The 'Always, Sometimes, Never' Framework for Fictional Characters
Instead of the Wizard of Oz behind the curtain, you can find fictional rules and probabilities of likely character behaviour -- and you can use this to your advantage in writing...
(Note: I’ll be following this up with a Game Writing -specific Prompt in a few days, so for those of you interested in writing for that medium — stay tuned and think on this in the meantime! I’ve also just updated my Writing Workshop page with a number of graphics and diagrams to show how it all works, for anyone who may wish for more hands-on assistance)
On my workshop course, we work a great deal on writing skills in general, not just in interactive forms. One of the most important early exercises occurs during Phase 2, where students are trained one-to-one in various new ways of approaching their work.
The first step is on characterisation — a process I call ‘ALWAYS SOMETIMES NEVER’ (ASN). It is similar in many ways to the ‘questionnaire’ approach to coming up with characters that so many writers adore or revile, the inventorying of every possible piece of information about a hero that the audience might never ever get to see.
But here? It’s not about what we have. It’s… sort-of about who we ‘are’, something such questionnaire inventories do involve a little bit, but it’s not really the emphasis in ASN. What it is?
The ASN approach to characterisation is about what a person does. It is about action.
I’m going to look at an extremely popular well-known character first, to demonstrate some of the underlying theory behind ASN.
Iron Man - who is Iron Man?
He is Tony Stark
Most people who have seen the Avengers films will be able to give a fairly consistent answer to Tony Stark’s nature — brash, sarcastic, arrogant, fun, moral, intelligent, conscientious, etc. There will be traits beyond these that might vary based on individual interpretation, but we feel like we know who he is at his core. These are not things Tony has, these are a mixture of things he is and that he does (to some extent, inextricable).
Some of the traits we associate with him might be stronger in some contexts than others — for example, will he always display the emotion-driven pettiness that he showed in taking Captain America’s shield due to a revelation about a lie? No - he is not petty or overcome by his own feelings in a lot of other scenarios, at least not to the detriment of the greater good. We only see him like this in moments of great upset (a vaguely similar incident occurring when he arrives back on Earth in Endgame and begins mocking and attacking his co-workers for their failures). This is who Iron Man is on his worst day.
But a lot of the core traits - his brashness, arrogance, sarcasm, intelligence, his wish to do good at his core — all of these are still driving him, even if he can’t overcome his own bias. And a lot of other potential traits are completely absent to the point where if he had them, he would no longer be recognisable as the same character.
Put Ned Flanders in an Iron Man costume, and what do you get?
“Bla bla you aren’t a real writer if you can’t name the specific brand of candy that your 148 year old alien MC enjoyed when they were in their larval form, a backstory that will never be mentioned to your audience ever bla bla diddly diddly bla”
Or… maybe we can calm down a bit.
A lot of writers veer to extremes when it comes to talking about characterisation. They either steer clear of over-analysis and speak of them as if they’re real, with a semi-mysticism that defies schemas and structures — or, the complete opposite, and they fill out elaborate questionnaires about the birthday cake Super Alphonso Fishman had when he was 5 years old and the favourite colour of his great uncle.
Both of these approaches can help writers, but in my view they can be actively dangerous or limiting in certain situations and as learning tools. I feel like maybe 30% of the benefit of such styles of thinking are placebo in part — it gets you thinking more reverently or deeply about your characters, which prompts extra emotional investment and interest in your own mind, often leading to better writing. But what kind of view of humanity does the questionnaire approach follow? In my mind, it prioritises the equivalent of a noun-based conception of humanity — that Tony Stark is not arrogant and brash, precisely, but that Tony Stark has expensive sports cars and secret military technology.
You might immediately say that the two things I just associated with Tony Stark are of course suggestive of a number of personality traits. That’s right. And as a tool to examine a character you’re having difficulty with? It can be great - I would not even recommend against it in all situations, as sometimes you need to parse your ideas through a new lens to escape previous thought-traps and blocks.
Let’s imagine that the sports cars and military tech are the first things we know about Tony Stark in a film - let’s imagine we don’t even see him on screen for ages, we just see the objects and symbols associated with him. How do these affect the audience in that case? Well, they -tell us- about a character’s traits in a way that lets us work out things about them. We might end up making assumptions again (poor tions and umps), but there is a pleasure in well-handled revelations about fictional human beings. We talk about ‘twists’ in grand narrative senses of huge reversals and modifications to how we view X character, but these are like overdoses of the same substance you’re getting throughout an entire narrative structure. We are constantly experiencing micro-twists out of the negative space of the work — we are correcting and building our increasingly detailed spiritual/mental portraits of these people.
