Field Notes

synthesis

Hidden miscarriage risk is ubiquitous among top-ranked democracies, and Sweden’s correction channel is unusually opaque

StatusStableReviewdue UpdatedStands on12 sourcesSupportsnotes

Sweden ranks fourth of 143 countries in the 2025 WJP Rule of Law Index and sixth on its criminal-justice factor, behind only the other Nordics and New Zealand on the overall list.1 Rättssäkerhet in Swedish criminal cases documents that this certification coexists with wrongful-conviction mechanisms that Sweden’s own state inquiries exposed, a reopening gate among the strictest in Europe, and corrections that arrive through journalism rather than through any institution. The question this note answers is comparative: do other highly ranked democracies hide the same kinds of failure, or is Sweden exceptional?

The answer has three parts. First, the epistemic structure is ubiquitous: no country examined here can count its wrongful convictions, every country’s visible corrections measure the throughput of its Correction channel rather than its error rate, and every top-ranked country has documented miscarriages that its rankings could not detect. Second, the rankings are structurally blind to this everywhere: they are perception instruments with no variable that could register a hidden wrongful conviction in any country. Third, Sweden nonetheless stands out within its Nordic peer group on one dimension: the opacity of its correction channel. It is the only Nordic country with neither an independent or special reopening body nor published channel statistics, and the only one whose detected corrections depend entirely on unfunded outsiders. Sweden’s hidden flaws are not unique; Sweden’s refusal to count or fund the looking comes close.

The rankings cannot see the error rate

The WJP index builds its scores from a household poll of about 1,000 people per country and expert questionnaires averaging about 29 respondents per country. Its criminal-justice factor aggregates perceptions of investigative effectiveness, timeliness, corrections, impartiality, corruption-freedom, independence, and due process.2 No underlying variable counts wrongful convictions, tracks reopening applications or their outcomes, or audits closed cases against independent evidence. The index measures the system as it presents itself to the lawyers and citizens it surveys.

This is not a defect the WJP hides; it is the declared design. The consequence for this note is that a country’s high rank certifies that its institutions are experienced as competent and clean, not that their outputs are accurate. Versteeg and Ginsburg’s comparison of four major rule-of-law indicators found them correlated above 0.95 and argued the indicators converge on a generalized halo of impartial administration rather than measuring their distinct constructs — a critique that applies to the entire genre.3

The existence proof of the blind spot is the leaderboard itself. Japan ranks fifteenth overall while its courts convict 99.8 percent of tried defendants and its bail system holds unconfessed suspects hostage, as HRW documents.4 The United Kingdom ranks fourteenth overall; its most widespread miscarriage — hundreds of sub-postmasters prosecuted on the output of buggy accounting software — ran for fifteen years before anyone in authority looked.5 Iceland is a fixture near the top of every good-governance index, and its most famous case is six people convicted of two murders that may never have happened, on confessions extracted under up to 655 days of solitary confinement.6 Rankings like the WJP’s are real measurements of real things, but the things they measure are not the things Rättssäkerhet in its outcome sense is about.

Four correction-channel architectures

Every system examined here privileges finality and then makes an exception to it. What differs is who holds the gate, what opens it, who is equipped to knock, and whether anyone counts the knocks. Four architectures cover the examined countries:

Architecture Countries Gatekeeper Publishes statistics Funded detection
Independent investigative commission Norway (2004), England/Wales (1997), Scotland (1999), New Zealand (2020), Canada (2024, not yet operational) Commission with own investigators and counsel funding Yes, annually The commission itself
Prosecutorial review United States (CIUs), Netherlands (Procurator General + ACAS) The side that obtained the conviction Partial, via registries Innocence organizations and journalism
Special court inside the judiciary Denmark (Klageret, 1939), France (révision formation, 2014) Senior judges Sparse Journalism and advocates
Ordinary courts, no special body Sweden, Finland, Germany, Japan The convicting hierarchy itself Sweden: none Journalism and private advocacy only

