Get early access
Watch a run

One run, step by step, and one step you operate

A Grade 11 class asked what is documented about three aviation incidents that circulate online as UFO cases. The answer sits at the top of this page, then the five steps that made it, then the same machinery on a second subject. Step three is live, so the approval there is one you make yourself, on a list you can change.

Where this run ends

The record thinned as LoQuery checked it

Three aviation incidents went in. The table came back with one row empty and no row clean, and every row carries a verdict, one word for how well the located documents support it, with a score from 0 to 1.

The empty row is the result the class learned the most from. The five steps below are how the table got made, and the same machinery then runs a second subject.

  1. Describe a category, so LoQuery builds the list
  2. Discovery runs in waves, then stops itself
  3. Approve the list before anything runs. This one is live
  4. LoQuery flags an item, then researches it anyway
  5. The table from the top, now earned
Research Workbench Table View
Filter Confirmed clean 0 With caveats 2 Candidates 0 Filtered 1
Item name Score Verdict Why Caveats DateOfficial explanationPrimary sourceStill unexplained
Kinross incident 2 attempts 0.71 confirmed with caveats Score 0.71; a USAF determination is on record and is contested, and no government or military file establishing it was located: the safety board's analysis, findings and recommendations were withheld under FOIA. source_contested uncalibrated_authority_domain 23 November 1953 aviation-safety.net/wikibase/161691 USAF: the F-89C was tracking an off-course RCAF C-47 and was lost over Lake Superior en.wikipedia.org/wiki/Felix_Moncla None located. The safety board's findings were withheld under FOIA; ASN WikiBase entry 161691 is user-submitted, not the USAF report aviation-safety.net/wikibase/161691 No. A determination was issued, and the pilot of the RCAF flight named in it, Gerald Fosberg, denied being intercepted en.wikipedia.org/wiki/Felix_Moncla
Shag Harbour incident 1 attempt 0.82 confirmed with caveats Score 0.82; the located source for the records field is a curated research list rather than the investigating agency's own file, so its only record-group holdings are RG77, National Research Council records, rather than the RCMP or DND file the question asked about. uncalibrated_authority_domain 4 October 1967 en.wikipedia.org/wiki/Shag_Harbour_UFO_incident No determination reached; investigated by the RCMP and Canadian Forces en.wikipedia.org/wiki/Shag_Harbour_UFO_incident A Library and Archives Canada research list; its only record-group holdings are RG77, National Research Council bac-lac.gc.ca/…/list/43130 Yes. No conclusion was reached en.wikipedia.org/wiki/Shag_Harbour_UFO_incident
Lake Michigan 1994 incident 3 attempts 0.18 mismatch Low score (0.18); zero of 2 evidentiary field(s) filled. The 1994 reports are a cluster of sightings across several towns, so no single incident record resolves. Not Found Not Found Not Found Not Found
Where the run ends: one row empty, no row clean. The steps below are how it got here.

Cell values, sources and caveats here are real and open to checking. The scores, attempt counts and source totals show the shape of a run and get replaced once a captured session record lands.

Step one

Describe a category, so LoQuery builds the list

One line and four column names are the whole input, and LoQuery works out the search terms, the sources and the domain for itself.

The plan LoQuery wrote names six preferred sources, none of them typed by anyone, so you can start on a subject you have no list for. The full plan sits under the form.

Three words used throughout. Discovery is the stage that finds the items when you have described a category instead of listing one. Six stages run a job and they carry working names: Task Doer fetches and extracts, and Evaluator scores what came back against the document it came from.

How LoQuery writes its own plan →

Discovery
Research Configuration Ready
  1. Direct Research
  2. Discovery
  3. Analytical Soon
Aerial incidents over Canada or the Great Lakes that a government or military body formally investigated
date, official explanation on record, primary source document, still listed as unexplained
Auto-detect (default)

Domains you specify are negated in every search, vetoed in ranking and scoring, and dropped from the fetch queue. If a starved run falls back to one anyway, the result carries a caveat saying so.

