Using Torch For Research And Osint: A Workflow

Start with a question, not a search box

Search engines reward exploration, but research needs a method. Before opening Torch, write down the question you are trying to answer, the population you are trying to describe, and the decision the data will support. That framing determines which queries are relevant and which results should be kept.

Build a query set

A single query is rarely enough. Build a small, documented set of queries that covers synonyms, spellings, and related terms. Include narrow queries for specific names and broad queries for themes. Run the same set on more than one engine where possible, and keep the query text exactly as submitted.

Capture results consistently

For every result you keep, record the engine, the query, the date, the position, the title, the onion address, and the snippet. A spreadsheet or a structured database is better than screenshots alone, because it allows deduplication and later analysis. If you use screenshots, store the associated query and date alongside them.

Deduplicate across sources

The same service may appear under several addresses, mirrors, or spellings. Normalise addresses before comparing them, and keep a note of aliases. A result that appears on three engines is not three independent findings; it may simply be the same indexed page seen three times.

Account for index age

Torch is an older engine, so some of its results describe services that existed years ago. That is useful for historical questions and misleading for current-state questions. Decide in advance whether your database records historical presence, current availability, or both, and label each entry accordingly.

Verify before recording

Verification is separate from discovery. Check whether the service is reachable, whether the content matches the snippet, and whether the address is stable over time. Record the verification method and result. If you cannot verify a service, record it as unverified rather than presenting it as a confirmed entry.

Manage provenance and retention

Provenance is what makes a database useful to other researchers. Store enough context that someone else could repeat your search and understand your decisions. Decide in advance how long to keep raw results and how to handle sensitive material. Minimise collection of personal data and avoid retaining content that is not necessary for the research question.

Report limitations

Describe the engines you used, the dates, the query set, and the verification rules. State that coverage is partial and that absence of a result is not proof of absence. These limitations are part of the finding, not an apology for it.

Continue with verifying Torch results and onion sources or the Torch link guide.

Last updated: October 2026. Site map.