Three-Act Structure
Your second act is ninety pages long and nothing in it costs anybody anything. That is a structure problem, and three acts is how you find it. The model is a description of stories that work, not a rule someone is enforcing.
What the three acts actually do
Aristotle's Poetics notes that a whole has a beginning, a middle and an end. That is an observation, not an instruction. The version screenwriters argue about is much younger: Syd Field's Screenplay (1979) laid a 120-page script over three parts of roughly a quarter, a half and a quarter, with the first act ending near page 30 and the last beginning near page 90. A page of standard screenplay format runs about a minute of screen time, so those page numbers are also clock readings.
Strip the page counts away and what is left is three jobs:
- Act one asks. It shows what normal looks like, who has to live in it, and then makes a question unavoidable. Not "will they be happy" but something specific enough to be answered on the last page.
- Act two costs. The protagonist pursues the answer, the opposition responds, and the situation gets worse in ways the protagonist caused. This is where a story is made or lost.
- Act three pays. The question gets answered, at a price, using only what the story already put on the table.
None of that tells you what happens. It tells you what the pages are for, which is the only thing a structural model is good at.
Act breaks are doors, not milestones
An act break is not a percentage. It is a moment after which the previous situation is no longer available. The test is blunt and it works: could your protagonist go home and resume their old life? If the answer is yes, the door has not shut, whatever page you are on.
- The first break closes the exit. The offer is accepted, the body is found, the letter is opened. Something is now true that cannot be untrue, and the protagonist is committed, ideally by a choice rather than by weather. A break where things merely happen to someone reads as passive, and readers feel that long before they can name it.
- The second break is usually the point where the original strategy runs out. Everything the protagonist knew how to do has been tried and failed, and the last piece of information, or the last thing they refused to give up, arrives. Sometimes that is a low point. It does not have to be, and forcing an "all is lost" beat into a story that does not want one is a common way to make a third act feel borrowed.
The midpoint
The midpoint is the least discussed and most load-bearing beat in the model. Its job is structural: it stops the long middle from being one continuous corridor by splitting it into two halves that are not the same. It usually does that in one of four ways.
- A reversal. The situation flips. The hunter is being hunted.
- A revelation. Something the protagonist believed turns out to be false, and the first half has to be reread in that light.
- A false victory or false defeat. They get the thing and it is not the thing. Or they lose it, and losing it frees them.
- A change of stance. The protagonist stops reacting and starts acting, which is why the second half of a good middle feels faster at the same scene length.
Ask one question at the halfway mark: does anything here change what my protagonist is trying to do, or what they think is true? If nothing does, you have found the real reason act two drags, and it is not the length.
Why act two sags
Act two is half the story and has the least shape of its own, so it fails more often than the other two combined. The causes are boringly consistent:
- Nothing is forcing anything. No clock, no antagonist applying pressure, no consequence for waiting. The writer invents reasons for events, and invented reasons read as invented.
- Scenes sit in a sequence instead of a chain. The useful test comes from South Park's writers: between any two consecutive beats you should be able to say therefore or but. If the honest connector is and then, you have a list rather than a plot.
- Repetition at one temperature. The same confrontation three times without escalation. Each individual scene works; the sequence does not.
- The protagonist has no local want. They have a goal for the book but nothing they need by Thursday, so the scenes have no shape to take.
The fixes are the inverse: give the middle a deadline, remove options rather than adding obstacles, and escalate cost instead of volume.
Where the model lies
Three acts is a good diagnostic and a bad generator, and most of the damage done in its name comes from using it the second way. Three honest limits:
- It describes proportion, not content. The fifteen-beat commercial sheets are one interpretation of the shape, not the structure itself, and a story that fails to hit beat nine on schedule is not broken.
- Plenty of forms use a different count. American network television is written in four to six acts because the breaks were sold to advertisers, and that architecture outlived the ad breaks on streaming. British drama frequently runs without act divisions at all.
- Writing to hit a page number is how you get a hollow act break. The door has to be earned by what came before it. Placing it correctly and not earning it is worse than placing it late.
Using it on a draft you already have
Structure is easiest to see with the prose out of the way. Reverse-outline the draft, one line per scene, then read the list alone and answer five questions:
- Where does the protagonist lose the option to walk away, and how far into the book is that?
- What changes at the halfway mark?
- Can you say therefore or but between each pair of consecutive scenes?
- Does the ending use only ingredients the story already planted? Discovering in act three that you needed a scene in act one is the most expensive find in the job.
- What are the actual proportions? Measure them rather than estimating. Slow starts are almost always arithmetic.
This is the reverse-outline pass from how to outline a novel, pointed at shape instead of content. And a moved scene is a moved fact: everyone who knew about it now knows at a different time, which is where structure hands off to continuity.
Structure in Skarvia
Skarvia's Outline board is where this work happens. Beat cards carry a title, a synopsis and a status that cycles idea, draft, done, and you drag them into order, so marking the two act breaks and the midpoint as their own beats makes the shape of the book visible at a glance. The same outline also exists as a treatment, a long-form prose view where a Section heading marks a beat, for when a bulleted list is hiding a missing connection. Your scene list sits beside you in the editor while you draft, and Insights in the Review tab reports runtimes, locations and who talks, so act proportions are measured rather than estimated.
And the bright line holds here as everywhere: Skarvia never writes a word for you. It will not generate your beats, name your midpoint, or tell you what happens on page 60. Structure is the most personal engineering in a book, and a tool that guesses at it hands you somebody else's story. Skarvia is the staff. It keeps the board, the bible and the facts straight while you decide what the story is.