The plan as written · four sections, none of them typed

Aviation history · Discovery Plan 4 sections

  1. 1 · Research subject event
  2. 2 · Entity type & domain event · domain: aviation history
  3. 3 · Discovery progress research · item_1/3
  4. 4 · Preferred sources aviation-safety.net, gov, canada.ca, archives.gov, nasa.gov, mil
What the class typed, and what LoQuery wrote from it.
Step two

Discovery runs in waves, then stops itself

The class did not type the three incidents, because working out which ones exist is the first half of the job and the harder half of it, and without a tool it happens by hand. LoQuery runs discovery in waves instead, and logs the reason it stopped, so you can see every query it tried and every candidate it threw away. The rejected column is the one worth reading.

Workflow Discovery
Discovery 5 candidates kept · 4 rejected · 3 waves
  1. Wave 1 Broad authoritative

    Ten queries, most anchored with a site: operator to national archives, defence and transport departments, accident databases.

    Queries
    10
    URLs
    34
    Kept
    3
    Rejected
    2

    Both rejections were page furniture. An article heading and a place name.

  2. Wave 2 Seed-based

    Queries built from what wave 1 found, plus everything already tried, so the wave is pushed toward what it has not seen.

    Queries
    8
    URLs
    29
    Kept
    2
    Rejected
    1

    Two more surfaced, both government programmes rather than incidents. The rejection was a duplicate under a different spelling.

  3. Wave 3 Deep dive

    No site: operators. Domain-specific phrasing aimed at whatever the first two waves did not reach.

    Queries
    6
    URLs
    21
    Kept
    0
    Rejected
    1

    Nothing new. A wave that returns nothing, after one that barely moved, is the signal to stop.

Discovery stopped rather than opening a fourth wave: the target was met and the last wave returned nothing. It logs that decision as budget_satisfied_await_confirmation.

Sources here are real; scores and attempt counts are illustrative until a session capture lands.

Step three

Approve the list before anything runs

This one is live. Strike a candidate below and the count on the button follows. Discovery narrowed 84 URLs down to five names, and two of those five turned out to be government programmes rather than the incidents the question asked for. LoQuery scored them lowest and kept them, so the call is yours.

Discovery Approval
Approve discovered items 5 candidates after 3 waves · 4 rejected by validation

You asked forAerial incidents over Canada or the Great Lakes that a government or military body formally investigated

Strike anything that does not belong. Nothing is researched until you proceed, so this is the cheapest place in the whole run to be strict.

Research of 5 Keep looking Stop here
Rejected before you saw them
  • Top 10 Canadian UFO cases A page heading, not a thing you could research.
  • Wilbert Smith A person. The query asked for incidents.
  • Manitoba A location. Rejected at the metadata pre-filter.
  • Library and Archives Canada An institution, and one of the sources being read.
Strike the two programmes and the button reads three.

Sources here are real; scores and attempt counts are illustrative until a session capture lands.

Step four

LoQuery flags an item, then researches it anyway

Entity screening read the 1994 item before any research started. LoQuery ran it because you approved it, and the warning travels with the row into the results table below. You met the empty 1994 row at the top of this page; this warning is why it arrived already doubted.

Workflow Item warnings

Item warnings · 1

Lake Michigan 1994 incident: Entity screening flagged 'Lake Michigan 1994 incident' (Resolves to a cluster of reports across several Michigan towns rather than one documented incident with a single record.) Researching it anyway because you approved it. Treat its results with this in mind.

Step five

The table from the top, now earned

This is the table the page opened on, reached the way the class reached it. A student's own browser fetched twenty-four pages across eight domains, so no server of ours went near a single one of them.

Every filled cell carries the document behind it, and the row LoQuery could not fill says in its own words why it stayed empty. The Kinross row is the other one worth reading. Its primary source cell reads “None located”, because ASN WikiBase entry 161691 is user-submitted rather than the USAF report. A structured aviation database with serial numbers in it is still a wiki. So you can mark a gap rather than grade a guess.

