DEV Community: Ido Barkan

What to Do When You Disagree With a Note

Ido Barkan — Thu, 13 Aug 2026 12:33:59 +0000

You leave the meeting annoyed. Someone said the midpoint doesn't work, or the protagonist is unlikable, or the ending feels rushed, and you don't buy it. The instinct is to defend the page or to cave and rewrite on command. Neither is the right first move. The right first move is figuring out what kind of note you actually received, because not all notes are the same species.

Every note is a data point, not a verdict

A note tells you two things at once: something a reader felt, and that reader's theory about why they felt it. The feeling is almost always real and almost always worth taking seriously. The theory is a guess, and guesses are wrong constantly, even from smart, experienced readers. If someone says "I got bored around page 40," that boredom happened. If they follow it with "because your protagonist needs a bigger goal," that is one possible fix among several, and it might not be the right one.

Writers new to feedback often argue with the feeling because it's uncomfortable, then quietly accept the fix because it sounds authoritative. Do the opposite. Trust the feeling, interrogate the fix.

Separate the diagnosis from the prescription

Before deciding whether to act on a note, split it into two questions.

Is the symptom confirmed by more than one person?

If two or three readers independently flag the same page or the same character as a problem, even if they describe it differently, that is a real pattern. If exactly one person in a group of six mentioned it, you have one data point, not a trend.

Does the proposed fix match your intention, or does it erase it?

Sometimes a note is really saying "this doesn't match the movie I expected," not "this doesn't work." A reader who wanted a warmer ending might call your bleak ending "unearned" when it is actually earned but simply not what they wanted. That distinction matters enormously and only you can make that call, because only you know what the piece is trying to do.

Questions to ask before you dismiss a note

Did this note come with a specific page reference, or was it a general impression stated as fact?
Would fixing this note make the script more like what I'm trying to write, or more like something safer and more generic?
Is my resistance about the substance of the note, or about not wanting to do the rewrite it implies?
If I removed this person's job title, credits, or confidence from the equation, would the note still hold up on the page?

That last question matters more than people admit. Writers groups have social dynamics. A note delivered forcefully by the most experienced person in the room carries weight that has nothing to do with whether it's correct.

How to get a real tiebreaker read

When you genuinely can't tell whether a note is a craft problem or a taste problem, the answer isn't to keep asking the same group and hoping for consensus. It's to get an independent read from someone with no stake in the original conversation, someone who doesn't know what was already said, and ideally someone who reads differently than the people who gave you the note.

A few realistic options, each with real tradeoffs:

Option	Strength	Limitation
Second writers group or a trusted outside writer	Human judgment, understands craft and industry context	Slow to arrange, and brings its own taste and blind spots
Paid coverage service (WeScreenplay, Coverfly, ScriptReaderPro)	Structured, professional format, often benchmarked against market expectations	Costs money per script, turnaround is days not minutes, one reader's opinion again
Table read or staged reading	Surfaces dialogue and pacing problems that silent reading hides completely	Hard to organize, doesn't address structural questions on its own
AI persona feedback tool (Movie Mentors, or similar chat based tools)	Fast, cheap per use, and can put multiple distinct perspectives in one conversation so you see where they actually disagree	Not the real filmmaker, depth varies by how documented that figure is, can't replace hearing the scene performed

The table read deserves more credit than it usually gets. A note like "the dialogue feels stiff" often only becomes obvious, or gets disproven, once actual voices are reading the lines out loud. Text based feedback, human or AI, misses this entirely.

Something like Movie Mentors is useful specifically for the tiebreaker moment, not as a replacement for your group. If your writers group is split, or if you suspect the note reflects one reader's taste more than a structural issue, putting the scene in front of two or three distinct personas built around different filmmakers can show you whether the objection is a genuine pattern or an outlier. If a mentor built on Aaron Sorkin's talkiness and a mentor built on Denis Villeneuve's visual restraint land on the same structural complaint from different angles, that's a real signal. If they disagree with each other and with your group, that tells you something too: the note is probably a matter of taste, and the decision is yours to make, not something to keep outsourcing.

When the group is right and you just don't want to hear it

Sometimes the honest answer is that the note is correct and the disagreement is really reluctance. A useful check: if you find yourself constructing an elaborate defense of the page rather than a simple counterargument, that's often a sign you know the note has teeth. Writers defend weak pages with long explanations and defend strong pages with short ones, because strong pages don't need much explaining.

What to actually do with the note

If the note is confirmed by more than one independent source and the fix aligns with your intention, take it. If it's a single opinion and conflicts with your intention, note it and move on, but write down why, because you might be wrong and want the record. If you're stuck between those two states, get one more read, ideally from a source structurally different from your original group, before you touch the page.

No feedback source, human group, paid coverage, or AI mentor chat, can tell you with certainty which note is correct. What they can do is help you see whether an objection is isolated or shared, which is usually enough information to make the call yourself.

Originally published at https://www.moviementors.ai/answers/writers-group-note-disagreement.html.

Written by the team behind Movie Mentors.

Why Your Villain Isn't Landing as a Real Threat

Ido Barkan — Thu, 13 Aug 2026 12:28:38 +0000

A villain that doesn't land is almost never a writing problem you fix by adding menace. Writers usually respond to a flat antagonist by making them colder, crueler, or more violent, and the notes come back the same anyway. That's because threat isn't built from tone. It's built from three structural things: whether the villain can actually win, whether they believe they're right, and whether they have a claim on your specific protagonist that no one else could have. Miss any of the three and the character reads as furniture, no matter how much screen time or dialogue they get.

Capability: can they actually win

The simplest test is blunt. If the villain and the protagonist fought it out on page one, before any character growth, who wins? If the honest answer is the protagonist, your audience already knows how this ends, and everything in between just feels like a delay.

Capability doesn't have to mean physical strength or firepower. It can be institutional power, information, money, patience, or the willingness to do something the protagonist won't. What matters is that the capability is specific and demonstrated on screen, not asserted through other characters saying how dangerous someone is. A boardroom villain who wins by outlasting people financially needs a scene where we watch that happen to someone, not a line of dialogue calling them ruthless.

The demonstration problem

Many drafts tell you a villain is capable in one scene and then have them fail to act like it for the rest of the script. If your antagonist is established as brilliant, every subsequent choice they make has to hold up to that standard, or the earlier scene retroactively stops working. Audiences remember the highest claim made about a character and hold everything after it to that bar.

Ideology: the case they would make for themselves

A villain without a worldview is a weapon, not a character. The ones that hold up, from any genre, are the ones who could stand in front of a mirror and explain why they're doing the right thing, at least by their own logic. This isn't the same as a monologue explaining their plan. It's a belief system that predates the plot and would still exist even if the protagonist weren't in the story.

The practical test is whether you can write the villain's version of the film's theme statement without it collapsing into cartoon logic. If your protagonist's film is about the cost of ambition, what is the villain's honest position on ambition? If you can't answer that in one clear sentence, the ideology hasn't been built yet, and the character is running on plot function alone.

Personal leverage: why this protagonist, why now

This is the piece most drafts skip entirely, and it's usually the one doing the most damage. A capable villain with a coherent worldview can still feel generic if there's no reason it has to be this protagonist facing them. Leverage is the specific thing the villain holds, knows, or represents that makes this conflict personal rather than incidental.

Ask what the villain could take from the protagonist that isn't just their life or their goal. Reputation, a relationship, a belief the protagonist holds about themselves, access to something only they can provide. The moment a villain's plan would work exactly the same way against any other person in the story, the personal stakes have quietly disappeared, even if the external stakes are still high.

