Most writing advice says: do your research and the details will be accurate. Nobody mentions that the research itself becomes a bigger problem than the writing.
Here's the paradox. A historical novelist spends three weeks researching 1920s Shanghai. Concession-system politics, rickshaw fares, Art Deco facades along the Bund. Thirty browser bookmarks, fifteen PDFs, half a notebook of handwritten notes, screenshots scattered across three apps. Every piece might be useful. Then she sits down to write a scene where a character pays for tea -- and can't find the price data. Was it in Notion? Google Docs? The phone memo app? Twenty minutes of searching later, she locates the note. But the sentence she was building is gone. The rhythm shattered.
This happens again. And again. Not because of laziness -- because research materials live in five different places with no connection to the writing project. When the cost of looking something up gets high enough, writers instinctively skip it, guess, or leave a blank they forget to fill. The research exists. The access doesn't.
Three patterns. Same root cause.
Three Ways Research Materials Spiral Out of Control
Information flooding. Everything might be relevant, so everything gets saved. Two weeks later: a mountain of unsorted, unindexed, tangled material. The detail is in there somewhere. Good luck finding it.
Research-writing fracture. Notes live in one app, the draft in another. Every verification requires a context switch -- find the app, search, locate, switch back. Twenty minutes per hour lost to toggling. That's not just inefficiency. It's the half-formed sentence dying in the gap between windows.
Source tracking collapse. "A rickshaw fare was about two jiao." Where did that come from? Three months later, the editor asks. Blank stare. Research it again from scratch? Wave vaguely at "some article"? Both options are bad.
One common root: research materials don't live with the writing project.
Building a Research System in Slima
The fix is counterintuitively simple -- make research a part of the project, not an attachment to it.
In Writing Studio's File Tree, the "Research" folder sits alongside "Drafts," "Characters," and "World." Not stored in a separate tool and linked over. Not managed in another app. It lives inside the project. Open the writing environment and the research is right there.
The folder structure should mirror the questions that come up while writing, not academic taxonomy:
Research/
├── Historical Background/
│ ├── 1920s-shanghai-overview.md
│ ├── concession-system.md
│ └── currency-and-prices.md
├── Locations/
│ ├── the-bund.md
│ ├── french-concession.md
│ └── city-god-temple.md
├── Daily Life/
│ ├── fashion-and-clothing.md
│ ├── food-culture.md
│ └── transportation.md
├── Character Prototypes/
│ ├── merchant-class.md
│ └── sing-song-girls.md
└── _source-index.md
Writing a scene where the character walks along the Bund and wondering what the street looked like in 1925? Go to "Locations/the-bund.md." Character pays a rickshaw fare? "Daily Life/transportation.md." The path itself is the map to the answer. No guessing, no rummaging.
Each research note needs a consistent internal structure too. Four sections, every time:
Summary -- one or two sentences. Enough to decide in three seconds whether this note is relevant to the current question.
Key Details -- digested highlights, not raw copied text. The original source might be five thousand words. This section distills it to three hundred.
Story Application -- the difference between a research note and a reading note. Which character encounters this detail? In which chapter? How does it get used? Write it down. Pull abstract research back into concrete writing decisions.
Sources -- book title, page number, URL, access date. When the editor asks three months from now, these four lines are the lifeline.
At the top level of the File Tree, create a _source-index.md file (the underscore keeps it sorted first) to centralize every cited source. Verification requires opening one file, not trawling through the entire folder structure.
Making Research and Writing Seamless
Access cost determines behavior.
High cost -- skip, guess, forget to fix later. Low cost -- check in three seconds, confirm, keep writing. The entire research system succeeds or fails on this line.
Split View (Cmd+ or Ctrl+) is the most direct way to lower that cost. The draft on the left says "the protagonist jumped off the rickshaw and tossed a handful of coins." The right side shows "transportation.md" -- confirming whether the fare is plausible. No app switching. No leaving the writing environment. Side by side, verify, move on to the next sentence.
