{
  "version": "https://jsonfeed.org/version/1.1",
  "title": "Licklider — Latest",
  "home_page_url": "https://www.licklider.ai/latest/",
  "feed_url": "https://www.licklider.ai/feed.json",
  "description": "Research, engineering, company news, and practical Blog articles from Licklider.",
  "language": "en",
  "authors": [
    {
      "name": "Licklider, Inc.",
      "url": "https://www.licklider.ai"
    }
  ],
  "items": [
    {
      "id": "https://www.licklider.ai/engineering/scipy-student-t-subnormal-tail-loss/",
      "url": "https://www.licklider.ai/engineering/scipy-student-t-subnormal-tail-loss/",
      "title": "SciPy returns zero for a representable Student-t tail",
      "summary": "A Welch test and two lower-level Student-t functions return zero for a positive subnormal probability that remains representable in binary64.",
      "content_text": "A Welch test and two lower-level Student-t functions return zero for a positive subnormal probability that remains representable in binary64.\n\nStatus: Closed as not planned on September 30, 2026 — MPArray recommended; no NumPy-backend fix merged\n\nPublished September 27, 2026 (UTC). Updated October 1, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/scipy-student-t-subnormal-tail-loss/",
      "date_published": "2026-09-27T17:37:21Z",
      "date_modified": "2026-10-01T03:59:39Z",
      "tags": [
        "Engineering",
        "Upstream report"
      ]
    },
    {
      "id": "https://www.licklider.ai/news/nomue-protocol-welch-successor-public-discussion/",
      "url": "https://www.licklider.ai/news/nomue-protocol-welch-successor-public-discussion/",
      "title": "nomue Protocol opens public discussion on an exact-value Welch successor",
      "summary": "Public comment is open on a proposed bounded successor for Welch checks, with exact-value evidence, explicit domain limits and a Contract-bearing Record format.",
      "content_text": "Public comment is open on a proposed bounded successor for Welch checks, with exact-value evidence, explicit domain limits and a Contract-bearing Record format.\n\nStatus: Public discussion open — proposed bounded successor; Release 1 unchanged; no new supported capability\n\nPublished September 27, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/news/nomue-protocol-welch-successor-public-discussion/",
      "date_published": "2026-09-27T16:53:10Z",
      "date_modified": "2026-09-27T16:53:10Z",
      "tags": [
        "News",
        "Public consultation"
      ]
    },
    {
      "id": "https://www.licklider.ai/research/historical-analysis-reuse/",
      "url": "https://www.licklider.ai/research/historical-analysis-reuse/",
      "title": "Reducing the cost of historical-analysis reuse decisions with nomue: a paired, equal-evidence LLM evaluation",
      "summary": "Adding nomue reduced API cost by 78.1% and task time by 62.9% on historical-analysis reuse decisions the ordinary-tool comparator also answered correctly.",
      "content_text": "Adding nomue reduced API cost by 78.1% and task time by 62.9% on historical-analysis reuse decisions the ordinary-tool comparator also answered correctly.\n\nStatus: Preprint v1 — not peer reviewed; synthetic Welch tasks; aggregate reproduction supplement available\n\nPublished September 23, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/research/historical-analysis-reuse/",
      "date_published": "2026-09-23T20:04:09Z",
      "date_modified": "2026-09-23T20:04:09Z",
      "tags": [
        "Research",
        "Publication"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/jstat-noncentral-t-probability-collapse/",
      "url": "https://www.licklider.ai/engineering/jstat-noncentral-t-probability-collapse/",
      "title": "jStat returns zero for a probability near 44%",
      "summary": "Independent integration and a mathematical lower bound expose a noncentral-t probability collapse that remains after an earlier convergence safeguard.",
      "content_text": "Independent integration and a mathematical lower bound expose a noncentral-t probability collapse that remains after an earlier convergence safeguard.\n\nStatus: Reported in jStat issue #300 — upstream confirmation pending\n\nPublished September 23, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/jstat-noncentral-t-probability-collapse/",
      "date_published": "2026-09-23T16:44:12Z",
      "date_modified": "2026-09-23T16:44:12Z",
      "tags": [
        "Engineering",
        "Upstream report"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/statsmodels-contingency-table-tail-loss/",
      "url": "https://www.licklider.ai/engineering/statsmodels-contingency-table-tail-loss/",
      "title": "statsmodels returns zero for a representable chi-square tail",
      "summary": "A reported p-value collapse for a representable chi-square tail led to a merged statsmodels repair with regression coverage for related contingency-table paths.",
      "content_text": "A reported p-value collapse for a representable chi-square tail led to a merged statsmodels repair with regression coverage for related contingency-table paths.\n\nStatus: Fix merged into statsmodels main in PR #10276 — not yet released\n\nPublished September 22, 2026 (UTC). Updated September 23, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/statsmodels-contingency-table-tail-loss/",
