The Statistic Was Still Known Internally

The visible absurdity is the hook. The structural finding is the second answer. The cognition exists inside the apparatus; the admissibility-to-the-public-record was withdrawn while the trend grew. This essay treats the case as a worked specimen of three corpus diagnostics operating in interaction: The Record Gate (cognition held below the binding threshold), Causal Talisman (the talismanic explanations occupying the discourse where contribution data would have been examined), and Structured Underspecification (the discourse vocabulary within which the case is adjudicated). The corpus discipline produces a specific artifact when applied to a multi-cause discourse — the contribution table — which is what symmetric application of the Causal Talisman diagnostic to the Espoo case yields. The table is in §V.

I. The hook: the premature-baby line

Nurmi’s answer to the question about rising violence: yes, it has risen, but the pupil population has also grown, and the share of pupils needing demanding support has grown more. One factor that may affect this, he said, is that increasingly small premature babies can now be saved. He added that the causes should really be asked of other experts.

The article’s fact-check section quietly dismantles the medical claim’s magnitude. The Finnish Institute for Health and Welfare reports that very preterm babies — born before 32 weeks or under 1,500 grams — constitute roughly 0.8% of births in 2024. The Duodecim medical society notes that premature children’s behavioural profile tends toward conscientiousness rather than aggression. Nurmi himself describes the violence trend as a national and global phenomenon — which makes a Finland-specific neonatal trend a poor candidate for the operative driver without additional contribution accounting.

The reaction across Finnish-language X and journalism focused on this. Commentators called Nurmi’s claim desperate, incompetent, bottom-scraping. The discourse converged on the visible absurdity as the finding.

Treating the visible absurdity as the finding is itself the move that lets the structural finding pass unmentioned.

II. The Record Gate move

Buried in the same interview, the journalist asks about a separate matter: the annual count of violence and threat incidents the city had been publishing. The exchange runs roughly as follows.

The number of violence and threat incidents has been published yearly. Last year it was no longer published, even though the number had clearly been growing. Why?

Nurmi: The assessment report now covers different items. We continue to follow the number closely and develop our activities on that basis, even though they are not recorded in the report.

Why has the reporting been changed? Wouldn’t it be good to keep track of the volume of teacher-reported violence and threat incidents?

Nurmi: We have an enormous amount to report on. We want to continue following how the situation develops.

The annual count was published in previous years. In the year the trend was clearly growing, it was removed from the public assessment report. The cognition exists internally; the admissibility-to-the-public-record was withdrawn.

The Record Gate names this exact move: institutional cognition becomes binding-on-the-institution at staged thresholds (cognition → trace → file → record → contestable → binding → power-shifting). The underdiagnosed lever is where cognition is held below the binding threshold. Espoo rolled the violence statistic back from binding public record to internal monitoring — exactly when the statistic would have most strongly constrained the institution.

Nurmi’s defence operates through the slot-function shape: the slot (the assessment report) still exists, therefore the function (violence tracking) is performed; the cognition still happens inside the apparatus, therefore the apparatus is doing its work. The fact that the cognition is no longer admissible-to-the-public-record is treated as a format change, not a structural move.

The headline finding is here. Espoo did not stop knowing the violence statistic. Espoo stopped making the statistic publicly binding. The talismanic explanations are what made that structural move discursively tolerable.

In the vocabulary of Feedback Authority, the downgrade is precise: the annual statistic, when published in the assessment report, sat at the procedural grade — an institutional duty to publish a recurring number that another actor (council, opposition, press) could review against trend. Removing it leaves the cognition at the advisory grade — internally noted, externally non-binding, no actor obliged to answer the trend. The architecture continues to use procedural-sounding language (“we continue to follow,” “we develop our activities on that basis”) while the mechanism delivers at advisory. This is the canonical conflation pattern feedback-authority names.

III. Causal displacement: three explanations protecting one conclusion

Nurmi’s premature-baby line was followed within twenty-four hours by a LinkedIn post from Laura Liukko, a doctoral researcher in education at the University of Helsinki, rejecting the premature-baby explanation and substituting societal malaise driven by the current government’s welfare cuts. The comment thread under Liukko’s post produced a third position from Lea Tornberg, an educator marketing a pedagogical-training product, asserting that there has never been a resource shortage — the real resource is teacher technique.