The table’s lesson is that the visible exoneration count of a country is largely a readout of which row it occupies. Commissions with investigators and counsel funding find hundreds of cases. Prosecutorial units find what prosecutors are willing to see, supplemented by whatever private innocence infrastructure exists. Court-internal systems find almost nothing unless journalists or families do the investigating. And no architecture produces a number that can be read as an error rate, because every channel observes only what entered it.7

The Nordic natural experiment

Norway and Sweden are the controlled comparison: same legal family, same evidentiary culture, same welfare-state context, and until 2004 the same court-internal reopening structure. Norway’s scandal sequence — Liland acquitted in 1994 after NRK journalism demolished his axe-murder conviction, the Moen cases, Rødseth — drove it to lift reopening out of the courts into an independent commission. Sweden’s scandal sequence — Bergwall, the Felaktigt dömda cases, Kaj Linna — produced reports and doctrinal adjustments but no institutional change.8

The throughput difference is an order of magnitude:

  • Norway, 2004-2021: 3,065 applications, 433 cases reopened — roughly 20 reopenings a year in normal years, 15-18 percent of merits-reviewed applications, in a country of 5.4 million.9
  • Sweden: 40-60 granted resning applications per decade between 1950 and 1988, in a country now of 10.4 million; in 2015, 383 applications decided, 7 granted (3 percent), with fact-based innocence claims granted at 0.9 percent (1 of 110).1011

Sweden’s grant pattern is not merely low; it is asymmetric in a way that says who the channel serves. Across 5,732 researcher-assembled applications (1935-2020), prosecutor applications were granted at about 98 percent, private applicants at 6.7 percent, unrepresented private applicants at 2.5 percent, and unrepresented applicants pressing factual claims at Högsta domstolen 2005-2010 at 1.25 percent.12 The channel is open to the state and nearly sealed to the unaided convicted person.

Norway’s commission is not a refutation of the hiddenness thesis; it is its best confirmation. Even this gold-standard channel denied Viggo Kristiansen’s Baneheia application six times before granting the seventh, after a defense-commissioned DNA report, a book, and a documentary had made the error undeniable outside the system. The commission’s own expert then found what the defense’s expert had found eleven years earlier, and Kristiansen was acquitted in December 2022 of a double child murder for which he had spent two decades in preventive detention.13 The Torgersen case — Norway’s longest-running contested conviction — was refused again in 2021 and remains unreopened.9 An independent commission with investigators is a better gate, not a window into ground truth.

The other Nordics have their own cases

Every Nordic country has produced a flagship miscarriage with the same family resemblance to Sweden’s Bergwall case: suspect-shaped investigation, confession or child-witness centricity, and correction driven from outside.

Iceland. In the Guðmundur and Geirfinnur case, six young people were convicted in 1980 of murdering two men who had vanished in 1974 — no bodies, no forensic evidence, no witnesses, only confessions extracted under solitary confinement lasting up to 655 days, with drugs, sleep deprivation, and water torture documented in the case of the alleged ringleader. Iceland’s Supreme Court rejected reopening in 1998. A government working group in 2013, a special prosecutor, and Gísli Guðjónsson’s memory-distrust analysis preceded the acquittal of five of the six in September 2018, forty-four years after the events; the sixth, Erla Bolladóttir, has never had her perjury conviction reopened. Amnesty reported in 2023 that Icelandic police still use pre-trial solitary confinement excessively.6

Finland. Anneli Auer was twice convicted of murdering her husband and twice saw the verdicts overturned, then was convicted with her former partner Jens Ihle of sexual offences against her own children on allegations that emerged after the children entered foster care — the same allegation-generation mechanism the Swedish Felaktigt dömda report documented. Both were definitively acquitted in 2025 after roughly four years in prison, and their pending compensation claims are the largest in Finnish history. Finnish wrongful-imprisonment compensation payouts quadrupled over the decade to 2017.1415

Denmark. Denmark’s channel is court-internal like Sweden’s, through Den Særlige Klageret, a special complaints court created in 1939 — older than any commission but structurally closer to Sweden’s model than to Norway’s: it sits inside the judiciary, shares a secretariat with the Supreme Court, and by design cannot reopen a case for general failure of due process, only on enumerated grounds.16 Denmark’s exoneration visibility is correspondingly low: its flagship EUREX entry, Ove Thomas Hansen, dates to a 1992 acquittal.

