| name | sherlock-research |
| description | Investigate problems systematically using Sherlock Holmes' method: observe, deduce, hypothesize, and verify. . Use when working with sherlock research. |
| domain | research |
| author | oyi77 |
| license | Apache-2.0 |
| subdomain | research |
| tags | ["analysis","investigation","research","sherlock"] |
| version | 1.0.0 |
Sherlock Research
When to Use
Trigger phrases:
- "sherlock research"
- "Help me with sherlock research"
Use cases:
- When the task matches this skill's domain expertise
When NOT to use:
- For tasks outside this skill's scope
When NOT to Use
- When the answer is already known and documented
- For time-sensitive decisions that cannot wait for thorough research
- When the topic is outside your domain of competence
Overview
Sherlock Research applies the deductive method of Sherlock Holmes — systematic observation, disciplined inference, and evidence-driven conclusion — to any investigation or research problem. The core premise is that the human mind naturally jumps to conclusions; a structured methodology counteracts this tendency by forcing each step: gather all relevant facts before forming hypotheses, then test each hypothesis against the evidence, discarding those that do not fit.
The methodology rests on three pillars. First, observation without preconception — entering an investigation with an open mind, recording data before interpreting it. Second, hypothesis generation — proposing multiple competing explanations for the observed facts, including the least obvious one. Third, elimination — systematically ruling out hypotheses that contradict evidence until only one plausible explanation remains. This mirrors Holmes's maxim: "When you have eliminated the impossible, whatever remains, however improbable, must be the truth."
In practice, Sherlock Research is not limited to criminal investigation. It applies to competitive intelligence, due diligence, technical debugging, strategic analysis, journalistic research, and any domain where partial information must be assembled into a coherent picture. The key capabilities are disciplined source triage, bias awareness, chain-of-evidence documentation, and confidence-calibrated reporting.
The framework also acknowledges its limits: it works best with accessible evidence, struggles when key data is permanently unavailable, and depends on the investigator's domain knowledge to generate plausible hypotheses. The practitioner must balance thoroughness with timeliness — perfect information is rarely attainable, and the goal is actionable certainty, not absolute certainty.
Workflow
- Establish baseline facts — Before forming any hypothesis, collect and document every known fact relevant to the question. Use primary sources only at this stage. Record dates, actors, locations, sequences, and quantities. Note what is unknown (gaps) alongside what is known.
- Generate competing hypotheses — Propose at least three distinct explanations for the observed facts. Avoid anchoring on the first plausible explanation. For each hypothesis, write down what additional evidence would confirm it and what would refute it.
- Rank sources by reliability — Classify each source on a credibility spectrum: primary documentary evidence > corroborated eyewitness > single eyewitness > hearsay > anonymous claim > circumstantial inference. Weight conclusions accordingly. Flag sources with known bias or conflicts of interest.
- Cross-reference across independent lines — Compare evidence from unrelated sources. A claim supported by two independent, reliable sources is far stronger than the same claim repeated ten times from a single origin. Look for convergent evidence — multiple weak indicators pointing in the same direction can be as strong as a single strong indicator.
- Seek disconfirming evidence — Actively search for data that contradicts your leading hypothesis. This is the most counter-intuitive and most important step. If you cannot find disconfirming evidence after genuine effort, your leading hypothesis gains credibility. If you avoid looking, you are rationalizing, not investigating.
- Eliminate hypotheses — Rule out hypotheses that contradict established evidence. The surviving hypothesis is not necessarily true — it is simply the least wrong. If multiple hypotheses survive, identify what additional evidence would separate them and go collect it.
- Document the chain of reasoning — Produce a written record that traces from raw evidence through inference to conclusion. Include dead ends and discarded hypotheses — they protect against future re-investigation of settled points. Assign confidence levels (low/medium/high) to each conclusion, and be explicit about what would change your assessment.
Source Evaluation
- Authority — Is the source credible and expert?
- Currency — Is the information recent and relevant?
- Objectivity — Is there bias or conflict of interest?
- Accuracy — Can claims be verified independently?
Output Format
- Executive summary (1-2 paragraphs)
- Key findings (bullet points)
- Detailed analysis (sections with evidence)
- Recommendations (actionable next steps)
- Sources and methodology
Common Pitfalls
- Confirmation bias — The most pervasive trap. Once you form a hypothesis, you naturally seek evidence that supports it and discount evidence against it. Mitigation: deliberately spend as much time searching for disconfirming evidence as confirming evidence.
- Anchoring on the first source — The first piece of information you encounter shapes your entire framework. Subsequent evidence is interpreted through this lens. Mitigation: seek out the strongest opposing view first, before you read the supporting argument.
- Overweighting vivid evidence — A single dramatic anecdote outweighs statistics in our judgment. A compelling story carries more weight than aggregate data. Mitigation: ask "what is the base rate?" before acting on any individual case.
- False precision — Treating uncertain numbers as exact. An estimate of $1.5M implies precision that "between $1M and $2M" honestly conveys. Mitigation: report confidence intervals and margins of error. Be explicit about what you do not know.
- Premature closure — Stopping investigation once you have a plausible answer. Most errors come from the first conclusion that fits, not the wrong answer after thorough search. Mitigation: adopt a "three good hypotheses" rule — do not conclude until you have tested at least three distinct explanations.
Anti-Rationalization Table
| Rationalization | Reality |
|---|
| "First result is good enough" | Deep research finds better answers. Keep digging. |
| "I do not need to verify sources" | Unverified sources lead to wrong conclusions. Always cross-check. |
| "Research is a one-time thing" | Markets change. Research needs to be continuous, not one-off. |
| "This evidence confirms what I already suspected" | That is the moment to test the opposite hypothesis hardest. |
| "The data looks clean — no need to verify further" | Errors hide in what you are not looking at. Scrutinize from multiple angles. |
| "I do not have enough information to start" | You rarely will. Start with what you have, document gaps, and fill them iteratively. |
Process
- Frame the problem — Convert the vague question into a specific, testable hypothesis. Write down what evidence would confirm, disconfirm, or leave the question unsettled.
- Map the evidence landscape — Inventory available data sources: public records, databases, expert contacts, physical evidence, digital artifacts. Rank by reliability and relevance.
- Collect systematically — Gather evidence in order of fragility (most volatile first). Document chain of custody. Tag each piece with source, timestamp, and collection method.
- Analyze and cross-reference — Compare claims across independent sources. Identify contradictions. Look for what is missing — absent evidence is often as telling as present evidence.
- Conclude and communicate — Assign confidence levels to each finding. Prepare a narrative that walks from evidence to conclusion, not conclusion to evidence. Archive raw data for future re-examination.
Verification
Monetization
- Private investigation and due diligence services — Offer structured investigation for divorce, fraud, background checks, and asset tracing. High-value, project-based billing ($200–500/hour or flat per case).
- Competitive intelligence consulting — Businesses pay for structured intelligence on competitors: pricing moves, hiring patterns, product roadmaps, partnership signals. Retainer model ($2K–5K/month for weekly briefs).
- Investigative journalism / freelance research — Sell in-depth investigations to publications or produce subscriber-funded research reports. Narrative format, single-topic deep dives. Revenue from publication fees, syndication, and reader donations.
- Training and methodology licensing — Package the investigation framework as a workshop or certification. Sell to corporate security teams, journalism programs, and law enforcement agencies. Product model: curriculum + workbooks + case studies.
- Research-as-a-Service (RaaS) — Subscription-based ongoing research for startups and SMBs: customer discovery, market validation, competitor tracking. Monthly retainer with scheduled delivery of research briefs.