Three actor classes, three explanations. Each declines to engage the others’ position. Each protects a different load-bearing conclusion that also happens to be consistent with the third position’s load-bearing conclusion, when read structurally:

• Nurmi’s premature-baby + global-trend explanation lets the apparatus avoid implicating Espoo’s resource and placement decisions
• Liukko’s societal-malaise + austerity explanation lets the educational-progressive coalition avoid implicating the inclusion-policy reform that dissolved special-needs classes
• Tornberg’s technique-not-resources explanation lets her training product remain marketable as the solution

The three explanations look like disagreement. Structurally, none of them lands on the operative-mechanism candidate the teachers themselves name: the inclusion-policy reform. All three explanations route the causal weight elsewhere — to medical-demographic factors, to national-policy austerity, to teacher competence. None of the three engages the magnitude of the inclusion-policy contribution.

This is the displacement pattern. It is not coordination — Liukko explicitly attacks Nurmi; Tornberg disagrees with both. It is convergent displacement: each actor reaches for the talisman their substrate’s reward environment pays, and none of the talismans implicates the operative-mechanism candidate the practitioners name. The Response Vector is the architectural reason: institutional pressure routes through the cheapest available channel, and a talismanic cause-name (formal gaming + incidence shifting in the vector’s vocabulary) is cheaper to produce than a contribution table that would implicate the apparatus producing it.

The displacement makes the Record Gate move tolerable. If the apparatus’s published causal narrative is premature babies-or-malaise-or-technique, the apparatus does not need to publish the data that would let an external observer test those causes against the inclusion-policy candidate. The withdrawal of the violence statistic from the public record is consistent with — and protected by — the multiple talismanic explanations occupying the discourse where the data would have been examined.

IV. The uncomputed mechanism

The teachers, organised through the Finnish teachers’ union (OAJ), name a different cause: the abolition of special-needs classes is overwhelmingly the largest cause of the rise. Students who previously would have been in segregated special-needs classes are now mainstreamed into ordinary classes. The teachers report this directly.

Whether the inclusion-policy reform is in fact the operative mechanism is an empirical question that the available data cannot definitively answer. The teachers may be right; they may be partly right; they may be naming one important factor among several. The diagnostic move this essay makes is not “the teachers are correct”; it is “the operative-mechanism candidate the teachers name is the one for which contribution accounting has not been published.”

The inclusion-policy contribution would be measurable. Comparing per-student incident rates across schools and classes with different inclusion-policy implementation profiles, controlling for student-population composition, would produce bounded contribution estimates. The data exists internally — Espoo knows which schools implemented the inclusion reform on which timeline, and Espoo knows incident rates per school. The data is not in the public assessment report. The contribution analysis has not been produced or published.

Internal evidence in Nurmi’s own interview suggests the teachers are closer to the operative mechanism. Asked whether protective equipment is used, Nurmi confirms it is — on a small scale, in the demanding-support special classes. The journalist asks whether this is radical; Nurmi acknowledges it is.

The legal context Nurmi does not invoke: the Finnish Occupational Safety and Health Act establishes the employer’s general duty of care (§ 8) and risk-assessment duty (§ 10), with personal-protective-equipment provisions in § 15. The hierarchy is explicit: elimination and isolation of risk come before protective equipment; protective equipment is last-resort mitigation when the prior controls cannot remove the hazard. A workplace where protective equipment is needed because the hazard cannot be eliminated or isolated is by definition not a workplace where the resources are in order — the elimination and isolation options have already failed.

The internal contradiction is visible on the page of Nurmi’s own interview. Inside the Closure Machine names why the contradiction survives: defensibility-not-optimality is the institutional acceptance criterion. The statement is defensible — technically not false, hedged with “small scale,” legally framed by Finland’s compulsory-education duty (which obliges municipal schools to accept any pupil within their catchment) — so it survives even when its internal contradiction is visible on the page. In the diagnostic gates of Bad Equilibria Are Not One Thing, this is an object-gate failure: the apparatus cannot form “inclusion-policy-produced violence spike” as a stable object because forming it would convert defensible-and-not-false into operative-and-binding.

The inclusion-policy reform is also a fake-closure case in the sense of Corrective Closure Ownership: a high-irreversibility decision (abolishing special-needs classes affects multi-year student cohorts) was administratively closed without a named actor legally required, funded, and able to notice when the violence trend diverged from expectation, reopen the inclusion-policy implementation, and account for what changed. The Espoo case’s second answer is what closure-without-corrective-ownership looks like several years into implementation: the decision binds; the verification rule was not predeclared; no actor owns reopening.