The commission countries still miss systemic error

England and Wales built the model commission in 1997, and the UK record shows both what a commission can do and what it cannot.

The CCRC’s long-term referral rate is about 3 percent of applications, and it fell below 1 percent in 2016-18 — a level one solicitor told a government review was “quickly moving towards the point of vanishing.” Success rates of referred appeals run 64-79 percent, so the gate is calibrated to refer only near-certainties.1718 A 2026 University of Exeter study of hundreds of CCRC case evaluations found the new-evidence requirement leaves the CCRC structurally unable to refer exactly the convictions most likely to be wrong: those resting on the weakest original evidence, where nothing new exists to find. Andrew Malkinson spent seventeen years in prison on such a case before DNA cleared him.19 This is the English form of the Swedish Billy Butt problem: the denied set marks where the channel’s threshold sits.

The Post Office Horizon scandal then demonstrated the channel’s scaling limit. More than 900 sub-postmasters were prosecuted between 1999 and 2015 on shortfalls produced by defective software whose known faults were never disclosed. The CCRC referred 77 cases over four years — its largest-ever investigation — and the remaining hundreds of convictions were finally quashed not case by case but by an Act of Parliament in May 2024, after the government’s own compensation advisory board concluded the individual-case machinery was structurally incapable of processing a systemic error.520 The lesson generalizes: every correction channel examined here is built for retail error and fails silently on wholesale error, and the UK’s could not see its largest miscarriage as a class until journalism, litigation, and a public inquiry forced the frame to change.

New Zealand’s commission (2020) and Canada’s (legislated December 2024, not yet operational) are the same institutional diffusion: each followed a case the old channel fumbled — Teina Pora in New Zealand, whose confessions the Privy Council found unreliable after two trials and twenty years;21 David Milgaard in Canada, where ministerial review had referred twenty cases in two decades and every public inquiry since 1989 had recommended an independent commission.22 Australia remains the counterexample: no state has a commission, and outside South Australia and Tasmania the only route is petitioning the Attorney-General for the prerogative of mercy — executive gatekeeping that an Australian Human Rights Commission submission warned converts an innocence claim into a public-relations campaign.23

The restrictive-novum continent

The Netherlands shows that loosening a finality gate by statute changes less than the statute promises. Dutch revision was historically so restrictive that new expert evidence routinely failed the novum test; a 1992 study found 346 revision requests in 1979-91 with 71 granted, mostly minor cases.24 After the Schiedam Park Murder — in which the Supreme Court dismissed a revision request in September 2004 for lack of new evidence while the prosecution service sat on information identifying the real perpetrator, who had confessed a month earlier — the Netherlands widened the criterion in 2012 and added a pre-revision investigation procedure.2425 The five-year official evaluation found fewer applications than before, a novum criterion still too strict to admit the apparently unsafe Arnhem Villa Murder convictions, an advisory committee acting as an over-strict pre-filter, and access problems from mandatory representation.25 The legislature had considered and rejected an independent review commission as unnecessary bureaucracy.

Germany and Japan mark the restrictive end. Germany has no review body; Wiederaufnahme applications go to the courts, and the Mollath case shows the pattern: a man held seven years in forensic psychiatry on a fabricated-diagnosis commitment, whose reopening the Landgericht Regensburg refused until journalism and a federal constitutional complaint forced the Oberlandesgericht to grant it; he was acquitted in 2014 and compensated at a statutory €75 per day.26 Japan’s retrial system is nearly sealed: five death-row retrial acquittals since 1945, and the Hakamada case needed forty-four years of petitions, the disclosure of six hundred pieces of withheld evidence, and a court finding that investigators had fabricated and planted evidence, before the world’s longest-serving death-row prisoner was acquitted in September 2024.27 France’s Outreau affair — thirteen innocent people sent through assize trials, more than twenty-five cumulative years of pre-trial detention, one suicide in custody — produced a parliamentary inquiry in 2006 whose report estimated, in its introduction’s framing, that of 20,000 remand prisoners some 2,000 would probably eventually be recognized innocent: “as many Outreaus of which we will probably never hear.”28 That sentence, written by a parliamentary commission about its own highly ranked legal system, is the hidden-figure thesis stated from inside a state institution.