      "date_published": "2026-09-22T12:57:15Z",
      "date_modified": "2026-09-23T11:59:24Z",
      "tags": [
        "Engineering",
        "Upstream report"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/preserving-welch-confidence-intervals-near-zero/",
      "url": "https://www.licklider.ai/engineering/preserving-welch-confidence-intervals-near-zero/",
      "title": "Preserving Welch confidence intervals near zero",
      "summary": "A public-source verifier repair retains precision before nearly equal quantities are subtracted, with independent reference checks around the boundary where the repair takes over.",
      "content_text": "A public-source verifier repair retains precision before nearly equal quantities are subtracted, with independent reference checks around the boundary where the repair takes over.\n\nStatus: Merged source repair; not in the published npm 0.2.1-rc.1; no Protocol scope or tolerance change\n\nPublished September 19, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/preserving-welch-confidence-intervals-near-zero/",
      "date_published": "2026-09-19T09:18:14Z",
      "date_modified": "2026-09-19T09:18:14Z",
      "tags": [
        "Engineering",
        "Implementation note"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/keeping-independent-checks-visible/",
      "url": "https://www.licklider.ai/engineering/keeping-independent-checks-visible/",
      "title": "Keeping independent checks visible when verification fails",
      "summary": "An experimental Holm verifier preserves eligible Record checks when the caller’s expected context differs, and explains why dependent checks cannot run.",
      "content_text": "An experimental Holm verifier preserves eligible Record checks when the caller’s expected context differs, and explains why dependent checks cannot run.\n\nStatus: Unissued candidate.5 development checkpoint merged; implementation and Research Gate remain open; no additional public support\n\nPublished September 19, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/keeping-independent-checks-visible/",
      "date_published": "2026-09-19T09:18:14Z",
      "date_modified": "2026-09-19T09:18:14Z",
      "tags": [
        "Engineering",
        "Implementation note"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/student-t-df1-center-precision/",
      "url": "https://www.licklider.ai/engineering/student-t-df1-center-precision/",
      "title": "Preserving Student-t probability near zero at one degree of freedom",
      "summary": "A closed-form calculation keeps small but representable probability differences from disappearing and gives the public nomue verifier an independent regression check for a known numerical failure.",
      "content_text": "A closed-form calculation keeps small but representable probability differences from disappearing and gives the public nomue verifier an independent regression check for a known numerical failure.\n\nStatus: Implemented in @licklider/nomue-verifier 0.2.1-rc.1; SciPy issue #25667 remains open; no Protocol scope change\n\nPublished September 18, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/student-t-df1-center-precision/",
      "date_published": "2026-09-18T03:16:02Z",
      "date_modified": "2026-09-18T03:16:02Z",
      "tags": [
        "Engineering",
        "Implementation note"
      ]
    },
    {
      "id": "https://www.licklider.ai/news/nomue-protocol-release-5-public-discussion/",
      "url": "https://www.licklider.ai/news/nomue-protocol-release-5-public-discussion/",
      "title": "nomue Protocol opens Release 5 discussion on shared evidence for study design and selection timing",
      "summary": "Researchers and developers can help define how declared study design and selection timing should appear in a common verification report across analysis families.",
      "content_text": "Researchers and developers can help define how declared study design and selection timing should appear in a common verification report across analysis families.\n\nStatus: Public discussion open — conditional family scope; research and design holds remain; no new supported capability\n\nPublished September 17, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/news/nomue-protocol-release-5-public-discussion/",
      "date_published": "2026-09-17T00:51:14Z",
      "date_modified": "2026-09-17T00:51:14Z",
      "tags": [
        "News",
        "Public consultation"
      ]
    },
    {
      "id": "https://www.licklider.ai/research/whole-submission-welch-verification/",
      "url": "https://www.licklider.ai/research/whole-submission-welch-verification/",
      "title": "Comparing three tool-assisted AI configurations for whole-submission verification of Welch reports — preprint v1.0",
      "summary": "A 648-session fixed-panel study compares general Python, a stable Welch wrapper, and a bound verifier for complete five-field report correctness.",