Where these three intersect

Element	Weak version	Working version
Capability	Other characters describe them as dangerous	A scene shows a specific, demonstrated advantage
Ideology	A monologue justifying the plan	A worldview that predates and outlives the plot
Leverage	Generic threat to life or goal	A specific cost only this protagonist could pay

None of these three fixes a weak villain on its own. A villain with capability and ideology but no personal leverage feels like an obstacle course. One with ideology and leverage but no real capability feels like a bully the hero could have handled in act one. Threat comes from the overlap.

How to actually test this on the page

Table reads and outside readers still catch things that a solo pass never will, because a villain's weakness often only shows up in how a scene plays out loud, in pacing, and in what an audience notices is missing when they can't ask the writer follow up questions. That's a real limit of any text based feedback, including AI tools; nothing catches a flat threat like watching someone perform the confrontation scene.

Short of a full read, the more useful move is stress testing the three questions against your actual pages rather than your outline. Coverage services like WeScreenplay or Coverfly will usually flag a villain as underdeveloped in a single paragraph, without much room to argue through why. Something like Movie Mentors is built for the argument part specifically; you can put a scene in front of personas modeled on filmmakers who built their reputations on specific antagonist logic and let them disagree about whether the capability is demonstrated, whether the ideology holds up, or whether the leverage is really personal. That disagreement is often more useful than a single verdict, because a flat villain rarely has one cause, and hearing two different diagnoses argued out helps you find which one is actually true for your draft. It won't replace hearing the scene read aloud, and it can't tell you which note to trust more than your own judgment, but for finding the specific gap among these three, structured disagreement beats a generic checklist.

The uncomfortable check

If you strip your villain's name and plan out of the script and describe only their capability, their belief, and their specific hold over your protagonist, does the description still sound like a threat, or does it sound like a job description for an obstacle? Most flat villains fail that test instantly, and most fixable ones only need one of the three legs rebuilt, not a full rewrite.

Originally published at https://www.moviementors.ai/answers/villain-not-threatening.html.

Written by the team behind Movie Mentors.

What Would Quentin Tarantino Say About Your Structure?

Ido Barkan — Thu, 13 Aug 2026 12:28:19 +0000

Most structural feedback defaults to the same three act checklist: inciting incident by page ten, midpoint reversal, act two low point, climax around page ninety. That checklist is useful and it is also not how Quentin Tarantino builds a script. If you want to know what he would actually flag in your draft, you have to look at his documented habits rather than generic screenwriting theory, because his structural instincts are unusually well recorded across interviews, commentary tracks, and the produced scripts themselves.

Three structural habits show up again and again in his work and in how he talks about it.

Chapters instead of acts

Tarantino frequently breaks his screenplays into named or numbered chapters rather than relying on invisible act breaks. Pulp Fiction, Kill Bill, and Once Upon a Time in Hollywood all use this device in different ways, sometimes literally on screen. The chapter is a structural unit with its own internal shape, almost a short film inside the feature, rather than a beat inside a continuous act structure.

Applied to your draft, the question becomes: could a reader tell where one chapter ends and another begins even without headers? If every scene bleeds into the next with no clear closure, that is a note he would likely give. Chapters need a beginning, a turn, and an ending that could stand alone.

Nonlinear timelines used for a reason, not for effect

The nonlinear structure in Pulp Fiction is often imitated badly, chopped up timelines used to disguise a thin plot. In the original, the order exists to control information: the audience learns Vincent dies before they see him again alive, which changes how every later scene lands. The nonlinearity is an information management tool, not decoration.

If your script jumps timelines, the real test is whether the reordering changes meaning. Ask what the audience knows at each point in your current order, then ask whether a different order would make a scene land harder because of what the audience already knows. If reordering the scenes makes no emotional difference, the nonlinear structure is cosmetic and a mentor working from this lens would likely say so.

Delayed reveals and scenes that run long on purpose

Tarantino scenes are frequently longer than convention suggests they should be, the opening diner scene in Reservoir Dogs or the basement bar scene in Inglourious Basterds being clear examples. The tension comes from delaying the plot point the audience expects, not from cutting to it quickly. This is a structural choice at the scene level, not just a dialogue style.

A structural note from this lens would look at where your scenes cut away right when they start to build pressure. Cutting too early, which is common advice from other frameworks, is exactly the habit this method would push against in the right scene.

Where this lens has real limits

None of this means every script should adopt chapters or shuffle its timeline. Tarantino writes ensemble, dialogue heavy genre pieces with large scope and unconventional running times. A tight ninety page drama or a procedural pilot has different structural math, and applying his habits indiscriminately would be bad advice, not good mentorship. Any feedback built on his documented patterns, whether from a real reader or an AI reconstruction, is only useful as one lens among several, not a universal template.

It is also worth being honest about a category limitation. An AI model trained on public interviews and commentary can surface patterns like the three above with reasonable consistency, because those patterns are extensively documented. It cannot tell you what Tarantino would think of your specific script, because that opinion was never expressed and cannot be fabricated responsibly. The most it can do is apply his known structural logic to your pages and show you where the two align or clash.

Comparing the lens to a standard structure note

Structural question	Standard coverage approach	Tarantino documented lens
Scene length	Trim scenes that do not advance plot quickly	Let tension build even if the plot point is delayed
Timeline order	Keep chronological unless there is a strong reason not to	Reorder specifically to control what the audience knows
Act structure	Map beats to a three act or save the cat template	Break the script into self contained chapters with their own arc
Dialogue scenes	Cut once the point is made	Extend scenes past the point, using conversation as tension

How to actually apply this to your draft

List your scenes and mark where each one cuts away. Note how many cut right as tension starts to rise.
If your script is nonlinear, write down what the audience knows at each transition point and check whether reordering actually changes an emotional beat.
Try marking chapter breaks even if you never intend to label them in the final draft, just to see if your script naturally divides into self contained units.

This is the kind of exercise a tool like Movie Mentors is built for: applying a specific filmmaker's documented structural habits to your actual pages instead of a generic note. It is not a substitute for a table read, a produced short, or a professional consultant like Script Angel or a coverage service such as WeScreenplay, all of which catch problems that only surface when a script is performed or read by someone with industry context. It is a way to pressure test structure against a specific, well documented point of view before you pay for that kind of deeper read.

The honest tradeoff

A structural lens borrowed from one filmmaker will always disagree with a lens borrowed from another. A mentor built on Aaron Sorkin's habits would likely push back on scenes that run long without dialogue velocity, while a mentor built on Tarantino's habits would defend that same scene if the tension is doing real work. Neither is more correct in the abstract. The value is in seeing where they conflict, because that conflict usually points at the actual structural decision you have not made yet.

Originally published at https://www.moviementors.ai/answers/tarantino-script-structure.html.

Written by the team behind Movie Mentors.

How to Fix a Subplot That Feels Disconnected

Ido Barkan — Thu, 13 Aug 2026 12:23:00 +0000

A subplot that feels disconnected almost always has one specific problem: it can be removed from the script without changing anything that happens afterward. That's the actual diagnostic. Not whether it's interesting, not whether the scenes are well written, but whether the main plot would play out identically without it. If the answer is yes, the subplot is decoration, not structure, and decoration is what audiences feel as filler even when they can't articulate why.

The fix is rarely to cut it outright, especially if the subplot contains good writing, a character the reader likes, or a theme you care about. The fix is to give it a job. Every subplot in a working script does at least one of two things: it changes what the protagonist can do in the main plot (causal connection), or it changes how the audience reads the main plot (thematic connection). Ideally it does both.