The United States: industrialized detection, industrial-scale production

The US is the only country that built a detection industry: a national registry, roughly seventy innocence organizations, and prosecutorial Conviction Integrity Units. The result is the world’s largest visible error count — 3,646 exonerations since 1989, 147 in 2024, with official misconduct in 71 percent and perjury or false accusation in 72 percent — and the registry’s own finding that the entire growth in annual exonerations tracks the proliferation of professional exonerators, not changes in adjudication.29 America’s visibility is high because its channel is funded, which is exactly the comparative point: visibility is produced.

The dark figure is correspondingly specific. About 95 percent of US convictions come from guilty pleas, and the registry concedes that innocent defendants who plead guilty face barriers so high they are “so unlikely to be exonerated at all” that it cannot see them.30 The only calibrated population estimate — at least 4.1 percent of death sentences are false convictions, per the survival analysis in Wrongful conviction — covers the one corner of the system under maximum scrutiny. Cassell’s skeptic counter-estimate, 0.016-0.062 percent, argues the innocence literature overstates the rate by orders of magnitude.31 For this note the dispute matters less than its shape: both sides extrapolate from channel-selected samples (exonerations, death sentences), and neither has a denominator the channel did not select. The most transparent system in the comparison still cannot state its own error rate; it can only state that its officials are involved in most of the errors it catches.

Why high-trust systems are self-sealing

The comparative record supports a stronger claim than “every system hides errors.” High-trust systems face a specific epistemic trap:

flowchart LR
  A[Narrow correction channel] --> B[Few visible corrections]
  B --> C[High perceived reliability]
  C --> D[Finality doctrine appears validated]
  D --> A
  E[Scandal plus independent gatekeeper] -.->|can break the loop| D

Each element is documented somewhere in this note. The narrow channel suppresses the correction count (Sweden’s 1.25 percent, the CCRC’s “vanishing” referral rate, Japan’s five retrials in eighty years). The suppressed count is then cited as evidence of accuracy — the WJP’s experts perceive a reliable system, the Dutch legislature cited trust in the judiciary when defending its restrictive novum, and Sweden’s Supreme Court presents finality as the starting point. The perception justifies keeping the channel narrow, because widening it is framed as undermining confidence that the loop itself manufactured. Norway’s history shows the loop’s breaking condition: a scandal the frame could not absorb (Liland, via journalism), followed by moving the gate outside the institutions being checked.

Sweden has absorbed scandal after scandal — the JK’s two rättssäkerhet projects, the Bergwall commission, Kevin, Da Costa, Linna — without the loop breaking, because each was metabolized as an unfortunate exception rather than as evidence about the channel. The prediction this framework makes is that Sweden’s true error profile cannot be inferred to be better or worse than Norway’s was in 2003, when Norway also believed its wrongful convictions were vanishingly rare — because Norway had not yet built the institution that found hundreds of cases to reopen.

Does Sweden stand out?

In kind, no. Every examined democracy lacks ground truth for historical guilt, privileges finality, routes correction through a threshold-shaped channel, and carries documented miscarriages invisible to its rankings. The failure mechanisms are shared too: confirmation-biased investigation, confession centricity, expert overreach, non-disclosure, and institutional defensiveness appear in the Dutch Schiedam inquiry, the French Outreau inquiry, the Norwegian Liland and Baneheia histories, the Icelandic working group, the Canadian inquiries, and the US registry’s misconduct statistics as surely as in Sweden’s JK reports. Journalism-dependent correction, which feels like a Swedish pathology, is also the documented pattern in pre-commission Norway (Liland), Germany (Mollath), Iceland (the Reykjavik confessions), New Zealand (Pora), and the Netherlands (Schiedam).