      "content_text": "A 648-session fixed-panel study compares general Python, a stable Welch wrapper, and a bound verifier for complete five-field report correctness.\n\nStatus: Preprint v1.0 — not peer reviewed; fixed constructed panel; reproducibility package available\n\nPublished September 16, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/research/whole-submission-welch-verification/",
      "date_published": "2026-09-16T05:35:06Z",
      "date_modified": "2026-09-16T05:35:06Z",
      "tags": [
        "Research",
        "Publication"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/making-verification-results-useful-to-agents/",
      "url": "https://www.licklider.ai/engineering/making-verification-results-useful-to-agents/",
      "title": "Making verification results more useful to research agents",
      "summary": "nomue development improves difficult calculations, makes completed checks explicit, and preserves the meaning of historical results as versions change.",
      "content_text": "nomue development improves difficult calculations, makes completed checks explicit, and preserves the meaning of historical results as versions change.\n\nStatus: Product development update — integrated candidates and internal milestones; no new hosted release or public verifier support\n\nPublished September 15, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/making-verification-results-useful-to-agents/",
      "date_published": "2026-09-15T09:01:29Z",
      "date_modified": "2026-09-15T09:01:29Z",
      "tags": [
        "Engineering",
        "Implementation note"
      ]
    },
    {
      "id": "https://www.licklider.ai/research/same-test-different-p/",
      "url": "https://www.licklider.ai/research/same-test-different-p/",
      "title": "Same Test, Different p — preprint v1.0",
      "summary": "An audit of R, SciPy, and Julia shows why matching statistical decisions can still hide differences in definitions, numerical precision, and valid probability values.",
      "content_text": "An audit of R, SciPy, and Julia shows why matching statistical decisions can still hide differences in definitions, numerical precision, and valid probability values.\n\nStatus: Preprint v1.0 — not peer reviewed; reproducibility data and code available\n\nPublished September 15, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/research/same-test-different-p/",
      "date_published": "2026-09-15T08:01:43Z",
      "date_modified": "2026-09-15T08:01:43Z",
      "tags": [
        "Research",
        "Publication"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/statsmodels-welch-sumsquares-overflow/",
      "url": "https://www.licklider.ai/engineering/statsmodels-welch-sumsquares-overflow/",
      "title": "statsmodels loses finite Welch results when an intermediate sum overflows",
      "summary": "Four exactly represented observations per group make statsmodels overflow an intermediate sum, losing finite variances and Welch t-test results.",
      "content_text": "Four exactly represented observations per group make statsmodels overflow an intermediate sum, losing finite variances and Welch t-test results.\n\nStatus: Fix merged into statsmodels main in PR #10255 — not yet released\n\nPublished September 14, 2026 (UTC). Updated September 30, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/statsmodels-welch-sumsquares-overflow/",
      "date_published": "2026-09-14T06:28:07Z",
      "date_modified": "2026-09-30T00:50:00Z",
      "tags": [
        "Engineering",
        "Upstream report"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/separating-record-conformance-from-verification/",
      "url": "https://www.licklider.ai/engineering/separating-record-conformance-from-verification/",
      "title": "Separating format checks from verification results",
      "summary": "An experimental Holm verifier separates format and declaration checks from integrity, context and arithmetic results, preserving explicit outcomes for checks that did not run.",
      "content_text": "An experimental Holm verifier separates format and declaration checks from integrity, context and arithmetic results, preserving explicit outcomes for checks that did not run.\n\nStatus: Historical unissued candidate.4; candidate.5 now implements revised dependencies; formal adoption and support remain open\n\nPublished September 13, 2026 (UTC). Updated September 19, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/separating-record-conformance-from-verification/",
      "date_published": "2026-09-13T12:56:31Z",
      "date_modified": "2026-09-19T09:18:14Z",
      "tags": [
        "Engineering",
        "Implementation note"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/checking-welch-results-with-exact-rescaling/",
      "url": "https://www.licklider.ai/engineering/checking-welch-results-with-exact-rescaling/",
      "title": "Checking Welch results with exact rescaling",
      "summary": "Exact inputs and independent references expose a changed Welch p-value despite a finite result and no warning in a SciPy boundary test.",
      "content_text": "Exact inputs and independent references expose a changed Welch p-value despite a finite result and no warning in a SciPy boundary test.\n\nStatus: NumPy float64 repair PR #26209 open; current head is mergeable with 56 / 56 checks passing; issue #26169 open; not merged or released\n\nPublished September 12, 2026 (UTC). Updated September 30, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/checking-welch-results-with-exact-rescaling/",