Quick Open (Cmd+P) is faster still. Skip the folder hierarchy entirely -- type "currency" and jump to price-related notes. Type "bund" and land on the location description. Fuzzy search. A few keystrokes suffice.
But the real game-changer is the AI Assistant.
Press Cmd+Shift+A (Mac) or Ctrl+Shift+A (Windows) to open the AI Chat Panel. Ask questions in natural language -- not to the internet, but to the project's own research files.
Try this prompt:
Based on the contents of my "Research" folder, answer this question:
How much did a meal at an upscale restaurant cost in 1920s Shanghai? How many days of wages would this represent for an average worker?
Please cite specific source notes.
The AI Assistant pulls answers from already-vetted research notes, not random web results that might be wrong. Answer reliability depends on the quality of the research -- but at least the sources are traceable.
An even more practical scenario: mid-writing, an uncertain detail comes up. Tag it [RESEARCH: phone rates at the time] and keep the rhythm going. After accumulating a batch, clean them all up at once:
Please scan all instances of "[RESEARCH]" or "[TBD]" markers in my "Drafts" folder.
Compile them into a list showing:
- Which file, which section
- What needs to be researched
- Suggested research notes that might already have answers (if relevant notes exist)
Writing and research stop being two separate phases. Tag, keep writing, batch-process later. Flow and accuracy can coexist.
Avoiding Research Pitfalls
Two traps. They look like opposites on the surface. Both are equally lethal.
The first is research addiction.
It feels productive. Opening sources, taking notes, comparing references -- the brain generates an illusion of "making progress on the project." But the word count hasn't moved. The story hasn't advanced. Umberto Eco researched for two years before starting The Name of the Rose, but he knew exactly when to stop. Most people don't.
A simple test: can the basic details of the scene be written? If yes -- start writing. Uncertain details get a [RESEARCH] tag and life goes on. "Write while researching" beats "research then write" every time, because writing itself reveals what information is actually needed. Saves enormous amounts of blind collecting.
The second is info dumping.
Two weeks spent researching the currency system. The temptation to showcase all that homework is overwhelming. So readers slam into a five-hundred-word lecture on silver dollar minting history in the middle of a tense scene. Rhythm destroyed.
Readers need "He paid two silver dollars." Not "He paid two silver dollars, a coin type first minted in the late Qing Dynasty, weighing approximately 0.72 taels with 90% silver content..." Good research is invisible. Readers feel the world is real and believable without noticing the homework behind it. Iceberg theory -- one-eighth above water, but the seven-eighths below hold everything up.
One more thing easily overlooked: research shows how things actually were. But this is fiction. If a historical fact obstructs the story's emotional trajectory, adjustment is an option. Note the "intentional deviation" in the research file along with the reasoning -- that way it's clearly a creative choice, not an ignorant mistake.
The Value of Offline Research
Slima is offline-first. All research materials are stored locally. No dependency on any cloud service.
The cafe Wi-Fi goes down. The train enters a tunnel. No internet on the plane. Cloud-dependent workflows collapse in these moments. But the project's research materials are still there. Open Writing Studio -- draft on the left, notes on the right -- keep writing.
Security matters too. Research materials won't vanish because some online service changes its pricing or shuts down entirely. Those notes assembled over weeks or months sit on the local machine, protected by Version Control. Every edit is tracked. Any version is recoverable.
Past writers managed research with index cards and paper folders. The system was limited by physical space, search relied on memory, backup meant photocopying.
The tools are different now. Research materials and writing project live in the same place -- open and ready. The AI Assistant searches note contents in natural language, no manual digging required. Version Control tracks every change, no fear of accidental deletion or botched edits.
Put these to use, and research serves the writing instead of becoming an excuse to postpone it.
In the next article, we'll tackle multi-POV narrative management -- when a story has multiple point-of-view characters, how to ensure each perspective stays clear, consistent, and carries a distinct voice.