In degree, yes, on three specific dimensions:

  1. Opacity. Sweden is the only Nordic country with neither an independent or special reopening body nor published channel statistics. Norway publishes everything annually; Denmark’s special court publishes its decisions; Sweden cannot say how many people ask it for resning in a year.
  2. Asymmetry. Sweden’s 98-to-1.25 spread between prosecutor and unrepresented private grant rates has no documented equivalent in the comparison; Norway’s commission was built precisely to eliminate the equivalent asymmetry.
  3. Institutional null. Sweden has no review commission, no conviction integrity unit, no exoneration registry, and no innocence organization — every detected correction in the vault’s Swedish case file was produced by journalists or private advocates working without funding or powers.

Against that, Sweden is not at the restrictive extreme of the comparison: Japan’s channel is more sealed, Australia’s more political, and pre-2012 Netherlands’ was comparably narrow. Nor is Sweden distinctive in having a finality doctrine; it is distinctive in combining one with no instrument for seeing past it.

The honest ranking this evidence supports is therefore not a league table of error rates — no such table is constructible from channel data — but a ranking of how much a country can know about its own errors:

Tier Countries What the country can know
Structured visibility Norway, England/Wales, Scotland, (Canada, NZ pending) Channel throughput, grounds, and outcomes, published annually; errors the commission can catch
Funded partial visibility United States Registry-scale discovered errors, skewed to serious crimes and active CIU counties; nothing on pleas
Statutory visibility, weak channel Netherlands, Denmark, France Decision-level data, little investigation capacity
Darkness Sweden, Finland, Germany, Japan, Australia Individual scandals only, when outsiders force them

Sweden sits in the darkest tier with the systems most confident in their own outputs. That is the comparative finding: the problem the vault documents is ubiquitous, but Sweden’s inability to see it is near-maximal among peers that the rankings treat as equivalent.

Limitations and counterevidence

The skeptic position. Cassell argues US wrongful-conviction rates are far below the innocence literature’s figures, and his 0.016-0.062 percent range, if correct, would make hidden error a marginal problem.31 Two responses bound this: his estimate shares the extrapolation structure it criticizes (it assembles trial error rates, plea ratios, and exoneration proportions, all channel-selected), and it is computable only for the US, which is the one country with data to argue over. The Swedish dark figure is darker than Cassell’s darkest scenario precisely because Sweden offers no data on which any estimate could rest.

Front-end safeguards. Stridbeck and Magnussen argue Norwegian convictions are genuinely safer because of corroboration requirements and identification-procedure rules — production-side quality, not just channel-side visibility.8 The point stands against pure channel skepticism, but its force is limited: Sweden lacks several of those safeguards (free evaluation has no corroboration floor), and the JK reports documented failures in a system with its own safeguard list. Front-end differences shift the prior; they do not license confidence anywhere.

Selection in this note’s own evidence. Flagship cases cluster where journalism and researchers looked hardest (the Nordic cases, the UK, the Netherlands). Germany, France, Spain, Italy, and most of Europe have no registry at all; EUREX’s 144 cases across 20 countries reflect research attention as much as true error distribution.32 The comparison set here is weighted toward rich democracies because those are the countries Sweden is ranked among.

Small numbers. Swedish grant data come from researcher-assembled samples with method differences, not official statistics — which is itself the finding, but it means individual cells (like the 2-of-2 prosecutor figure for 2005-2010) are thin. The order-of-magnitude Nordic throughput contrast rests on different structures over different periods and is presented as that, not as a precise ratio.

What this changes for the vault’s Swedish critique

The international frame cuts in one direction against two misreadings. Against the misreading that Sweden is uniquely corrupt: every mechanism documented in Rättssäkerhet in Swedish criminal cases has a documented foreign counterpart, and several peer systems score worse on dimensions Sweden is criticized for. Against the misreading that everyone is like this, so the critique is overblown: Norway’s commission demonstrates that the same legal culture can build a channel that finds errors, and Sweden’s refusal to build one — after two state inquiries and a parliamentary commission documented the need — is a choice with measurable consequences for who gets found and who stays convicted.