Research Workbench Table View
Filter Confirmed clean 0 Candidates 0
Item name Score Verdict Why Caveats DateOfficial explanationPrimary sourceStill unexplained
Kinross incident 2 attempts 0.71 confirmed with caveats Score 0.71; a USAF determination is on record and is contested, and no government or military file establishing it was located: the safety board's analysis, findings and recommendations were withheld under FOIA. source_contested uncalibrated_authority_domain 23 November 1953 aviation-safety.net/wikibase/161691 USAF: the F-89C was tracking an off-course RCAF C-47 and was lost over Lake Superior en.wikipedia.org/wiki/Felix_Moncla None located. The safety board's findings were withheld under FOIA; ASN WikiBase entry 161691 is user-submitted, not the USAF report aviation-safety.net/wikibase/161691 No. A determination was issued, and the pilot of the RCAF flight named in it, Gerald Fosberg, denied being intercepted en.wikipedia.org/wiki/Felix_Moncla
Shag Harbour incident 1 attempt 0.82 confirmed with caveats Score 0.82; the located source for the records field is a curated research list rather than the investigating agency's own file, so its only record-group holdings are RG77, National Research Council records, rather than the RCMP or DND file the question asked about. uncalibrated_authority_domain 4 October 1967 en.wikipedia.org/wiki/Shag_Harbour_UFO_incident No determination reached; investigated by the RCMP and Canadian Forces en.wikipedia.org/wiki/Shag_Harbour_UFO_incident A Library and Archives Canada research list; its only record-group holdings are RG77, National Research Council bac-lac.gc.ca/…/list/43130 Yes. No conclusion was reached en.wikipedia.org/wiki/Shag_Harbour_UFO_incident
Lake Michigan 1994 incident 3 attempts 0.18 mismatch Low score (0.18); zero of 2 evidentiary field(s) filled. The 1994 reports are a cluster of sightings across several towns, so no single incident record resolves. Not Found Not Found Not Found Not Found
Open three sources and show where the machine got it wrong. The row LoQuery refused to fill is the most useful one in class, and the Kinross primary source cell is the second.

Sources here are real; scores and attempt counts are illustrative until a session capture lands.

After the run · 1 of 2

Same machinery, a different subject

An analyst listed three commodities traders and asked for four things about each one: the regulator, the penalty, the date of the order and the primary filing. Nobody wrote a line of code for enforcement. LoQuery worked the subject type, the domain and five preferred sources out of the question alone, so you can point it at a field nobody coded for.

Two runs, one shape of question. A list of things, and the same few facts about each of them. The shape fits a reading list, a sector's suppliers, every clinic in a region, anything you would otherwise chase across thirty tabs and paste into a spreadsheet.

Direct Research
Research Configuration Ready
  1. Direct Research
  2. Discovery
  3. Analytical Soon
Vitol Inc, Glencore Ltd, Freepoint Commodities
regulator, penalty amount, date of order, primary filing
Auto-detect (default)

Domains you specify are negated in every search, vetoed in ranking and scoring, and dropped from the fetch queue. If a starved run falls back to one anyway, the result carries a caveat saying so.

The plan as written · five preferred sources, none of them typed

Enforcement · Discovery Plan 4 sections

  1. 1 · Research subject organisation
  2. 2 · Entity type & domain company · domain: enforcement
  3. 3 · Discovery progress research · item_1/3
  4. 4 · Preferred sources sec.gov, justice.gov, cftc.gov, gov, courtlistener.com
The form the analyst typed, and the plan LoQuery wrote from it.
After the run · 2 of 2

Every figure carries the filing it came from

Three rows, twelve cells, a link on each. Defending one figure never means re-running the job. Not one came back clean. The Vitol penalty combines a Justice Department and a Brazil resolution, and a parallel CFTC order issued the same day offsets part of it.

Adding the two headline figures gives a total nobody owed, which is exactly the mistake a bibliography at the end of a report lets a reader make. LoQuery read nineteen pages across five domains here. Three of the five domains are the federal agencies themselves.

