and arrested. Must pay 10×10 gp fine for release.
  14. PC was robbed and is missing the most valuable thing that was in their possession last night. 1-2 They have no memory of when it happened. 3-4 They remember when and the general circumstances. 5-6 They know who did it.
  15. PC wakes up naked and tied to a bed. 1-2 It’s his bed. 3-4 It’s not his bed. 5-6 It’s in a monster lair.
  16. PC wakes up in the stocks in the city square.
  17. PC wakes up in the woods near town, naked. There’s a dead, half-eaten carcass nearby and there’s blood on the PC’s face and hands. 1-2 It’s a deer carcass. 3-4 It’s a skunk carcass. 5-6 It’s a humanoid body.
  18. PC wakes up with his head stuck in a soup bowl.
  19. PC joined a secret cult. 1-2 The cult is crazy but harmless. 3-4 The cult is crazy but dangerous. 5-6 The cult is dangerous, well-organized, and well-connected.
  20. PC gave all his available money to a crazy scheme. 1-2 Money and schemer are long-gone. 3-4 The schemer succeeds and returns in d4 weeks with 200% repayment. 5-6 Schemer returns in 1d4 days for more money and a strong quest hook.

Many results on this table are nasty. To make this fair, players must confirm with me their PC is getting drunk, and they’ll know a table roll is coming up in the morning.

Thanks to Joseph Teller and Tor-Ivar Krogsæter for contributing ideas.