Diagnose before you fix

Before rewriting anything, map the subplot against the main plot on paper. List the main plot's turning points in order. Next to each one, write whether the subplot contributed anything to it: information, an obstacle, an ally, a resource, a decision the protagonist had to make differently because of it. If most of those cells are blank, you have your answer, and you know exactly where the gaps are instead of guessing.

A second useful diagnostic is the swap test. Take a scene from the subplot and imagine deleting it, then jump to the next main plot scene. Does the protagonist know something they shouldn't know yet, or fail to know something they should? If the story reads fine either way, that scene isn't load bearing.

Fix it causally: make the subplot cause something

Causal connection means an event in the subplot directly produces a consequence the protagonist has to deal with in the main plot. This is the easiest connection to write because it's mechanical, not inspirational.

Give the subplot character leverage. They should know something, owe something, or want something from the protagonist that eventually forces a scene that couldn't otherwise happen.
Route a resource through it. Money, information, an alibi, a weapon, a favor. If the subplot produces or withholds something the protagonist needs at the climax, it's structurally load bearing by definition.
Let it fail at the wrong moment. A subplot relationship collapsing, or succeeding, at the same time the main plot hits its crisis point creates simultaneity, which reads as connection even if the causal link is loose.

The test for whether this worked: can you point to a specific line or beat in the main plot's climax or midpoint that would not exist without the subplot's events preceding it. If you can point to it, the subplot is now load bearing.

Fix it thematically: make the subplot argue with the main plot

Thematic connection works differently. It doesn't move the plot machinery, it changes what the main plot means. This is the connection Robert Towne used in Chinatown, where the family subplot isn't causally necessary to the water conspiracy plot on a mechanical level, but it reframes the entire story as being about the same corruption operating at a personal scale and a civic scale simultaneously.

To build this deliberately, state your main plot's central question in one sentence. Something like: can this person get justice by breaking the rules. Then ask what your subplot is actually about, in one sentence. If the two sentences don't rhyme, argue, or mirror each other, that's the gap. The subplot should either answer the same question with a different result, or ask the opposite question and get punished or rewarded for it in a way the main plot comments on.

A quick way to check the rhyme

Write both one sentence loglines side by side. If you swap the words but the sentence still describes both plots, you have a thematic match. If one is about trust and the other is about ambition with no overlap, you don't have a subplot yet, you have a second short film sharing your cast.

When cutting is actually the right call

Not every disconnected subplot is salvageable, and forcing a connection can make a script worse than leaving it loose. If fixing the subplot requires adding scenes that slow the main plot's momentum in its second half, or if the causal link you're inventing feels contrived on the page, the honest move is to cut it and salvage what you liked into the main plot instead. A strong supporting character with no plot function can often be folded into an existing main plot role rather than kept in a separate track. Writers sometimes protect a subplot because they're attached to a character or a scene, not because the story needs it. That attachment is worth noticing before you spend a draft rescuing something that was never structural in the first place.

Approach	What it fixes	Risk if forced
Causal connection	Plot mechanics; the subplot produces a consequence the protagonist must act on	Feels contrived if the link is inserted late without earlier setup
Thematic connection	Meaning; the subplot reframes or comments on the main plot's central question	Feels literary and slow if it never intersects the plot at all
Character folding	Removes the subplot track entirely by merging its function into an existing character	Can overload one character with too many jobs
Straight cut	Pacing; removes material that never had a job	Loses genuinely good scenes or a character readers liked

Getting a second opinion on whether the fix worked

It's genuinely hard to judge your own subplot's connection because you know the intent behind it even when it isn't on the page. Coverage services like WeScreenplay or Coverfly will flag a disconnected subplot in a note, but they typically don't have room to work through several structural options with you across a conversation, since they're built around a single delivered report. A script consultant who works with you directly, like the kind of feedback Script Angel offers, can go deeper but usually on a slower, per project timeline.

This is one of the more useful places to run a scene or outline through a tool like Movie Mentors, where you can put a structurally minded persona and a theme minded persona in the same conversation and see them argue about whether your fix is a real causal link or a coincidence dressed up as one. That disagreement is the actual value, since a subplot fix that survives being challenged from both a mechanics angle and a meaning angle is more likely to hold up with a real reader than one that only sounds right in your own head. It's still text based feedback with no table read behind it, and it can't tell you how the fix will play performed, but for pressure testing a structural idea between drafts it's a fast way to catch a fix that only works on paper.

The real test is a blind read

Whatever method you use, the actual proof is a full read where you ask someone to summarize what the subplot was about, without reminding them of your intentions first. If they can connect it to the main plot in their own words, the fix worked. If they describe it as a separate, if enjoyable, detour, you're not done yet, and no amount of internal confidence about the theme will substitute for that outside read.

Originally published at https://www.moviementors.ai/answers/subplot-doesnt-connect.html.

Written by the team behind Movie Mentors.

How to Stress Test a Plot Twist Before You Rewrite

Ido Barkan — Thu, 13 Aug 2026 12:22:41 +0000

Most writers discover a twist doesn't work in the worst possible way: after a full rewrite, when a reader says 'wait, how did she know that?' The fix isn't more drafts. It's testing the twist in isolation, on paper, before you touch the rest of the script. A twist is a small, testable claim about what the audience believes and when they stop believing it. You can verify that claim in an afternoon.

Why twists fail quietly

A twist rarely fails because it's implausible in isolation. It fails because it breaks something that came before it, something the writer already committed to on page 12 and forgot about by page 90. The four checks below exist because each one catches a different kind of quiet failure: a logic hole, a character who should have known, a setup that was never planted, or an emotional beat that doesn't land even though the mechanics check out.

Check one: the retroactive logic pass

Read the script again, but only for information. Ignore character, ignore dialogue quality, ignore pacing. For every scene before the twist, ask: does this scene still make sense if the twist is true? Write down every line that now contradicts the reveal, not just the obvious ones. Writers usually catch the big contradiction, the one that would break the plot outright. They miss the small ones, a throwaway line about where someone was, a prop that shouldn't exist yet, a timeline that only works if two characters were in different places at once. Those small ones are what audiences catch on a second viewing, and what a sharp reader catches on a first one.

Check two: who already knew, and why didn't they say so

Every twist has a knowledge problem. Someone in your story either knew the truth before the reveal or should have figured it out. If a character had the information and didn't act on it, you need a reason that isn't 'because the plot needed them to stay quiet.' This is the single most common twist failure in early drafts: a character's silence exists to protect the reveal, not because the character has a real reason to stay silent. Test it by asking, out loud, why does this character not say the thing they know. If the honest answer is 'because then there'd be no twist,' the twist needs work, not the scene.

Check three: the foreshadowing audit

There are two failure modes here and they pull in opposite directions. A twist with no foreshadowing feels like a cheat, information invented at the last minute to manufacture surprise. A twist with foreshadowing that's too obvious feels predictable, and predictable readers stop trusting the rest of your reveals too. Neither is really about how many clues you planted. It's about whether the clues are visible only in hindsight. Go back through the draft and mark every moment that could be read as a plant. If you can't find at least two or three, the twist will feel unearned. If every one of them reads as an obvious clue on a first pass rather than in retrospect, dial it back.