Research Workbench Table View
Filter Confirmed clean 0 Candidates 0 Filtered 0
Item name Score Verdict Why Caveats RegulatorPenalty amountDate of orderPrimary filing
Vitol Inc 2 attempts 0.79 confirmed with caveats Score 0.79; the USD 135m is a combined DOJ and Brazil resolution, and a parallel CFTC order the same day partly offsets against it. The figures are neither one number nor two that add up. parallel_action_unreported US Department of Justice justice.gov/…/vitol-inc-agrees-pay-over-135-million USD 135,000,000, a combined DOJ and Brazil resolution justice.gov/…/vitol-inc-agrees-pay-over-135-million 3 December 2020 justice.gov/…/vitol-inc-agrees-pay-over-135-million Deferred prosecution agreement (FCPA) justice.gov/…/vitol-inc-agrees-pay-over-135-million
Glencore Ltd 3 attempts 0.78 confirmed with caveats Score 0.78; the order settles with three entities collectively defined as Glencore, so the figure is not attributable to the named item alone. entity_scope_broader_than_item CFTC cftc.gov/PressRoom/PressReleases/8534-22 USD 1,186,345,850, penalty plus disgorgement, three entities collectively cftc.gov/PressRoom/PressReleases/8534-22 24 May 2022 cftc.gov/PressRoom/PressReleases/8534-22 CFTC order cftc.gov/PressRoom/PressReleases/8534-22
Freepoint Commodities 3 attempts 0.74 confirmed with caveats Score 0.74; a parallel CFTC order charges the same conduct under a different statute and is largely offset against this one, so neither figure alone is the answer and adding them is wrong. parallel_action_unreported US Department of Justice justice.gov/…/commodities-trading-company-98m USD 98,551,150, a DOJ penalty and forfeiture justice.gov/…/commodities-trading-company-98m 14 December 2023 justice.gov/…/commodities-trading-company-98m Three-year deferred prosecution agreement, District of Connecticut justice.gov/…/commodities-trading-company-98m

Sources Discovered 19 sources 5 domains

Most-used sources

  • www.justice.gov × 6
  • www.cftc.gov × 6
  • www.sec.gov × 3
  • www.courtlistener.com × 2
  • www.reuters.com × 2
Somebody asks where the 1.186 billion came from, and the answer is a link on that cell and a caveat saying the figure covers three entities.

Sources here are real; scores and attempt counts are illustrative until a session capture lands.

The controls

Steer the run without restarting it

The aviation run got a validation rule. The enforcement run got a manual verdict, because the Glencore figure covers three legal entities and one of them is not called Glencore. Neither restarted the run. The one that was sent did not take effect on send either, and the form says so on the line under the button.

Picking from LoQuery's list is the fast path for someone who already knows what they want. A drawer under the form takes a typed sentence instead. A student typing one, and the reading LoQuery gave back →

Workflow User input injection

User input injection

22 actions, each aimed at a stage you choose

  • Force include URL
  • Exclude domain pattern
  • Raise confidence threshold
  • Lower confidence threshold
  • Force rerun item
  • Set verdict manually
  • Override verdict reasoning
  • Pause pipeline
  • Resume pipeline
  • Boost URL (pre-extraction)
  • Reject URL (pre-extraction)
  • Restore dropped URL
  • Flag item as priority
  • Flag item to skip
  • Agree with verdict
  • Boost source authority
  • Demote source authority
  • Add search query
  • Add sources (bulk CSV)
  • Extend entity definition
  • Add validation rule
  • Add knowledge triple

Target stage

  • Discovery drains at wave boundary
  • Research drains at attempt boundary
  • Evaluation drains at verdict boundary

Validation rule

only count a source as primary if it is a government or military file

Submit injection

Queued for application at next safe checkpoint. Watch for the applied-row update.

User input injection

22 actions, each aimed at a stage you choose

  • Force include URL
  • Exclude domain pattern
  • Raise confidence threshold
  • Lower confidence threshold
  • Force rerun item
  • Set verdict manually
  • Override verdict reasoning
  • Pause pipeline
  • Resume pipeline
  • Boost URL (pre-extraction)
  • Reject URL (pre-extraction)
  • Restore dropped URL
  • Flag item as priority
  • Flag item to skip
  • Agree with verdict
  • Boost source authority
  • Demote source authority
  • Add search query
  • Add sources (bulk CSV)
  • Extend entity definition
  • Add validation rule
  • Add knowledge triple

Target stage

  • Discovery drains at wave boundary
  • Research drains at attempt boundary
  • Evaluation drains at verdict boundary

Item index (zero-based)

0

Asserted verdict

ACCEPT / REVISE / REJECT

Your reasoning

Why are you asserting this?

Submit injection

One form, two runs. The fields under the action list change with the action you pick. On the left, a validation rule typed and sent. On the right, the same form the moment you choose a manual verdict, its three fields still empty.

Run one yourself

Opening to a small first group, so tell us what you would point LoQuery at first.

One email when it opens. No newsletter unless you ask for one.