The comparative evidence also sharpens the vault’s central epistemic claim. Correction channel argued that correction counts measure channels. The international record is that claim’s field test: channel throughput varies by an order of magnitude across countries whose true error rates no one can compare, and the country that built the best window found its own errors immediately. Sweden’s low visible exoneration rate is therefore evidence of exactly one thing: the shape of Sweden’s gate. The same is true, with less force, of everyone else.

Open research

  • Obtain Den Særlige Klageret’s grant statistics and grounds distribution for a numeric Denmark row.
  • Request EUREX’s per-country dataset (the site’s country counts are script-rendered and could not be fetched).
  • Establish Finland’s reopening application volume and whether Finland has debated a commission.
  • Import SvJT 2021/574 on Norway’s reopening law history (the 1969 rejection of the Danish model, the 1998 CCRC catalyst article).
  • Track Canada’s Miscarriage of Justice Review Commission operationalization and first published statistics.
  • Track the CCRC’s Capture-system referrals and any statutory response to the Exeter findings.
  • Check whether Japan publishes retrial-petition volumes beyond the five known death-row grants.
  • Ground Baneheia case and Guðmundur and Geirfinnur case as full event notes if they attract more vault use.

  1. World Justice Project, Sweden country profile, 2025 Rule of Law Index (saved copy): overall rank 4/143 (0.85), criminal justice rank 6/143. Overall top four: Denmark, Norway, Finland, Sweden; on the criminal-justice factor Finland ranks 2nd and Norway 3rd. ↩

  2. World Justice Project, Rule of Law Index 2025 methodology snapshot (saved copy). Data sources: General Population Poll (~1,000 respondents/country) and Qualified Respondents’ Questionnaires (average ~29 expert respondents/country); criminal justice sub-factors 8.1-8.7 as described in the section above. ↩

  3. Mila Versteeg and Tom Ginsburg, “Measuring the Rule of Law: A Comparison of Indicators,” Law & Social Inquiry 42, no. 1 (2017): 100-134, doi:10.1017/lsi.2017.57. Pairwise correlations among three of the four indicators exceed 0.95; the authors attribute convergence to measurement strategy (expert perceptions, information constraints) rather than to a shared concept. ↩

  4. Human Rights Watch, Japan’s “Hostage Justice” System (2023) (saved copy). Conviction rate 99.8 percent of tried cases per 2021 Japanese Supreme Court statistics; “hostage justice” (hitojichi-shiho) is the domestic critics’ term. Japan’s 2025 WJP overall rank: 15/143. ↩

  5. Criminal Cases Review Commission, “Post Office ‘Horizon’ cases” (2025): “the biggest single series of wrongful convictions in UK legal history”; 77 convictions referred by the CCRC; the Post Office (Horizon System) Offences Act 2024 quashed the remaining convictions by statute. UK 2025 WJP overall rank: 14/143. ↩↩

  6. Gísli H. Guðjónsson, “The Science-Based Pathways to Understanding False Confessions and Wrongful Convictions,” Frontiers in Psychology 12 (2021), doi:10.3389/fpsyg.2021.633936; Iceland Monitor, “All found innocent in Guðmundur and Geirfinns case,” 27 September 2018; BBC, “The Reykjavik Confessions”. Tryggvi Leifsson spent 655 days in solitary; Sævar Ciesielski 1,533 days in custody total; Erla Bolladóttir’s perjury conviction stands. Amnesty International (2023) on continued excessive pre-trial solitary confinement, cited via the case’s public record. ↩↩

  7. Correction channel — the vault’s concept note formalizing why discovered-error counts measure the channel, not the error rate, and why an institution cannot calibrate itself. ↩

  8. Stridbeck and Magnussen, “Prevention of Wrongful Convictions: Norwegian Legal Safeguards and the Criminal Cases Review Commission,” University of Cincinnati Law Review 80 (2012/2013) (saved copy). For the commission’s legal history, see also “Gjenåpning av straffesaker i Norge,” Svensk Juristtidning 2021 s. 574 (not yet imported). ↩↩