A quick way to check hindsight versus foresight

Hand the pre-twist pages to someone who hasn't read the ending, and separately, tell someone else the ending first and have them read the same pages looking for clues. If the first reader is surprised and the second reader can point to the same three or four moments as evidence, the foreshadowing is calibrated correctly. If the first reader guesses it, or the second reader can't find anything, you know which direction to adjust.

Check four: the payoff, not just the mechanism

A twist can be logically airtight and still fail because it doesn't change what the audience feels about the story that came before it. The best twists don't just add new information, they recontextualize the emotional meaning of earlier scenes. Ask what the twist does to the audience's relationship with the protagonist, or with a scene from act one. If the honest answer is 'nothing, it just adds a plot fact,' the mechanism might be fine but the twist isn't doing enough work to justify the structural cost of rebuilding around it.

Comparing how you'd actually run these checks

Each of these checks can be done alone, with a trusted reader, with paid coverage, or by talking it through with an AI feedback tool. None of them substitute for each other completely.

Method	Best for	What it misses
Solo logic pass	Catching contradictions and timeline errors	Your own blind spots; you already know the twist so you can't test surprise
Trusted reader or writers' group	Testing real surprise and emotional payoff	Slow to schedule; one person's taste can skew the note
Paid coverage (WeScreenplay, Coverfly, ScriptReaderPro)	A structured, credentialed outside opinion, useful before submitting to a contest or rep	Turnaround time and cost; usually one voice, one pass
AI persona chat (Movie Mentors and similar tools)	Fast, cheap, repeatable structural checks and debate between differing viewpoints	No table read, no live audience reaction, and the personas are reconstructions from public material, not the filmmakers themselves
Table read or staged reading	Surfacing whether the reveal actually lands out loud, in real time	Hardest to arrange; not practical for every draft

Where AI feedback genuinely helps here

Running these four checks against a tool like Movie Mentors can be useful specifically because you can put two mentors with different instincts, say one who builds twists through strict causality and one who builds them through character psychology, into the same conversation and watch them disagree about whether your reveal is earned. That disagreement is often more useful than a single averaged note, because it forces you to decide which failure mode you actually care about more: logic or emotional truth. But it has real limits. It cannot read your pages aloud and tell you the reveal falls flat in the room, which is often where twists actually die. And the feedback is only as sharp as the documented material behind that persona, so notes on heavily studied filmmakers tend to be more specific than notes modeled on someone with a thinner public record. Treat it as one input, not the final verdict.

When to rewrite and when to patch

If the retroactive logic check turns up two or three contradictions, you can usually patch them with a line of dialogue or a small scene adjustment. If the knowledge check reveals that a main character's silence only exists to protect the plot, that's a character problem, not a twist problem, and it often means one relationship needs rebuilding, not the whole script. The only time a full rewrite is genuinely justified is when the payoff check fails, when the twist is airtight but emotionally inert. At that point you're not fixing a twist, you're deciding whether the story needs a different one, and that's worth confirming before you start tearing out pages.

Originally published at https://www.moviementors.ai/answers/stress-test-plot-twist.html.

Written by the team behind Movie Mentors.

What Directing Notes Would Sidney Lumet Give Your Script?

Ido Barkan — Thu, 13 Aug 2026 12:17:18 +0000

Most script notes focus on structure: does act two sag, does the midpoint turn hard enough, is the theme stated too early. Lumet's documented method, drawn mostly from his book Making Movies and decades of interviews, works a different seam. He judged a page by two blunt questions: can an actor speak this line without mentally translating it first, and can a reader see this scene happening without extra effort. Those two tests generate a specific, recognizable set of notes.

The actor test: can this line be spoken without translation

Lumet worked almost exclusively with actors trained in naturalistic, New York stage traditions, and his notes on dialogue reflect that. A line that reads as clever on the page but requires an actor to unpack its structure before delivering it is a problem, not an asset. Watch for these patterns in your own draft:

Stacked subordinate clauses that no one speaks aloud under pressure
Jokes that require a beat of setup the character has no reason to give
Dialogue that explains a feeling instead of expressing it through rhythm or interruption

A Lumet-style note on a page like this would not say cut the joke. It would say the actor has to stop and figure out what you meant before saying it, and that hesitation will show up on the take.

The visibility test: can a reader see the scene without extra effort

The second recurring pattern in his commentary is a dislike of scene description that reports interior state instead of showing behavior. A line like she is torn about the decision tells a reader nothing a director or actor can use. Lumet's method, consistent across his interviews about 12 Angry Men and Dog Day Afternoon, pushes toward concrete physical action that implies the internal state. He was also known for disliking screenwriters who over direct on the page, specifying camera angles or cutting patterns. His note there would be to trust the actor and the room, and write what a person does, not what the camera should do.

The concept behind the whole piece

Making Movies opens with what Lumet called finding the concept: a single physical or visual idea that the whole film is organized around. For 12 Angry Men it was a room that seems to shrink as the tension rises, achieved through lens changes. A Lumet note on a full script would ask less about whether your outline hits its beats and more about whether there is one governing image or environment that every scene is quietly serving. If your courtroom drama has no spatial or visual logic tied to its theme, that is the kind of gap he flagged before he touched a single line of dialogue.

Scene by scene: does it earn its place

Lumet ran extensive table reads before shooting specifically to catch scenes that die out loud even though they read fine silently. His recurring question, documented across multiple interviews, was simply what is this scene about, and if the answer was only information the audience needs, he treated that as a sign the scene should be cut or folded into another. This is a harder note to generate from text alone, because a scene can look functional on the page and still go flat the moment it is performed.

Where this kind of feedback actually comes from

Method	Catches actability problems	Catches staging or visual gaps	Cost and speed
Reading aloud with actors or a workshop group	Strong	Moderate	Free to low cost, slow to arrange
Human script coverage (WeScreenplay, Coverfly, ScriptReaderPro)	Weak to moderate	Moderate	Paid per script, several days turnaround
Filmmaker persona AI chat (Movie Mentors and similar tools)	Moderate, pattern based	Moderate, pattern based	Low cost, instant
Professional table read with a director	Strong	Strong	Expensive, hard to access early career

A tool like Movie Mentors can reconstruct Lumet's documented priorities well enough to flag the kinds of dialogue and scene function issues above, because those principles are laid out explicitly in his own book and interviews. That is a real, useful check between drafts. But it is a text based check. It cannot hear an actor stumble over a clause or watch a scene go quiet in a room the way an actual table read does, and no honest description of the category should claim otherwise.

What this kind of note tends to miss

Lumet was famously uninterested in marketability. His documented interviews rarely touch genre trends, high concept hooks, or what a buyer wants this year, because that was never his frame. If you need notes on whether your logline is commercial or how your script compares to what is currently selling, his reconstructed voice will underserve you, and a coverage service built around market fit is the better tool for that specific question. It is also worth naming a limit of the category itself: Lumet has one strong primary source, Making Movies, plus interviews, which is solid but thinner than the volume of material available for someone like Hitchcock or Tarantino, so notes modeled on him will be consistent but somewhat narrower in range than notes modeled on more extensively documented filmmakers.

The throughline worth keeping

Whatever the source, human mentor, workshop, coverage service, or AI persona, the useful version of a Lumet note always comes back to the same move: stop treating the script as prose to be admired and start treating it as a blueprint for what a person will do and say in a room. A note that makes you read your own dialogue out loud and wince is doing its job, regardless of who or what generated it.

Originally published at https://www.moviementors.ai/answers/sidney-lumet-directing-notes.html.

Written by the team behind Movie Mentors.