      "date_published": "2026-09-12T05:05:29Z",
      "date_modified": "2026-09-30T00:50:00Z",
      "tags": [
        "Engineering",
        "Upstream report"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/when-a-verification-call-must-discard-its-result/",
      "url": "https://www.licklider.ai/engineering/when-a-verification-call-must-discard-its-result/",
      "title": "When a verification call must discard its result",
      "summary": "An experimental Holm verifier connects Record checks to shared execution budgets, operating-system limits and cleanup evidence before deciding whether a result can be returned.",
      "content_text": "An experimental Holm verifier connects Record checks to shared execution budgets, operating-system limits and cleanup evidence before deciding whether a result can be returned.\n\nStatus: Historical execution candidate preserved; candidate.5 successor has a merged development checkpoint; no additional supported capability\n\nPublished September 11, 2026 (UTC). Updated September 19, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/when-a-verification-call-must-discard-its-result/",
      "date_published": "2026-09-11T21:21:55Z",
      "date_modified": "2026-09-19T09:18:14Z",
      "tags": [
        "Engineering",
        "Implementation note"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/binding-holm-corrections-to-comparisons/",
      "url": "https://www.licklider.ai/engineering/binding-holm-corrections-to-comparisons/",
      "title": "Binding Holm corrections to the intended comparisons",
      "summary": "An experiment checks exact Holm adjustments together with the expected declaration and supplied p-values, including changes that leave the displayed answer unchanged.",
      "content_text": "An experiment checks exact Holm adjustments together with the expected declaration and supplied p-values, including changes that leave the displayed answer unchanged.\n\nStatus: Original binding experiment preserved; successor includes scoped numerical review and deterministic sort repair; no additional supported capability\n\nPublished September 11, 2026 (UTC). Updated September 19, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/binding-holm-corrections-to-comparisons/",
      "date_published": "2026-09-11T07:57:49Z",
      "date_modified": "2026-09-19T09:18:14Z",
      "tags": [
        "Engineering",
        "Implementation note"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/checking-factorial-probability-evidence/",
      "url": "https://www.licklider.ai/engineering/checking-factorial-probability-evidence/",
      "title": "Checking factorial probability evidence against the raw observations",
      "summary": "The factorial candidate now joins Record checks, exact probability evidence, a complete report and controlled execution in an independently reviewed readiness package.",
      "content_text": "The factorial candidate now joins Record checks, exact probability evidence, a complete report and controlled execution in an independently reviewed readiness package.\n\nStatus: Independently reviewed, unissued Release 4 candidate; D01/D07 amendment discussion open; no Protocol support\n\nPublished September 11, 2026 (UTC). Updated September 19, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/checking-factorial-probability-evidence/",
      "date_published": "2026-09-11T07:57:49Z",
      "date_modified": "2026-09-19T09:18:14Z",
      "tags": [
        "Engineering",
        "Implementation note"
      ]
    },
    {
      "id": "https://www.licklider.ai/blog/before-asking-an-agent-to-compare-groups/",
      "url": "https://www.licklider.ai/blog/before-asking-an-agent-to-compare-groups/",
      "title": "What to decide before asking an agent to compare several groups",
      "summary": "Specify the comparisons, research conditions, and outputs you need so an agent can propose an analysis that answers your question.",
      "content_text": "Specify the comparisons, research conditions, and outputs you need so an agent can propose an analysis that answers your question.\n\nStatus: Illustrative planning example; not a tested agent workflow\n\nPublished September 11, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/blog/before-asking-an-agent-to-compare-groups/",
      "tags": [
        "Blog",
        "Practical guide"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/checking-factorial-statistics-and-tail-bounds/",
      "url": "https://www.licklider.ai/engineering/checking-factorial-statistics-and-tail-bounds/",
      "title": "Checking factorial statistics without trusting rounded intermediates",
      "summary": "Exact arithmetic and probability bounds offer a path beyond scaling repairs, while a review shows why matching rounded answers does not certify an interval.",
