Logos Deep Search

Describe an argument in plain English and get the corpus cards that answer it, ranked.

Deep Search is a beta mode inside Logos for when you have an argument, not a keyword. You describe what you need in plain English, and instead of matching your words literally, it expands your request into many different searches across the same million-card library, then ranks the cards that actually answer you. Each result comes back with a one-line "why it matches" so you can see the reasoning. It draws from your Evidence Finder search allowance, so any signed-in account can use it, and heavier use is a Pro feature.

When to use it

Deep Search is the third mode of Logos, alongside the two keyword modes:

  • Cards and Header / Block (regular Logos) match the exact words you type against card taglines. Best when you already know the phrasing debaters use, like "states counterplan" or "AT: Cap K."
  • Deep Search is for when you know the argument but not the exact tagline. You write a full claim ("I need a card saying deterrence prevents war") and it finds the cards that make that argument, even when they are worded completely differently.

Use Deep Search when a plain keyword search comes back thin because the argument can be phrased a dozen ways, or when you want the single best cards for a claim rather than a raw list. Use regular Logos when you know the exact term and just want everything filed under it. Use Evidence Finder when the library does not have the argument at all and you need a brand-new card cut from the web.

How to describe what you need

Deep Search reads your own words to decide which direction the evidence should cut, so phrase the ask the way you would say it out loud. Three phrasings work especially well:

  • "I need a card saying " for support: "I need a card saying AI will lead to extinction." You get cards that assert the claim.
  • "I need evidence that " does the same thing, worded differently.
  • "Answers to " (or "responses to," "how do I beat") for the other side: "Answers to the states counterplan." You get answer cards ("AT:" blocks) that refute or turn the argument, not cards that run it.

Write a full, specific claim, not a bare topic. "AI regulation" is too vague to aim at; "I need a card saying AI regulation kills innovation" tells it exactly what to find. If your request is ambiguous, it returns cards from both directions so you can see support and answers together. There is no Aff/Neg toggle; the direction comes from how you phrase the ask.

  1. Open Logos from the Tools menu in the top nav (desktop) or the Logos tab in the bottom bar (mobile).
  2. Switch the mode toggle to Deep Search (it carries a Beta pill).
  3. Type your argument into the box, or tap one of the example chips, and press Enter. A short line of narration shows the work as it happens: reading your argument, searching several angles of the corpus, then judging the strongest matches.
  4. Each search is one turn. Deep Search never keeps a back-and-forth thread; every submission is a fresh search that replaces what was on screen.

Reading the results

  • The left pane lists the Ranked matches, best first. Each row shows the card's tag, citation, school, side, and its event (PF, LD, or Policy) so you can tell at a glance which activity it came from.
  • Under each match is a boxed Why it matches line: one sentence on how that card answers your argument. This is the reasoning for the ranking, not the card's own text.
  • Click a match to read the Full card on the right (full-screen on mobile). From the reader you can recolor the highlight and Copy the formatted card, exactly like regular Logos.

Starting over

  • To run a different argument, use Ask for another card at the top of the results, or just type a new claim in the box. Either way it starts a clean search; there is no conversation to continue.
  • If you switch to another Logos mode or another tab and come back, your last Deep Search results are still there. You only return to the empty entry screen after you choose Ask for another card.

Limits and cost

Deep Search shares the Evidence Finder search allowance rather than having its own number: 5 searches per month on the free plan, 200 per month on Pro. One Deep Search turn counts as one search, no matter how many angles it fans out into internally. Because it draws from the same pool, a Deep Search and an Evidence Finder search both come out of the same monthly total. When you hit the monthly limit, the upgrade prompt appears, the same as in Evidence Finder. See Plans & Limits for how the monthly reset works.

Deep Search is in beta, so the exact behavior and limits may change as it improves.

Common problems

  • No strong matches. Deep Search only returns cards that are actually in the Logos library. If nothing in the corpus answers your argument, it says so rather than inventing a card; try rephrasing the claim, or use Evidence Finder to cut a new one from the web.
  • The results lean the wrong way. The direction is inferred from your wording. If you wanted answer cards but got support (or the reverse), rephrase: lead with "Answers to..." for the opposing side, or "I need a card saying..." for support.
  • You hit the monthly limit fast. Deep Search and Evidence Finder share one allowance, so heavy use of either draws down the same total. Upgrading to Pro raises it from 5 to 200 per month.
  • Logos: the keyword search modes Deep Search sits beside.
  • Evidence Finder: fresh web search, and the shared search allowance.
  • Plans & Limits: how the monthly search limit works.