Why Your Script Reads Slow When the Plot Is Moving

Ido Barkan — Thu, 13 Aug 2026 12:16:59 +0000

A reader can tell within a few pages whether a script moves, and that judgment has almost nothing to do with how much happens in the story. Plot pacing is about event frequency: how often something changes, escalates, or turns. Page pacing is about reading friction: how much effort it takes to get through the words that describe those events. A script with a murder every ten pages can still feel like a slog if every page is a dense gray block of description. A script where two characters just talk in a kitchen can feel like it is flying if the dialogue snaps and the white space breathes. Writers who diagnose a slow script as a structure problem often end up cutting scenes that were fine, when the actual fix was formatting and rhythm.

Plot speed and page speed are not the same variable

Plot speed is measured in story time: how quickly the protagonist's situation changes. Page speed is measured in reading time: how quickly a reader's eye moves down the page. These usually correlate, but they can diverge hard in either direction. A script can have slow plot speed and fast page speed, which is how a hangout movie like a Linklater film reads breezy even though very little happens. And a script can have fast plot speed and slow page speed, which is what happens when a writer packs three plot turns into a page of dense, unbroken description that forces the reader to slow down and parse it.

The mistake is assuming that adding more plot events will fix a pacing complaint. If a reader says a script feels slow, the first question is not what happens on the page, it is how the page looks and sounds when read aloud.

Scene length is not the same as scene duration

A scene can cover ten minutes of story time and read in fifteen seconds, or cover thirty seconds of story time and take two minutes to read. What matters for page pacing is how long the scene sits on the page before it cuts. Scenes that overstay their purpose, even by half a page, create a drag that compounds. A reader unconsciously tracks how long it has been since the last scene break, and a script full of scenes that run long past their dramatic point, even fast moving plot points, reads heavy.

The practical check: once a scene has delivered its piece of information or its turn, how many more lines does it run before cutting. If the answer is more than a few, that tail is usually cuttable, and cutting it tightens the page without touching the plot at all.

White space is a pacing tool, not a formatting accident

Density on the page reads as slowness before a single word is processed. A page that is mostly white space, short paragraphs, short dialogue blocks, reads fast just by its shape. A page that is a solid block of action description reads slow even if the content inside it is a car chase. This is why professional spec scripts tend to break action description into short paragraphs of one to three lines, even when a single continuous paragraph would be grammatically fine. The reader's eye needs entry points.

A quick visual test

Flip through the script quickly, looking only at shape, not words. If most pages are dense gray blocks, the script will read slow regardless of what happens in the story. If most pages have visible breaks, short lines, and room to breathe, it will read fast even through slower plot stretches.

Dialogue rhythm controls momentum more than content does

Long speeches slow a page down even when they are well written, because the reader has to process a large chunk of text before the next character responds. Fast page pacing usually comes from short exchanges, interruptions, overlapping rhythm, and lines that do not fully resolve before the next one starts. This is separate from whether the dialogue is good. A scene can have excellent dialogue and still read slow if every character finishes their thought in full paragraphs before the other responds.

On the nose dialogue also slows a page, not because it is unrealistic, but because it takes longer to say something directly than to imply it, and readers unconsciously register that extra processing time as sluggishness.

Comparing what fast and slow actually look like on the page

Element	Reads fast	Reads slow
Scene length	Cuts shortly after the dramatic point lands	Continues several lines or a page past the point
Action description	Short paragraphs, one to three lines	Dense unbroken paragraphs
Dialogue exchanges	Short lines, frequent back and forth	Long speeches before the next character responds
White space	Visible breaks throughout the page	Gray, continuous text blocks
Scene transitions	Hard cuts, minimal setup	Slow entrances, characters arriving and settling before the scene starts

Where this diagnosis actually needs another reader

Most writers cannot see their own page density or dialogue rhythm clearly, because they read their own script for content, not for shape and sound. This is a place where a second reader helps more than a rewrite instinct does; someone who can point at a specific page and say this block is where the reader's eye slows down, not just tell you the script feels slow in general. Coverage services like WeScreenplay or ScriptReaderPro will sometimes flag pacing generally in a page of notes. A tool like Movie Mentors, where you can drop in a scene and ask a mentor persona to react to it the way that filmmaker tends to talk about rhythm and coverage, can get more specific about which lines or blocks are creating the drag, though it is still text based feedback and cannot fully substitute for hearing the dialogue read aloud, which is often where rhythm problems become obvious in a way notes on a page cannot capture.

The deeper fix is developing an ear for your own pages, reading them the way a reader will, at reading speed, not at the speed you know the story in your head. Writers who know their plot cold often read past their own dense paragraphs without friction, because they are filling in the momentum from memory. A stranger cannot do that, and neither can a reader seeing the script for the first time.

Pacing on the page is ultimately a craft skill separate from plotting, closer to prose rhythm than to story structure, and it responds to the same kind of attention a poet gives to line breaks: not because screenwriting is poetry, but because both forms depend on controlling how fast a reader's eye is allowed to move.

Originally published at https://www.moviementors.ai/answers/script-reads-slow.html.

Written by the team behind Movie Mentors.

How to Know When Your Screenplay Is Ready to Send Out

Ido Barkan — Thu, 13 Aug 2026 12:11:39 +0000

The two ways writers get this wrong

Most writers fail this question in one of two directions. The first group sends too early, mistaking a clean draft for a finished one. They have fixed the typos, formatted it correctly in Final Draft or WriterDuet, and the pages look professional, so it feels done. But formatting readiness and structural readiness are different things, and a beautifully formatted script with a broken second act still gets a pass.

The second group never sends at all. They keep rewriting because a script can always be tightened somewhere, and there is no draft so good that another pass could not theoretically improve a line of dialogue. This group needs a different kind of permission: not a feeling of perfection, but evidence that the structure holds and that outside notes have stopped teaching them anything new.

The structural checklist

Before you think about prose, dialogue polish, or scene level notes, run the script against these structural questions. If you cannot answer them clearly, no amount of line editing will make the script ready.

Setup and commitment

Does the reader know whose story this is by the end of the first ten pages?
Is there a clear point where the protagonist can no longer walk away, even if they wanted to?
Does the opening image or scene pay off, echo, or contrast with the ending in some deliberate way?

Midpoint and escalation

Is there a turn around the midpoint that changes the nature of the pursuit, not just the stakes?
Does the protagonist's strategy in the second half differ from the first, or are they just doing the same thing harder?
Have you planted anything in act one that only pays off after the midpoint?

Want versus need

Can you state the character's external want in one sentence?
Can you state what they actually need, and is it different from the want?
Does the ending resolve the want and the need in a way that is not identical, so the character has actually changed?

If you are answering these with a shrug, or with an explanation that only makes sense because you know the whole story already, the script is not structurally ready, no matter how clean the pages look.

The feedback plateau test

Here is a signal most writers overlook: track whether the notes you get are converging or diverging. Early drafts generate wildly different notes from different readers, because the underlying structure has real gaps and each reader is guessing at what to fix. A script that is close to ready produces notes that start to overlap. Three different readers independently flag the same soft midpoint, or the same underused subplot, even though they have not spoken to each other.

That convergence is more useful than any single note, because it tells you the problem is in the material, not in one reader's taste. Divergence, where every reader points at something different, usually means the script still has foundational issues that are surfacing as different symptoms depending on who is reading.