V. The contribution table — what computation would produce

Causal Talisman §III names the five-part operational test: share, rivals, discriminator, decision, defeater — what share does the named cause explain, against which rival causes, with what evidence would distinguish them, which actor would change which decision on what timescale, and what observation would lower confidence. The structured artifact this discipline produces against a multi-cause discourse places candidate causes in rows and the five fields in columns and requires the same evidentiary burden symmetrically across every cause regardless of which coalition deploys it. The contribution table is what mechanism-realist analysis produces; the talismanic discourse is what the substrate produces in its absence.

Applied to the Espoo school-violence discourse over the seven named causes. Each cause is its own block; the five fields appear on their own lines so the eye can scan a single field down all seven causes:

Survival of very small premature babies (Nurmi)

• Share: bounded small under plausible assumptions (very preterm babies around 0.8% of births × behavioural-disorder rate within that subgroup × violence-attribution); exact share unknown.
• Rivals: all six others below.
• Discriminator: incident rates by support-category and developmental-history variables; cohort-specific data.
• Decision: none at municipal level; at most national health-surveillance or long-range support-service planning.
• Defeater: no elevated incident rate among the relevant prematurity / developmental-history subgroup after controls.

Screen time at home (Nurmi)

• Share: unknown.
• Rivals: family-environment factors; school-environment changes; cohort effects unrelated to screen exposure.
• Discriminator: comparative incident rates by family screen-exposure profile; cohort comparison; intervention studies.
• Decision: none at municipal level directly; at most digital-literacy policy if magnitude proves large.
• Defeater: cohort comparison showing no association after family-environment controls.

National and global phenomenon (Nurmi)

• Share: a scope claim rather than an operative cause; if real, the residual after Finland-specific factors are subtracted.
• Rivals: all Finland-specific causes (composition, policy reform, resource changes).
• Discriminator: international comparative data showing the same trajectory across countries; cross-country covariate decomposition.
• Decision: none at municipal level by definition (the cause is exogenous to Espoo); at most national-education-agency or OECD coordination.
• Defeater: international comparison showing Finland diverges from comparable countries’ trajectories.

Societal malaise / welfare cuts (Liukko)

• Share: unknown.
• Rivals: inclusion-policy effects; immigration effects; cohort effects; family-formation changes.
• Discriminator: pre- vs post-policy-shift incident rates controlling for non-policy variables; service-availability against incident correlation; family-stress indicators; social-service waiting times.
• Decision: national welfare policy; municipal social-service capacity; integrated health-and-social-services planning.
• Defeater: pre- vs post-policy-shift incident comparison showing no change after service-availability controls.

Immigration and integration (social-media commentary)

• Share: unknown.
• Rivals: all other causes; specifically school composition and language-skill factors.
• Discriminator: incident rates by language background, school demographic profile, integration-stage variables (where lawful), denominator-adjusted incident decomposition.
• Decision: school catchment policy; integration support; language education capacity.
• Defeater: incident decomposition showing no concentration in language-background or integration-stage cases after denominator adjustment.

Abolition of special-needs classes (teachers’ union)

• Share: unknown; teachers identify it as the largest contributor.
• Rivals: all other causes; specifically welfare cuts and immigration / integration as overlapping in time and composition.
• Discriminator: before/after incident rates by placement model; school-level small-group availability; pupil movement after reform.
• Decision: inclusion-policy reversal; special-class capacity restoration; resource reallocation; national-education-agency policy review.
• Defeater: incident-rate change by placement model showing no correlation with inclusion-reform timing (controlled for lag and implementation intensity).

Teacher competence (Tornberg)

• Share: unknown.
• Rivals: resource adequacy; pupil profile; school-environment changes.
• Discriminator: pre/post training-intervention incident rates by teacher profile; comparison studies.
• Decision: teacher-training investment (specifying programme type, target teachers, expected effect size); pedagogical-product procurement.
• Defeater: training-intervention studies showing no incident reduction after teacher-profile and pupil-profile controls.

Symmetric application of the five-field standard yields a uniform finding: every cause has “unknown” in the share field. The discourse has produced no contribution accounting from any speaker. The pattern of unknowns is itself the table-level finding — the discourse is talismanic exchange, not adjudication.