You hear two people talking on a bus about their accounting business? It might be boring or neutral. You hear one person having half that conversation on the phone, and talking about legal consequences and deadlines and their fear without ever mentioning the accounting context? It’s going to be fascinating. Withholding information, if done correctly, creates a kind of invented ‘need’.
Sell me this pen
Wolf of Wall Street — Leo DiCaprio asks the table of his friends if anyone can sell him a pen, as a test of sales tactics. Most of the people who get given this? They end up talking about the virtues of the pen, its qualities, the things the pen is. The winner? He asks Leo DiCaprio to write his name down — DiCaprio can’t, as he doesn’t have a pen, so DiCaprio now needs a pen and wants one due to this feeling of a ‘lack’. This feeling that was entirely invented and engineered by the request to write a name down. This is the same thing with characters and their gradual revelation of details — we know a person is ‘incomplete’, and if there is enough interest surrounding other elements, our brains enjoy seeking out information to further complete and refine this person we might otherwise not have cared about.
We like surprises and seeing what people are capable of, how they respond in different situations, how they remain the same. The trope of a character seeming like they will do something nefarious - then doing the good thing instead.
A way of viewing this? The core to ASN?
Characters are more ‘verbs’ than ‘nouns’ (and even the ‘nouns’, the things they have or own or have done in the past, are partially just suggestive of the ‘verbs’ in a sense of what an audience member will find relevant about such info).
When we look behind the Wizard of Oz illusion of reality, characters are theorised rules. We refine our sense of these rules over time, sometimes objectively but almost always subjectively to some extent. Like micro-twists, if they contradict prior information we know about the character? Then if the story is successfully told, the character won’t ‘break’, it’s more that we’ll realise our prior theory didn’t account for both sets of information. We galaxy-brain it into a new character that makes sense with both models.
And if they do break?
Final Season of Game of Thrones Spoilers - you’ll most likely immediately know what I’m talking about here…
(skip down to next section to avoid!)
A certain queen in the final season of Game of Thrones is a good example of this.
You take a character who has been shown to commit troubling actings throughout the show, and then you have her commit a very troubling act in the finale. You might try and justify it by saying that it was foreshadowed all along. But this is a noun-based approach to the character. ‘She has a bunch of luxury cars so when she turns into a luxury car, it makes sense’. No, no it does not.
Every time Danaerys did something nefarious in earlier seasons?
A) We saw an arc leading up to it where Danaerys learned the brutality of the world she was in and the price of trust and gentleness. We were taught, cause and effect, what choices can cost and earn you in this world. So when Danaerys chooses accordingly? We view it as pragmatic and emotionally understandable and potentially linked to her inexperience and need to intimidate enemies.
B) Danaerys did the right thing in the end, and the experience isn’t so much a ‘twist’ on her character altering her, but more of a process of learning and character development that Danaerys herself is undergoing while her more core values still remain the same (or, gradually change but in a way we can follow). Often with these moments, they’re underscored by musical cues, friends and advisors telling her she did badly or she did well, and we get our little reversal.
When Danaerys commits the actions she commits in the finale?
It does not follow the rules - not the rules we’ve established over all those years of watching it.
Danaerys → bad thing (must be: pragmatic, emotionally understandable, potentially linked to inexperience and need to intimidate enemies) → // friends, advisors, background music seem to suggest a need for course correction // → Danaerys tries to make bad thing right and/or learns a lesson for the future, with accompanying friend/advisor/music emotional support.
Here, it’s:
Danerys → burns the shit out of some civilians with music really making it seem bad → no-one suggests course-correction or modification, friends either die-hard for or against → Danaerys doubles down on bad thing and seemingly has amnesia about every single lesson she (and the audience who follow her) have learned in so many seasons.
It upsets us not because civilians died. Who cares about nameless fictional civilians that much? No - she’s not a human being. She’s a character - and what we deeply care about here is the fact someone’s just stuck their hand inside our nicely tended-to fictional model of a person and yanked out all the wires. I guess our rules we spent so long refining, so we could guess and follow the behaviour of this person and enter into the make-believe dream of it all being real — they were just total garbage huh?