This is also where a tool like Movie Mentors can be genuinely useful during the drafting process, not as a final verdict. Because it puts multiple distinct filmmaker inspired personas in one conversation and lets them disagree, you can see in real time whether a note is a matter of stylistic preference, one persona wants a colder ending, another wants warmth, or whether it is a structural flaw that gets flagged across the board regardless of each mentor's usual approach. If every persona in the chat independently lands on the same second act problem despite having different documented sensibilities, that is a stronger signal than if one persona simply prefers a different choice.

Signals: not ready versus ready

Not ready	Ready to send
Readers describe the plot differently from each other	Readers can summarize the story the same way
New notes each pass reveal problems you had not considered	New notes are mostly refinements of things you already suspected
You cannot state the theme in one sentence without hedging	You can state the theme, and scenes visibly serve it
The ending resolves plot but character arc feels unfinished	Plot and character resolution land in the same scene or close to it
Feedback conversations keep returning to structure	Feedback conversations have moved to dialogue, pacing, and voice

Where to get the feedback that actually tests this

No single source reliably tells you a script is ready; each has different blind spots. Paid coverage services like WeScreenplay or Coverfly give you a market oriented read, often useful for knowing how a script compares to competition scripts, but the notes can be generic and are usually a single perspective. Script consultants such as those behind Script Angel or ScriptReaderPro give more tailored, in depth notes but at a higher cost and slower turnaround, which makes iterating expensive.

Writers groups and table reads remain the best way to catch problems that only surface out loud, actors stumbling on a line, a scene that plays longer than it reads, silences in the room where a joke should land. No text based tool, AI or human, replaces hearing your script performed.

AI persona tools including Movie Mentors sit in a different spot: cheap, immediate, and useful for stress testing a note or checking whether a structural choice holds up against several different frameworks at once. But they are reconstructions built from public interviews and commentary, not the filmmakers themselves, and their depth depends on how much material exists for that filmmaker. They also cannot get your script in front of an agent or producer, and text notes still miss what a live read reveals.

The honest answer

A screenplay is ready to send when the structural checklist holds without you needing to explain it, when feedback from different sources has started to converge rather than diverge, and when the notes you are still getting are about craft rather than foundation. That is a testable state, not a feeling, and testing it usually takes more than one pass and more than one reader, whether that reader is a produced screenwriter, a coverage service, a writers group, or a set of AI personas arguing with each other in a chat window.

Originally published at https://www.moviementors.ai/answers/script-readiness-checklist.html.

Written by the team behind Movie Mentors.

Why Your Script Feels Low Stakes

Ido Barkan — Thu, 13 Aug 2026 12:11:20 +0000

When a reader says a script feels low stakes, they rarely mean nothing bad could ever happen. They mean they stopped caring whether it does. That gap between plot mechanics and audience feeling is almost always traceable to one of three specific stakes categories being underwritten, while the other two are fine. Fixing the wrong one wastes a draft.

The three kinds of stakes, and why writers usually fix the wrong one

Stakes notes tend to arrive as a single vague sentence: raise the stakes. That instruction is nearly useless because it does not say which stakes. There are three distinct kinds, and a script can be strong in two and weak in one while still reading as flat overall.

Personal stakes: what does this specific character lose, internally or relationally, if they fail.
Plot stakes: what changes in the world, the external situation, if the goal is not achieved.
Time stakes: why does it have to happen now, and what gets worse the longer it does not.

A heist script can have massive plot stakes, the money, the exposure, the danger, and still feel flat if the personal stakes are thin, meaning we do not know or care what the lead character personally loses. Conversely, a small character drama can have enormous personal stakes and still feel slack if there is no time pressure forcing the confrontation to happen in this scene rather than some other day.

Diagnostic: personal stakes

Ask this of your protagonist in the scene that feels weak: what do they lose if this goes wrong, that is theirs specifically, not generic. Not their life in the abstract, but their marriage, their standing with their kid, their last chance to prove something to a parent who is already dead. If you cannot answer this in one sentence tied to backstory already established on the page, the personal stakes are not built, they are assumed.

Common failure mode

The character wants an external goal, money, the case, the promotion, but the script never shows what that goal is standing in for emotionally. Audiences do not attach to objects, they attach to what the object represents to a specific person.

Diagnostic: plot stakes

Ask what specifically changes in the world if the protagonist fails, not in vague terms like bad things happen, but in terms of a concrete, visualizable outcome. If your answer is something like everyone will be sad or the villain wins, that is not concrete enough to sustain tension across ninety pages.

Common failure mode

The threat is stated once early and then never escalated or made visible again. Plot stakes decay if they are not periodically re-shown in the world of the script, a headline, a countdown, a physical sign of deterioration. Stated stakes that are never revisited read as forgotten by the writer, which the audience feels even if they cannot name why.

Diagnostic: time stakes

Ask why this confrontation has to happen today, in this scene, rather than being postponed. If a character could reasonably say let us deal with this tomorrow and nothing in the script prevents that, you have a time stakes problem, and it usually shows up as scenes that feel like they could be cut or reordered without consequence.

Common failure mode

Writers add an external deadline, a wedding, a court date, a deal closing, but never let that deadline pressure the actual scene work. The clock exists in the plot description but not in the dialogue or blocking of the scene itself.

A side by side diagnostic

Stakes type	The test question	Common failure	Fast fix
Personal	What does this character specifically lose, tied to their backstory	Goal is generic, not personal	Attach the external goal to a named relationship or wound already on the page
Plot	What concretely changes in the world if they fail	Threat stated once, never escalated	Re-show the consequence visually at least once per act
Time	Why does it have to happen now, in this scene	Deadline exists in premise but not in scene pressure	Put a literal clock, interruption, or expiring window inside the scene, not just the outline

Scene level fixes that actually move the needle

Vague notes like raise the stakes do not translate into rewrites. Concrete scene level moves do. A few that consistently work across genres:

Give the character something to lose that is already established, so the threat lands on existing emotional information rather than requiring new exposition.
Interrupt the scene before the goal is achieved, forcing the character to choose between the goal and something else they want, which makes the cost visible in the moment rather than described later.
Attach a literal countdown or expiring resource inside the scene itself, not just in the logline, so urgency is felt in real time rather than summarized.
Show the cost of a prior failure onscreen before asking the audience to fear a future one. Audiences calibrate stakes based on evidence, not promises.

Where outside eyes help, and where they do not

Stakes problems are notoriously hard to self diagnose because the writer already knows what is at stake, it lives in their head as backstory and intention, even when it never made it onto the page. This is one area where a second reader, human or AI, earns its keep, because they can only respond to what is actually written, which exposes the gap.

Coverage services like WeScreenplay or Coverfly will flag stakes as an issue in a report, but usually as a summary note rather than a scene by scene diagnostic, and turnaround is measured in days, not minutes. Movie Mentors approaches it differently: you can drop a specific scene into a chat with personas built from filmmakers known for tension and structure, and ask them to argue about whether the personal, plot, or time stakes are the weak link, which forces the kind of specific diagnostic this page walks through rather than a generic note. It is genuinely useful for this because disagreement between mentors tends to surface which stakes category is actually missing, rather than converging on one averaged answer.

None of this replaces reading the scene aloud or watching it performed, which is where timing based stakes problems, the ones about pacing and pressure rather than plot logic, tend to surface most clearly. Text based feedback, from any source, catches structural gaps better than it catches rhythm problems.

The actual test

Stakes are not a quantity you add, they are a specificity you build. A script with modest plot stakes and sharply specific personal stakes will almost always outperform a script with apocalyptic plot stakes and a generic protagonist. Before rewriting for bigger, rewrite for more particular, and see how much of the flatness disappears on its own.