      "content_text": "Exact arithmetic and probability bounds offer a path beyond scaling repairs, while a review shows why matching rounded answers does not certify an interval.\n\nStatus: Reviewed research components now connected experimentally and archived; no additional supported capability\n\nPublished September 10, 2026 (UTC). Updated September 11, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/checking-factorial-statistics-and-tail-bounds/",
      "date_modified": "2026-09-11T07:57:49Z",
      "tags": [
        "Engineering",
        "Technical method"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/scipy-studentized-range-tail-loss/",
      "url": "https://www.licklider.ai/engineering/scipy-studentized-range-tail-loss/",
      "title": "A nonzero Studentized-range tail disappears in SciPy",
      "summary": "An exact special-case reference shows SciPy returning zero for a probability near 0.00000002, with no warning in the recorded runs.",
      "content_text": "An exact special-case reference shows SciPy returning zero for a probability near 0.00000002, with no warning in the recorded runs.\n\nStatus: Additional reproducer reported — upstream confirmation pending\n\nPublished September 10, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/scipy-studentized-range-tail-loss/",
      "tags": [
        "Engineering",
        "Upstream report"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/scipy-welch-anova-weight-sum-overflow/",
      "url": "https://www.licklider.ai/engineering/scipy-welch-anova-weight-sum-overflow/",
      "title": "Exact rescaling can reverse SciPy’s Welch ANOVA decision",
      "summary": "At an extreme input scale, SciPy’s Welch ANOVA changes a p-value from 0.02650 to 0.05611, crossing the 5% threshold without losing input information.",
      "content_text": "At an extreme input scale, SciPy’s Welch ANOVA changes a p-value from 0.02650 to 0.05611, crossing the 5% threshold without losing input information.\n\nStatus: NumPy float64 repair PR #26209 open; current head is mergeable with 56 / 56 checks passing; issue #26146 open; not merged or released\n\nPublished September 10, 2026 (UTC). Updated September 30, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/scipy-welch-anova-weight-sum-overflow/",
      "date_modified": "2026-09-30T00:50:00Z",
      "tags": [
        "Engineering",
        "Upstream report"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/agricolae-regw-treatment-labels/",
      "url": "https://www.licklider.ai/engineering/agricolae-regw-treatment-labels/",
      "title": "Renaming treatment groups changes agricolae’s REGW result",
      "summary": "With observations and group membership unchanged, renaming groups changes a p-value from 0.0363 to 0.0791 and reverses a 5% decision.",
      "content_text": "With observations and group membership unchanged, renaming groups changes a p-value from 0.0363 to 0.0791 and reverses a 5% decision.\n\nStatus: Reported by email — upstream confirmation pending\n\nPublished September 10, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/agricolae-regw-treatment-labels/",
      "tags": [
        "Engineering",
        "Upstream report"
      ]
    },
    {
      "id": "https://www.licklider.ai/news/nomue-protocol-release-3-public-discussion/",
      "url": "https://www.licklider.ai/news/nomue-protocol-release-3-public-discussion/",
      "title": "nomue Protocol opens Release 3 public discussion for independent groups and multiple comparisons",
      "summary": "Researchers and developers can comment on how a multi-group analysis should declare its design, comparisons, results, and error-control claims.",
      "content_text": "Researchers and developers can comment on how a multi-group analysis should declare its design, comparisons, results, and error-control claims.\n\nStatus: Public discussion open — supplied-source proposal; method adoption and numerical support remain pending\n\nPublished September 9, 2026 (UTC). Updated September 10, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/news/nomue-protocol-release-3-public-discussion/",
      "tags": [
        "News",
        "Public consultation"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/power-scaling-and-factorial-f-statistics/",
      "url": "https://www.licklider.ai/engineering/power-scaling-and-factorial-f-statistics/",
      "title": "What power-of-two scaling can and cannot repair",
      "summary": "Exact references show when rescaling recovers a factorial F calculation, and when lost inputs or rounding residuals require a different numerical decision.",
      "content_text": "Exact references show when rescaling recovers a factorial F calculation, and when lost inputs or rounding residuals require a different numerical decision.\n\nStatus: Reviewed and steward-accepted bounded research; no additional product support\n\nPublished September 9, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/power-scaling-and-factorial-f-statistics/",
      "tags": [
        "Engineering",
        "Technical method"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/games-howell-approximation-and-guarantees/",
      "url": "https://www.licklider.ai/engineering/games-howell-approximation-and-guarantees/",
      "title": "What “approximate” means for Games–Howell comparisons",