She would never do this - and timing is important here. If the action had taken place much earlier in the story in exactly this way, it wouldn’t have been as much of a problem for us - which proves the point about our emotional investment in learning these rules. All later encounters would have been assessed against this early one.
END OF SPOILERS
A character ‘BREAKING’ in some way is usually dependent on narrative timing.
Early ‘out of character’ decisions do not necessarily feel ‘out of character’, as our mental model of the person at hand has not yet solidified. We can’t say they would ‘always’ or ‘never’ do certain things with surety, as we don’t know them well. There is no betrayal, because we don’t love them.
There is a lot that can be said and written here about the implications for how we should write characters in light of all of this - but for me, there is a simple takeaway and simple tool we can use intermittently throughout writing, particularly for characters who aren’t working or who need differentiation from others.
THE FRAMEWORK:
SOMETIMES - anything a character is capable of doing, thinking, saying, acting like, etc, in different situations and different moods (while still obeying the two other categories below).
ALWAYS - anything that is true of a character in 99% of situations. If a character ever contradicts something in this list, there should be a hugely compelling reason for them to do so.
NEVER — anything that a character would never do.
‘Sometimes’ is usually at least double the length of always/never (TIP: think of the character responding to the same event/news/behaviour from someone else on one of the best days of the character’s life, VS the worst).
HOW TO USE IT:
There are a number of exercises and extra ways in which this idea can be used, but this is the framework at its core. It can be deployed at any stage of writing — whether in advance of creating a character, in the early stages, or even as a retrofitted tool to get a better idea of the person you’re already dealing with.
Indeed, these principles will naturally emerge from ‘things’ and status quos you may have already written about for your character. Likewise for the audience, these principles will emerge from and be refined by the ‘things’ the character is associated with in the story.
Another way to view it all? Individual pieces of characterisation and questionnaire-inventory style info in a narrative are all akin to pieces of evidence in a criminal trial. The ASN? That’s the crime itself and the motive for that crime, all wrapped up in one inextricable knot.
It can be incredibly useful also to test this against other characters you have not created but which you enjoyed elsewhere. If you get a ‘strong sense of’ a film or game or book protagonist you love? If you feel like you really know them? I guarantee you’ll be able to write some interesting stuff down for an ASN analysis. And when you get a nice, fairly detailed list? Compare to your own characters and — if you can’t write nearly as much or lack the same complexity — this can really help people figure out ways in which to enhance their own characters.
As for video games…
But Greg, you might ask — didn’t you mention games earlier? And isn’t that workshop you run an interactive-writing focussed class? How could you use a thing like this to make games, where the audience can quite literally control the decisions of characters?
How can you possibly have a ‘character’ like the ones you love following in films or linear games, one you get to know and who surprises you and makes you feel a great deal of emotion, when the player is controlling that person? How can ALWAYS/SOMETIMES/NEVER rules work for someone whose behaviour is, to some extent, down to the audience?
Spoiler for the upcoming, aforementioned game writing prompt article that will not only give you a fun exercise to work through, but also some analysis of a range of classic video game characters as examples: one of the last things I said above is the clue.
“Someone whose behaviour is — to some extent — down to the audience.”
Bonus fun exercise for the D&D and TTRPG players among my readers:
Another fun thing you can try for practice and/or fun? If you play D&D or TTRPGs, have a little side character sheet going where you do a mix of adding ASN traits from scratch and modifying them in response to the way you roleplay your character. If you’ve written something in ‘never’ and have to contradict or alter it, this will be a lot easier earlier on, and likely cause you more pause and hesitation if it’s been some long-held long-written part of that column after months of roleplaying. This? This will show you precisely what an audience member feels at potential late-stage contradictions, but in a way that still feels quite close to a writing-type decision on your part.
This isn’t precisely what I’m going to be talking about in the upcoming game writing prompy follow-up to this piece — TTRPGs have always been their own very strange fun lovable kettle of fish in this strange sea of story types — but in strangeness, we can expose a lot of things about storytelling that might be harder to see elsewhere when we’re blinded by familiarity and convention. So I always recommend such experiences not just for fun, but as a great creative practice to engage in alongside people’s work if they can manage to do so.