Games Like Her Story: How to Handle Information Overload Without Losing the Plot
A field workflow for investigators facing clue overload in database-search mysteries.
In games like Her Story, overload is not a flaw in the design. It is the design.
The archive keeps speaking while your model of the case keeps changing. Every new clip can reopen a timeline, reframe a motive, or quietly break a theory you thought was stable.
The risk is not missing one clue. The risk is letting unranked evidence become unearned certainty.
Why These Games Feel So Mentally Expensive
Database-search mysteries force investigators into an endless foraging loop: search, skim, extract, then search again with sharper language. That loop is productive, but it is also cognitively expensive because your question evolves while you work.
Unlike linear stories, evidence often arrives out of order, and retrieval is typically constrained. You cannot rely on “I’ll just watch everything in sequence” because sequence is exactly what the system withholds. That means progress depends less on volume and more on ranking: what changes your next move, what can wait, and what should be archived.
When that ranking discipline collapses, judgement starts to bend. Investigators fall back on the earliest coherent narrative because it feels efficient. In practice, that is how suspect lock-in begins.
Signal Triage Model
Signal Means Action Value, Not Moral Truth
Treat each new artefact as intake, then ask a single operational question: Does this change what I do next?
High-signal evidence should force movement. It opens a new search branch, creates or breaks a relationship, or directly contradicts something you were treating as stable. Medium-signal evidence may matter later, but cannot yet carry a decision. Low-signal evidence is atmospheric, repetitive, or currently unanchored.
The key is not deletion. The key is placement.
- HIGH: pin as one testable claim with provenance.
- MEDIUM: park in a Lead Ledger with a promotion trigger.
- LOW: archive with one label so it stays retrievable without dominating attention.
This model protects the caseboard from emotional weight. "Interesting" and "actionable" are not the same thing.
Triage Card (One Per Artefact)
ID:
One-line extraction (<=140 chars):
Entities: Person | Place | Object | Time
Hooks: 2-5 search terms (+ 1 synonym set)
Signal: HIGH | MEDIUM | LOW
Provenance: Observed | Stated | Inferred
Next action (exactly one): Search | Link | Lead Ledger | Contradiction check
If you cannot write the one-line extraction, park the artefact and continue intake. Unclear notes are how clutter enters the system.
The Three-Pass Review That Keeps the Case Legible
Most investigators do not lose the plot because they are bad at deduction. They lose it because their review rhythm is inconsistent. A simple recurring cadence solves more than another hour of raw intake.
Pass One: Scan
Scan is deliberately fast and timeboxed. The objective is exposure and first-pass categorization, not interpretation theater. Move through new material, write triage cards, capture hooks, then stop when the time window ends.
Pass Two: Structure
Structure converts intake into a caseboard you can actually reason with. Promote only high-signal cards, maintain a timeline spine with explicit anchors, keep a relationship web only for links that change decisions, and quarantine medium-signal leads in a ledger.
This is also where you run a redundancy sweep. Duplicate nodes and decorative links feel productive, but they usually increase cognitive drag without increasing insight.
Pass Three: Contradiction
Contradiction is your anti-lock-in pass. Write the dominant narrative in three sentences, then actively pressure it.
- What would prove this wrong?
- Which “facts” are actually weakly sourced claims?
- Which timeline anchors are inferred rather than explicit?
Generate one rival scenario that explains the same anchors, then run one or two targeted re-checks. Keep this pass sharp and brief. Contradiction should be a scalpel, not a second intake session.
Failure Patterns to Watch Early
| Failure pattern | What it looks like in practice | Correction |
|---|---|---|
| Note hoarding | Endless capture, no movement | Force triage placement and run weekly merge/delete sweep |
| Timeline drift | Recalled order keeps changing | Keep a timeline spine with confidence flags |
| Suspect lock-in | One story becomes untouchable | Schedule contradiction pass and maintain one rival scenario |
| Keyword fixation | Same terms, same dead ends | Add synonym sets each session and retire stale hooks |
| False certainty | Assertions presented as facts | Stamp every claim as Observed, Stated, or Inferred |
Caseboard Hygiene
Before ending any investigation session, verify five things:
- Every new artefact has a triage card.
- Only high-signal claims are pinned.
- Provenance is visible on every critical node.
- The timeline marks uncertainty instead of pretending precision.
- The next search question is explicit.
Five checks are enough. You want consistency, not ritual bloat.
A Quiet Bridge to Palladian
This is the class of problem we are building around in Palladian: keeping investigation reasoning clear under heavy evidence flow.
Our evidence board and AI interrogation loop are designed to hold provenance, contradiction pressure, and uncertainty in the same working surface. Not to eliminate complexity, but to keep complexity inspectable.
If you want to test this workflow in a live case, start in the case archive and then defend your theory through the interrogation loop.
A case is rarely lost in one dramatic error.
It is usually lost in small organizational failures repeated long enough to feel normal.
Rank the signal.
Keep the rhythm.
Defend only what survives contradiction.