  9. Norwegian Criminal Cases Review Commission annual report 2021 (saved copy): 263 applications and 82 reopenings in 2021; 3,065 received and 433 reopened 2004-2021; Baneheia reopened February 2021; Torgersen application refused June 2021. ↩↩

  10. Dennis Martinsson, “Exoneration in Sweden: Is It Not about Time to Reform the Swedish Model?”, Erasmus Law Review 2021:4 (saved copy): JK-project figure of 40-60 grants per decade 1950-1988; no official Swedish statistics; reform proposals. ↩

  11. Sara Hellqvist, “The Narrow Road to Exoneration — the Incidence, Characteristics and Outcomes of Wrongful Conviction Claims in Sweden over a One-Year Period,” Bergen Journal of Criminal Law & Criminal Justice 5, no. 2 (2017), doi:10.15845/bjclcj.v5i2.1461: 216 applications decided in 2015 after exclusions, 7 approved (3 percent), fact-based claims 1 of 110 (0.9 percent), law-based claims 6 of 55 (11 percent). ↩

  12. Moa Lidén, “Resning i brottmål i Sverige,” Svensk Juristtidning 2021 s. 508-532 (saved copy), Table 1: 5,732 applications, 1935-2020. Prosecutor applications granted ~98 percent across periods; unrepresented private applicants 2.51 percent; HD 2005-2010 fact claims 1.25 percent (15/1201). ↩

  13. News in English, “State apologizes for grave miscarriage of justice,” 21 October 2022; NRK, “DNA-ekspert: Burde vært gjenopptatt for 13 år siden,” 23 October 2022; Dagbladet, “Viggo Kristiansen frifinnes,” 21 October 2022. Kristiansen applied seven times; the commission’s reopening vote was 3-2; acquitted by Borgarting Court of Appeal 15 December 2022; the “two perpetrators” DNA reading is now attributed to contamination. ↩

  14. Yle News, “Wrongfully imprisoned Jens Ihle and Anneli Auer seek millions in compensation,” 4 August 2026: Ihle claims over €2 million, Auer €1.3 million; acquittals final after the prosecutor general declined to appeal. ↩

  15. Yle News, “Wrongful imprisonment claims in Finland quadruple in 10 years,” 7 May 2018: ~€3 million paid in 2017 versus €720,000 in 2007; spike driven by five individuals; typical large claims follow aggravated drug or sexual-offence charges. ↩

  16. Jens Frederik Hansen, “Danish Criminal Procedure — Chapter 12: Resumption” (author’s chapter publication, 2018): the Klageret “cannot order resumption in case of a failure to apply due process of law” — no general unsafe-conviction safety valve. On the court’s composition (three judges, an academic jurist, a practising attorney) and history, Stanley Anderson, “The Special Court of Complaints in Denmark,” Scandinavian Studies in Law (1978). ↩

  17. Ministry of Justice, Tailored Review of the Criminal Cases Review Commission (2018): long-term referral rate 3.3 percent; 0.8 percent in 2016/17 and 0.5 percent in the first three quarters of 2017/18; the “vanishing” quotation at 7.30. ↩

  18. CCRC, Annual Report and Accounts 2023-24: 25 referrals (4 percent of closed cases); 19 of 24 referred appeals allowed (79 percent); ~1,400-1,600 applications per year. ↩

  19. University of Exeter, “Restrictive rules surrounding CCRC review are preventing the efficient identification of some miscarriages of justice,” 10 February 2026: review of 400+ CCRC case evaluations by the Evidence-Based Justice Lab; convictions resting on the weakest trial evidence are the hardest to refer because nothing “new” exists; Andrew Malkinson cited as the example. ↩

  20. Horizon Compensation Advisory Board, paper on the appeals system (2024): absent systemic change, “the palpably unjust convictions of several hundred individuals appear likely to be maintained — and that constitutes a major affront in a civilised state”; recommends treating the cohort as a systemic case. ↩