Three differential observations the table surfaces:

Asymmetric scrutiny in the discourse, not in the table. Public reception in May 2026 decomposed Nurmi’s premature-baby claim extensively (the 0.8% figure, Duodecim’s behavioural profile, the speaker’s own global-trend disclaimer); Liukko’s welfare-cuts claim, Tornberg’s technique claim, and the right-coded immigration claims received much less of the same scrutiny. By the contribution-table standard each is in the same category — named cause without share, rivals, discriminator, decision, or defeater. The differential scrutiny tracks coalition alignment of the analysts doing the scrutiny, not analytic merit. This is the asymmetric-default failure mode the Causal Talisman symmetric-burden guardrail names.

Differential mechanism plausibility is a separate operation. The five-field standard surfaces that all seven causes are uncomputed; it does not by itself adjudicate which are more mechanistically plausible than others. Nurmi’s premature-baby claim is structurally talismanic on the standard alone — magnitude failure, decades-distal, no municipal lever, externalisation to medical progress, conflict with the speaker’s own global-phenomenon framing, no stated defeater. The other causes are unresolved by the standard alone — neither confirmed talismanic nor decomposed enough to defend. The intuition that “welfare cuts are current with identifiable actors and bounded mechanisms, unlike decades-distal medical progress” is a mechanism-plausibility judgment imported separately by the analyst. The judgment may be correct; it is not generated by the table. The table identifies where contribution accounting is missing; differential mechanism judgment is a different analytical operation that must itself satisfy symmetric-burden discipline before its differential finding is trustable.

The strongest row for operational measurement is the abolition of special-needs classes. Not because the teachers’ cause is necessarily the operative one, but because the discriminator field is more concrete than other rows: before/after incident rates by placement model is in-principle measurable from data Espoo already holds internally. The other causes either require data Espoo does not have (international comparison; family-environment profile data) or implicate actors outside Espoo’s decision-locus. If Espoo were to produce one contribution analysis, the teachers’ candidate is the cheapest to compute and the most directly relevant to municipal decisions.

The decomposition is genuinely Espoo’s responsibility. Per Nurmi’s own statement in §II, the violence/threat data exists internally. What is missing is the public structured decomposition the table specifies. The contribution table is what would convert the discourse from talismanic exchange into mechanism-realist analysis — and the structural finding of §II (the public statistic was withdrawn during trend growth) is the institutional move that prevents the table from being produced.

VI. The therapeutic-vocabulary backdrop

The therapeutic-vocabulary framework that teachers operate within is an aggression self-management programme hosted on the public mental-health portal Mielenterveystalo, reproduced through teacher training and adjacent services. Its structure carries the Structured Underspecification pathology: a principled criterion (violence is everything that subordinates or exploits another person) plus a behavioural enumeration (verbal berating, name-calling, insulting, invalidating, mocking, presented as examples of psychological violence), with no specified mapping rule between the two, plus a self-defence doctrine that authorises legitimate response to the same class of speech acts the enumeration lists as psychological violence. The overlap is exact: verbal berating appears in both lists, and the self-defence doctrine’s catch-all phrase — speech that oppresses another’s rights — restates the principled criterion. The architecture is in fact triple-warranted: the principle level collapses physical and psychological violence into one unmarked category, the self-defence trigger fires on either, and the prohibition forbids self-defence using either — with no proportionality rule between the levels. The framework’s dual warrant lets a teacher’s calibrated verbal response to student behaviour be classified as psychological violence under the enumeration reading, while the same speech acts receive principle-reading exemption when the teacher is the recipient. The architecture makes teacher-level practitioner judgment legible as misconduct or as professional judgment depending on which reading the enforcing administrator selects.

VII. What could actually move it

The Espoo case is also a clean telos-gap instance: the harm (rising violence + teachers carrying PPE-warranted risk + unmeasured pupil-cohort effects) is real and known to practitioners, but no actor’s mandate, budget, KPI, power, incentive, constituency, or trigger contains the inclusion-policy-contribution-to-violence as an actionable object. The city board owns education budget; the wellbeing-services county owns violence-response services; the Ministry of Education owns national policy; the regional occupational-safety authority owns workplace safety. Each owns part of the chain; none owns the joint vector. The Record Gate withdrawal is the institutional move that keeps the harm outside every actor’s admissible-object intersection.

