What the registry contains

176 canonical entities across 11 collections. Scenarios compose them, archetypes are presentation-only overlays, and the formal substrate is always recomputed.

Situations, formats, and tools: usage guide.

Mechanisms by agency zone

• Candidate action6 · 21%
• Intermediary dependent8 · 29%
• Exogenous (no agency)14 · 50%

By nature and agency zone

Every mechanism, cross-tabbed by what kind of force it is and who could remove it. Unlike the /archetypes grid, every cell here is counted from the registry — an empty one is a claim (no mechanism of that nature is that removable), not a placement nobody got to yet.

Counts by funnel stage

Traces the atlas cannot place

Where a mechanism operates and where its trace is read are different questions, and only the first is recorded for 0 of 68 emissions. 0 of those conflict outright: the observation is never seen at any stage the mechanism acts on, so no intersection could stand in for the missing answer.

Causes no observation can pin down

A cause is settled only when nothing else emits everything it does. For 10 of 28 that never holds: another cause leaves every trace this one leaves, so it survives whatever is observed. Listed with what covers it.

• mech.automated_keyword_qualification_filterAutomated Keyword / Qualification Filternever ruled outmech.location_or_timezone_compliance_constraint
• mech.bench_priority_fillBench-Priority Fillnever ruled outmech.pre_selected_internal_candidate
• mech.bid_conditional_talent_poolBid-Conditional Talent Poolnever ruled outmech.speculative_sourcing_talent_pooling_without_opening
• mech.employment_gap_downranking_biasEmployment Gap Downranking Biasnever ruled outmech.automated_application_expiration_timeout
• mech.experience_age_grading_mismatchExperience-Age Grading Mismatchnever ruled outmech.unstated_compensation_band_discrepancy
• mech.genuine_technical_skill_shortfallGenuine Technical Skill Shortfallnever ruled outmech.domain_specificity_over_weightingmech.portfolio_work_artifact_misinterpretation
• mech.pre_selected_internal_candidatePre-Selected Internal Candidatenever ruled outmech.bench_priority_fill
• mech.recruiter_volume_quota_incentive_distortionRecruiter Volume & Quota Incentive Distortionnever ruled outmech.bid_conditional_talent_poolmech.speculative_sourcing_talent_pooling_without_opening
• mech.speculative_sourcing_talent_pooling_without_openingSpeculative Sourcing / Talent Pooling Without Openingnever ruled outmech.bid_conditional_talent_pool
• mech.stronger_competing_candidate_in_final_cohortStronger Competing Candidate in Final Cohortnever ruled outmech.bench_priority_fillmech.pre_selected_internal_candidate

8 of 21 observations are consistent with exactly one cause. The rest narrow to a set, never to one.

• obs.communication_mismatch_or_tone_friction_in_panel
• obs.compensation_band_reduced_or_altered_mid_process
• obs.feedback_naming_as_absent_something_the_submitted_work_contains
• obs.multiple_interview_reschedulings_or_interviewer_no_show
• obs.rejection_after_the_application_sat_pending_for_months
• obs.rejection_naming_a_jurisdiction_or_work_eligibility_ground
• obs.rejection_naming_a_specific_industry_sector_as_required
• obs.take_home_assignment_exceeding_reasonable_stated_scope

Causes the registry cannot tell apart

Each group below leaves the same trace. No observation in the registry separates its members, so narrowing to one of them has in fact narrowed to all. 2 groups, 4 of 28 mechanisms. This is a limit of the observation vocabulary, not a defect in the causes.

• • mech.bench_priority_fillBench-Priority Fill
  • mech.pre_selected_internal_candidatePre-Selected Internal Candidate

  shared traceobs.generic_closer_alignment_rejection_templateobs.position_closed_after_final_interview_without_hireobs.rejection_naming_an_internal_hire_as_the_outcome

• • mech.bid_conditional_talent_poolBid-Conditional Talent Pool
  • mech.speculative_sourcing_talent_pooling_without_openingSpeculative Sourcing / Talent Pooling Without Opening

  shared traceobs.complete_silence_after_submissionobs.republished_job_posting_with_refreshed_date_and_identical_requirement_bodyobs.unsolicited_recruiter_outreach_followed_by_ghosting

What this atlas does not cover

A registry is worth what its gaps are worth knowing, so they are counted here rather than described. Every number below is derived from the registry on each build.

• Gates with no proposed change0

  A barrier the atlas documents but for which it names nothing anyone could do. 0 of 16.

  None.

• Causes with no proposed change0

  A mechanism no intervention targets. 0 of 28.

  None.

• Entries citing no published source4

  4 of 96. An entry without a citation still carries its evidence level, which says how far it is supported.

  • obs.feedback_naming_as_absent_something_the_submitted_work_contains
  • obs.rejection_naming_a_specific_industry_sector_as_required
  • mech.bench_priority_fill
  • mech.probation_used_as_extended_de_facto_interview

• Stages a candidate can observe nothing at

  Not a gap to fill: before a role is posted, and inside the hiring manager’s own review, there is by construction nothing for a candidate to witness. It is recorded because the silence is the finding.

  • Client approval

Two limits the counts do not show. The registry describes technical hiring, and the era evidence is drawn from US macro series, so the figures on the eras page describe one market. Neither is a claim about hiring everywhere.

Substrate Topology & Algebraic Pattern Emptiness (SPEC §5)

Every contradiction pattern is formally evaluated over the substrate graph. A pattern is classified as computed_empty if its joint condition set is provably unsatisfiable under monotonic evaluation, and prose_asserted if its satisfying set depends on unobserved hidden state.

Workflow Latency Bounds & Dwell Turnaround Matrix

Turnaround bounds defined across all canonical workflows (4 state machines). Dwell exceeding the maximum bound signals a stalled state driven by hidden queue eviction, ghost postings, or batching.

Financial Flow Conservation & Solvency Verification

Parametric resource flows across all 108 financial records and 9 flows.