  21. Pora v The Queen [2015] UKPC 9 (Judicial Committee of the Privy Council, 3 March 2015): convictions quashed because Pora’s confessions, explained by his then-undiagnosed fetal alcohol spectrum disorder, could not be relied on; the case catalyzed the Criminal Cases Review Commission Act 2019 (in force 1 July 2020). ↩

  22. Department of Justice Canada, “David and Joyce Milgaard’s Law receives Royal Assent,” 18 December 2024 and the LaForme/Westmoreland-Traoré report, A Miscarriages of Justice Commission (2022): the Minister directed 8 new trials and 12 new appeals in the 20 cases remedied since 2003; inquiries since the 1989 Marshall inquiry uniformly recommended an independent commission. ↩

  23. New Zealand government review Q&A summarizing Australian arrangements; B. Moles, “South Australia and Tasmania” appeal-rights analysis; Victoria University Law and Justice Journal, comparative CCRC analysis. South Australia’s 2013 second-appeal statute (fresh and compelling evidence, substantial miscarriage) followed in Tasmania 2015; other states rely on Attorney-General referral petitions. ↩

  24. Chrisje Brants, “Wrongful Convictions and Inquisitorial Process: The Case of the Netherlands,” University of Cincinnati Law Review 80 (2012): 1069ff: 346 revision requests 1979-1991, 71 granted (~21 percent), mostly minor offences; the Schiedam Park Murder chronology, including the Supreme Court’s September 2004 dismissal and the prosecution’s handling of the real perpetrator’s August 2004 confession. ↩↩

  25. Holvast, Nan, and Lestrade, “Between Legal Certainty and Doubt,” Erasmus Law Review 2020:4 (saved copy): the 2012 novum widening, the ACAS pre-investigation procedure, and the three persisting problems (novum strictness, ACAS pre-filtering, representation costs). ↩↩

  26. Bundesverfassungsgericht, press release 56/2013 on the Mollath constitutional complaint; Der Spiegel, “Gericht ordnet Wiederaufnahme an,” 6 August 2013; tagesschau FAQ on the 2014 retrial acquittal and compensation. The 2003 HypoVereinsbank audit surfaced in 2012 corroborated the “paranoid delusion” itself. ↩

  27. AP, “Japanese court acquits the longest-serving death row inmate,” 26 September 2024 and Amnesty International’s statement the same day: fifth postwar death-row retrial acquittal; the retrial court found evidence fabricated and planted; 44 years from final verdict to retrial ruling. ↩

  28. Assemblée nationale, rapport n° 3125 (2006), commission d’enquête sur l’affaire d’Outreau (628 pages, 221 witnesses); the 2,000-of-20,000 figure is the report introduction’s own extrapolation from remand statistics, cited here as that commission’s framing, not as a measured prevalence. See also the Inspection générale des services judiciaires report on the case’s judicial handling. ↩

  29. National Registry of Exonerations 2024 annual report: 3,646 exonerations 1989-2024; 147 in 2024; misconduct, perjury, and professional-exonerator shares as stated. The registry site blocked automated download; figures are from the report’s published text. ↩

  30. National Registry of Exonerations, Guilty Pleas and False Confessions: guilty pleas account for ~95 percent of US convictions; 10 percent of non-drug exonerations include guilty pleas; innocent defendants who plead guilty “are so unlikely to be exonerated at all.” ↩

  31. Paul G. Cassell, “Overstating America’s Wrongful Conviction Rate? Reassessing the Conventional Wisdom About the Prevalence of Wrongful Convictions,” Arizona Law Review 60 (2018): 815, doi:10.2139/ssrn.3276185: assembled estimate of 0.016-0.062 percent; the contrary population estimate (at least 4.1 percent of death sentences) is Gross et al. 2014, discussed in Wrongful conviction. ↩↩

  32. European Registry of Exonerations: 144 exonerations across 20 European countries as of 2025, cases back to 1970; the registry itself notes Europe’s problem is “very much unknown and difficult to assess” compared with North American registries. ↩

Built on 12 sources (8 archived here, 4 external).

Working out connections…