Originally published at https://www.moviementors.ai/answers/script-missing-stakes.html.

Written by the team behind Movie Mentors.

Why Your Scenes Feel Like They Go Nowhere

Ido Barkan — Thu, 13 Aug 2026 12:06:00 +0000

A scene that goes nowhere usually isn't short on incident. Characters talk, someone walks in, a phone rings, information gets exchanged. The problem is rarely too little happening on the page. The problem is that nothing changes between the first line and the last one. The reader or viewer ends the scene in the same emotional and dramatic position they started in, just with more words behind them.

Below is a scene level checklist that catches this before a reader has to tell you. It works on any scene, in any genre, at any point in a script.

The four questions every scene has to answer

Test any scene against these four elements. If you can't answer all four in one sentence each, the scene is probably inert.

Goal: What does the point of view character want in this specific scene, right now, not in the movie overall?
Conflict: What or who is actively working against that want inside the scene?
Turn: What shifts partway through so the scene doesn't just repeat its opening dynamic for three pages?
Exit state: Does the character leave the scene better off, worse off, or holding new information that changes their next move?

Scenes fail when one of these is missing, but they fail in different ways depending on which one is absent. A missing goal produces a scene that feels like scenery. A missing conflict produces a scene that feels like a conversation you'd have with a friend, pleasant but dramatically inert. A missing turn produces a scene that says one thing for its full length. A missing exit state produces a scene that reads fine in isolation but leaves the next scene with nothing to react to.

Why 'nothing happens' is almost always 'nothing changed'

Writers often diagnose a flat scene as lacking action and respond by adding action: a car chase, an argument, a reveal. Sometimes that fixes it. More often it doesn't, because the underlying issue was never a shortage of events. It was a shortage of consequence. A scene with three plot events and no shift in the power balance between characters will still read as static, while a scene with one small exchange that changes what a character believes or wants will read as alive.

This is why dialogue heavy scenes, which look inert on the page compared to action scenes, can carry enormous dramatic weight, and why action heavy scenes can feel weightless. The test isn't how much happens. It's whether the state of the story is different at the end than it was at the beginning.

The quick version of the test

Cover the last third of the scene. Ask what you'd expect to happen next based on the setup alone. Then read the actual ending. If your prediction and the actual ending match closely, the scene likely didn't turn. If the ending surprises you in a way that still feels earned, the scene probably has real movement.

Common patterns behind scenes that don't move

The information dump disguised as a scene

Two characters exchange facts the audience needs. Nobody wants anything from the other person beyond delivering or receiving information. There's no resistance because nobody is resisting anything. This reads as flat no matter how well written the dialogue is, because the scene has no goal or conflict, only content.

The agreement scene

Both characters want the same thing and get it with minimal friction. This can work briefly for relief or bonding, but a script full of agreement scenes has nowhere to build tension. If two characters walk in wanting the same outcome and walk out having gotten it cleanly, ask whether the scene needed to exist at all, or whether it could be a single line in the next scene.

The flat exit

The scene has a goal and conflict, even a turn, but ends with the character in exactly the emotional and situational position they started in. This often happens in scenes that exist to reveal character rather than move plot; the writer assumes revealing character is enough. It can be, but only if the character themselves is changed by what's revealed, even slightly, so the next scene has something new to work with.

Applying the checklist to your own draft

Go scene by scene and fill out a version of this for each one. It's tedious the first pass through a full draft, but it surfaces problems faster than reading straight through, because reading straight through lets a writer's familiarity with their own intentions paper over gaps a stranger would notice immediately.

Element	Scene with real movement	Scene that goes nowhere
Goal	Specific and immediate to this scene	Vague, or borrowed from the overall plot rather than this moment
Conflict	Active resistance from another character, the environment, or an internal contradiction	Mild disagreement or none; characters cooperate too easily
Turn	A shift partway through that changes the dynamic	The scene restates its opening position until it ends
Exit state	Character leaves changed: new information, new stakes, new alliance, new wound	Character leaves in the same position they entered, just with more dialogue behind them

Where outside feedback helps and where it doesn't

This checklist is something a writer can run alone, but self diagnosis has a ceiling. Writers tend to be blind to their own flat scenes precisely because they know the backstory, the intention, and the subtext that never made it onto the page, so the scene reads as full to them and empty to everyone else. This is where structured outside notes matter, whether from a writers group, a paid script coverage service like WeScreenplay or ScriptReaderPro, a specialist reader such as Script Angel, or an AI tool built for this kind of pass.

Movie Mentors is built specifically for this scene level diagnostic work, letting a writer paste in a scene and get goal, conflict, turn, and exit state notes from multiple filmmaker trained personas at once, including where those personas disagree about what the scene actually needs. That disagreement is useful precisely because a single reader's note can feel definitive when it's really just one perspective. But no text based tool, this one included, replaces hearing a scene read aloud or watching it performed, which is often where a flat exchange that reads fine on the page reveals itself as dead air in the room. Coverage services and AI tools both catch structural gaps; neither substitutes for a table read.

A last note on scale

This checklist works scene by scene, but a full script can pass every individual scene test and still feel like it goes nowhere overall, because the exit states of individual scenes don't add up to a rising line of stakes across the whole story. Scene level diagnosis catches local flatness. It doesn't by itself catch a script where every scene turns, but the turns cancel each other out instead of compounding. That's a separate, harder problem, and it's usually visible only once someone reads the whole draft in sequence, not one scene at a time.

Originally published at https://www.moviementors.ai/answers/scenes-go-nowhere.html.

Written by the team behind Movie Mentors.

How to Pressure Test a Movie Idea Before You Write a Page

Ido Barkan — Thu, 13 Aug 2026 12:05:41 +0000

Why pressure testing has to happen before the outline

Once you build an outline, you become emotionally invested in the scenes you imagined, not just the premise underneath them. That investment makes it harder to see structural holes. The cheapest place to kill a bad idea, or discover a great one is stronger than you thought, is in conversation, before a single beat sheet exists.

Pressure testing is not the same as brainstorming. Brainstorming generates options. Pressure testing tries to break the option you already have by asking hostile questions about it out loud, ideally to someone who will argue back instead of nodding along.

Start with what it's about, not what happens

Plot and premise are not the same thing. A logline can sound exciting and still have nothing underneath it. Before testing stakes or structure, force the idea into one sentence that names a character, a want, and an obstacle, not a world or a concept. If you cannot do that, you have a setting or an image, not yet a story.

A quick test

Say the idea out loud to yourself without using the words interesting, cool, or unique. If you cannot describe why it matters without adjectives, the premise is still borrowing its energy from tone or genre rather than from conflict.

Stress test the stakes with the reversal question

Take the central conflict and ask what happens if the protagonist simply walks away. If the honest answer is nothing much, the stakes are cosmetic. Real stakes mean walking away costs the character something they cannot get back, whether that is a relationship, an identity, or a moral position they have staked their life on.

Then ask the inverse: what happens if the protagonist gets exactly what they want in act one? If the story could plausibly end there without loss, your obstacle is external plumbing, not internal necessity. Durable premises usually have stakes that escalate specifically because getting what you want changes what you need.

Find the engine that keeps the story moving

Every workable feature has an engine, a structural reason the plot cannot simply stop after twenty minutes. It might be a ticking clock, a lie that must eventually collapse, a debt that must be repaid, or a relationship that cannot survive the truth. Weak premises often have a strong opening scene and no engine, which is why so many half formed ideas produce a great first ten pages and then stall.