The case is unusually tractable because the structural finding is contained, documented, and procedurally addressable:

• Workplace-safety complaint to the regional authority. Finland’s regional occupational-safety authorities have consequential authority on workplace safety, including conditional fines. The protective-equipment admission in Nurmi’s own interview is direct evidence for a safety complaint. A formal complaint citing the Occupational Safety and Health Act §§ 8, 10, and 15 would oblige the authority to investigate. This is the most direct procedural route available.
• Journalistic follow-up on the reporting withdrawal. Not on the premature-baby absurdity. A focused piece asking why the assessment-report format was changed exactly during the trend growth would create reputational cost on the city board high enough to potentially restore the statistic.
• Municipal resident initiative. Espoo residents file a formal initiative requiring the city council to consider restoring annual violence-incident reporting.

The structural finding (statistic withdrawal during trend growth) is more actionable than the rhetorical finding (premature-baby absurdity). The latter discharges as outrage and fades; the former is admissible to the procedural objects above. The contribution-table artifact §V is what each of these procedural objects can carry as substantive content — the table is the deployable that the discourse has not produced and that the procedural objects would oblige the institution to produce.

VIII. Why the pattern matters beyond Espoo

The shape recurs across contained domains where a multi-cause discourse meets an institution holding the relevant data internally but not publicly: a visible rhetorical absurdity absorbs attention; a hidden Record Gate failure or refusal-to-compute event lets the apparatus avoid the substantive decomposition; multi-actor talismanic explanations converge on protecting the apparatus’s preferred conclusion through structurally different routes; the operative-mechanism candidate named by practitioners is dismissed across actor classes through different talismans; institutional response capacity that should detect the structural move is thin or captured. The Attenuation Layer describes the parent mechanism — the persistent institutional pattern that routes unresolved value conflicts toward whoever can least convert their damage into named blame, with school violence as one of its principal worked corridors; the Espoo case is an instance where the binding repair (aggregate visibility) was once in place and then revoked. Espoo is unusually clean because all elements are on the public record — the apparatus has the data, the talismanic explanations are circulating, the practitioner candidate is named, the Record Gate move is documented in the same interview that produced the absurd quote. Other domains (health-and-social-services reforms, state-grant redistribution, eldercare provisioning) produce structurally identical cases with substrate-specific carriers; the architecture is general, the named carriers interchangeable.

IX. Close

Espoo did not stop knowing the violence statistic. Espoo stopped making the statistic publicly binding. That sentence is the load-bearing finding. The premature-baby absurdity, the societal-malaise cluster, the teacher-technique frame are the rhetorical theatre that made the structural move tolerable. The contribution table in §V is the artifact mechanism-realist analysis produces against a multi-cause discourse; its absence in the Espoo public discourse is what made the Record Gate withdrawal survivable. The procedural objects in §VII are reachable; whether they are reached depends on whether the structural finding survives the rhetorical theatre that obscured it. The first move toward that survival is naming the headline finding correctly — not “Nurmi said a stupid thing about premature babies” but “Espoo removed the violence statistic from public binding cognition exactly when the trend was growing, and the talismanic discourse around the case made the withdrawal discursively tolerable.”

Related:

• The Causal Talisman — companion diagnostic essay. This worked-case essay carries the artifact-form (the contribution table §V) and the symmetric-burden guardrail in operation.
• The Attenuation Layer — the parent mechanism. The Record Gate move §II names is Repair 1 (aggregate visibility) actively revoked; the contribution-table absence §V is the attribution-gradient diagnostic; the Nurmi protective-equipment admission is the responsibility-without-power asymmetry in §IV.
• The Record Gate — the structural move §II names: cognition held below the binding threshold.
• Structured Underspecification — the discourse vocabulary the case is adjudicated within (§VI backdrop).
• What Bureaucracy Is — the Slot-Function Fallacy fold names the form Nurmi’s defence takes.
• Inside the Closure Machine — defensibility-not-optimality as institutional acceptance criterion; names why Nurmi’s internally contradictory statement survives.
• The Refusal to Compute — institutional failure mode behind the missing contribution accounting.
• The Procedural Object — the form §VII repair routes must take to compel institutional response.
• Feedback Authority — the procedural→advisory downgrade of the violence statistic (§II) is a worked instance of the conflation pattern.
• The Response Vector — the architectural reason the talismanic discourse is cheaper than a contribution table (§III).
• Bad Equilibria Are Not One Thing — object-gate failure in §IV: the apparatus cannot form “inclusion-policy-produced violence spike” as a stable object because forming it would convert defensible into binding.
• Corrective Closure Ownership — the inclusion-policy reform as fake-closure case (§IV).
• The Telos Gap — cross-silo unowned harm in §VII; the Record Gate move keeps the harm outside every actor’s admissible-object intersection.