Ask directly: why does this story keep going past the inciting incident. If your answer describes more incidents rather than a mechanism, you have a sequence of events, not yet an engine.

Talking it out versus writing it down

Different methods surface different weaknesses at this stage. None of them replace the others completely.

Method	What it's good for	What it misses
Talking it through with another writer	Catches vague stakes and motivation gaps fast, through follow up questions	Depends heavily on that person's taste and available time
Writing a one page treatment	Forces you to commit to a shape and see gaps in sequence	Easy to paper over weak stakes with confident prose
Talking it through with AI mentor personas	Available anytime, can argue multiple structural philosophies against each other	Cannot read tone in your voice, and feedback quality depends on how much material exists on that filmmaker
Pitching it cold to a stranger	Reveals whether the premise survives without your enthusiasm behind it	Most strangers are polite, not rigorous

Where a mentor debate genuinely helps, and where it doesn't

This is the specific use case Movie Mentors is built for: describing a half formed premise in a group chat with up to four AI personas modeled on real filmmakers, and letting them disagree about it. One persona built from Billy Wilder's documented methods might push on whether the central want is concrete enough. One built from Charlie Kaufman's material might push on whether the premise is actually about anything beneath its plot. Watching them contradict each other surfaces the tradeoff you would otherwise only discover on page eighty of a draft.

It is not a substitute for a human collaborator who knows your specific voice, and it cannot get your script in front of an agent or producer. The personas are reconstructed from public interviews, commentary, and produced screenplays, not the filmmakers themselves, so they will sometimes be wrong or disagree with each other, and you still have to be the one who decides which note survives. Coverage services like WeScreenplay or Coverfly, or a script consultant like Script Angel, are built for evaluating pages you have already written, not for arguing with a premise you haven't drafted yet. That gap, feedback before the outline exists, is where a structured debate format earns its keep.

Know when to stop testing and start writing

Pressure testing has a failure mode of its own: infinite deliberation. If your premise survives the reversal question and you can name a real engine, further interrogation mostly produces diminishing returns and encourages you to plan instead of write. The point of testing early is to catch fatal flaws cheaply, not to arrive at a guaranteed idea. No amount of conversation, human or AI, removes the risk that the idea only reveals its real problems once you're forty pages into a draft and the characters start behaving on their own terms.

Originally published at https://www.moviementors.ai/answers/pressure-test-half-formed-idea.html.

Written by the team behind Movie Mentors.

Stanley Kubrick's Screenwriting Philosophy, Explained

Ido Barkan — Thu, 13 Aug 2026 12:00:20 +0000

Kubrick is one of the few major directors who left behind almost no formal essay on screenwriting craft, yet his production history and interviews reveal a consistent set of habits. He did not think of a screenplay as a finished literary object. He thought of it as the current draft of an idea that would keep changing until the negative was cut.

The script was a draft, never a blueprint

On most Kubrick productions, pages kept changing during shooting. The Shining is the most documented example: Diane Johnson and Kubrick wrote a script, but dialogue and even scene intentions shifted on set, partly through improvisation with Jack Nicholson and Shelley Duvall. Full Metal Jacket went through extensive rewriting with Michael Herr during and after principal photography. Kubrick's public comments consistently frame the written script as a starting position, not a contract he felt bound to honor once cameras rolled.

This matters for how you read his films as craft models. The clean, symmetrical dialogue you see on screen is often the result of iteration during production, not a single elegant first draft. Writers who study only the shooting script and assume it arrived that way are missing half the process.

He worked with novelists, not career screenwriters

Kubrick's key collaborators, Jim Thompson on The Killing, Arthur C. Clarke on 2001, Diane Johnson on The Shining, Michael Herr on Full Metal Jacket, were novelists or journalists first. This was not incidental. Kubrick has described preferring writers who thought in prose and character psychology over writers trained in industry beat sheets. He wanted people who could generate dense material he could then cut down, rather than people who arrived with a structure already locked.

What this meant in practice

Sessions with these collaborators often produced far more material than made it to screen: alternate scenes, extended backstory, dialogue that was later stripped to almost nothing. Kubrick's method leaned on over generation followed by aggressive subtraction, which is closer to how a novelist revises than how a spec screenwriter is often taught to outline first and write lean from the start.

Structure followed theme, not a formula

Kubrick's films do not map cleanly onto three act templates. Full Metal Jacket is split into two distinct halves with different settings, casts, and tones. 2001 has almost no traditional dialogue driven midpoint. Barry Lyndon is structured as two long parts separated by a literal narrator's intertitle. This was not carelessness. Kubrick talked about wanting structure to emerge from the material and the idea he was exploring, rather than forcing material into a pre set shape.

This is a real point of disagreement among screenwriting teachers. Structure focused approaches, the kind associated with Syd Field or Blake Snyder, treat beat placement as close to a rule. Kubrick's practice suggests structure can be discovered through drafting and even through editing, and that a film can hold together on tone and idea rather than on midpoint and act break mechanics.

Ambiguity was a deliberate choice, not a failure to explain

Kubrick was famously resistant to explaining what his films meant, including 2001's ending and The Shining's supernatural logic. In interviews he argued that a film should work on an audience emotionally and viscerally before it works intellectually, and that over explaining collapses that effect. For screenwriters, the craft lesson is not simply be vague. It is that Kubrick built ambiguity through withheld information and unresolved visual motifs, while still keeping the underlying mechanics of each scene, who wants what, what changes, functioning normally underneath.

Adaptation as demolition, not fidelity

Nearly every Kubrick film is an adaptation, and he was open about treating source novels as material to strip down rather than texts to honor. He changed Stephen King's ending in The Shining. He compressed Anthony Burgess's more moralizing final chapter out of A Clockwork Orange for the theatrical cut. His stated position was that a novel's interior psychology has to be translated into image and behavior, and that fidelity to plot events matters far less than fidelity to the underlying feeling he wanted the film to produce.

How this compares to more conventional screenwriting advice

Principle	Conventional screenwriting guidance	Kubrick's documented practice
The script	Should be locked before production to control budget and schedule	Treated as unfinished through shooting and editing
Structure	Follow act breaks and midpoint beats as reliable scaffolding	Let structure emerge from theme, sometimes ignoring standard beats entirely
Exposition	Clarify stakes and backstory early for audience orientation	Withhold explanation deliberately, trust image and behavior
Adaptation	Preserve the spirit and often the plot of the source	Extract one core idea, discard the rest, rewrite endings if needed

Neither column is simply right. Locking a script early protects a production budget in ways Kubrick, working with unusual creative control and time, did not have to worry about as much. A newer writer without that leverage still generally needs a script that reads as finished to get financed at all.

Where this leaves a working screenwriter

Kubrick's method is genuinely hard to imitate directly because it depended on control most writers never get: final cut, long shooting schedules, and the ability to keep rewriting after a studio would normally expect delivery. What is transferable is the underlying discipline: generate more material than you need, cut hard, resist explaining what a scene can show, and treat structure as something the story earns rather than something imposed on page ten regardless of content.

If you want to test how Kubrick's logic holds up against a more structure first mentor, a tool like Movie Mentors lets you put a Kubrick style persona in the same conversation as someone built on stricter structural principles and watch where their notes actually conflict on your pages. That kind of direct disagreement is useful for spotting where a script is relying on ambiguity as craft versus using it to cover a scene that simply is not working yet, though no chat based tool can replace watching the scene performed to find out for certain.

Originally published at https://www.moviementors.ai/answers/kubricks-screenwriting-philosophy.html.

Written by the team behind Movie Mentors.