      "summary": "Original-paper checks separate the construction of unequal-variance comparisons from simulation evidence and a guaranteed bound on false positives.",
      "content_text": "Original-paper checks separate the construction of unequal-variance comparisons from simulation evidence and a guaranteed bound on false positives.\n\nStatus: Reviewed and steward-accepted bounded research; no additional product support\n\nPublished September 9, 2026 (UTC). Updated September 9, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/games-howell-approximation-and-guarantees/",
      "tags": [
        "Engineering",
        "Technical method"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/comparing-with-control-or-best/",
      "url": "https://www.licklider.ai/engineering/comparing-with-control-or-best/",
      "title": "Comparing with a control and comparing with the best answer different questions",
      "summary": "Source review separates fixed-control tests, step-up and step-down calibration, and intervals that compare each treatment with the best of the others.",
      "content_text": "Source review separates fixed-control tests, step-up and step-down calibration, and intervals that compare each treatment with the best of the others.\n\nStatus: Reviewed and steward-accepted bounded research; no additional product support\n\nPublished September 9, 2026 (UTC). Updated September 9, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/comparing-with-control-or-best/",
      "tags": [
        "Engineering",
        "Technical method"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/implementing-multiple-testing-graphs/",
      "url": "https://www.licklider.ai/engineering/implementing-multiple-testing-graphs/",
      "title": "What must stay fixed in a multiple-testing graph",
      "summary": "Closed testing and graphical procedures make error control inspectable, but order, weights, stopping rules, and zero-level behavior still need precise definitions.",
      "content_text": "Closed testing and graphical procedures make error control inspectable, but order, weights, stopping rules, and zero-level behavior still need precise definitions.\n\nStatus: Reviewed and steward-accepted bounded research; no additional product support\n\nPublished September 9, 2026 (UTC). Updated September 9, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/implementing-multiple-testing-graphs/",
      "tags": [
        "Engineering",
        "Technical method"
      ]
    },
    {
      "id": "https://www.licklider.ai/news/nomue-protocol-release-4-public-discussion/",
      "url": "https://www.licklider.ai/news/nomue-protocol-release-4-public-discussion/",
      "title": "nomue Protocol opens Release 4 public discussion for two-factor experiments",
      "summary": "Researchers and developers can comment on a proposal connecting assumptions, results, and verification evidence for balanced two-factor experiments.",
      "content_text": "Researchers and developers can comment on a proposal connecting assumptions, results, and verification evidence for balanced two-factor experiments.\n\nStatus: Public discussion open — reviewed unissued candidate; separate D01/D07 amendment window; no Protocol support\n\nPublished September 9, 2026 (UTC). Updated September 19, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/news/nomue-protocol-release-4-public-discussion/",
      "date_modified": "2026-09-19T09:18:14Z",
      "tags": [
        "News",
        "Public consultation"
      ]
    },
    {
      "id": "https://www.licklider.ai/news/nomue-welch-limited-release-1/",
      "url": "https://www.licklider.ai/news/nomue-welch-limited-release-1/",
      "title": "nomue launches limited Release 1 for agent-callable Welch verification",
      "summary": "nomue now gives approved recipients a way to call bounded Welch statistical verification from compatible AI agents and HTTP clients.",
      "content_text": "nomue now gives approved recipients a way to call bounded Welch statistical verification from compatible AI agents and HTTP clients.\n\nStatus: Limited Release 1 — approved recipients only; no public registration\n\nPublished September 9, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/news/nomue-welch-limited-release-1/",
      "tags": [
        "News",
        "Product"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/what-multiple-comparison-procedures-guarantee/",
      "url": "https://www.licklider.ai/engineering/what-multiple-comparison-procedures-guarantee/",
      "title": "What a multiple-comparison procedure actually guarantees",
      "summary": "Original-paper checks separated overall tests, individual comparisons, and simultaneous intervals, giving future verification rules a more precise statement of what they protect.",
      "content_text": "Original-paper checks separated overall tests, individual comparisons, and simultaneous intervals, giving future verification rules a more precise statement of what they protect.\n\nStatus: Source-reviewed Release 3 research; bounded SR-F source acceptance recorded\n\nPublished September 8, 2026 (UTC). Updated September 9, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/what-multiple-comparison-procedures-guarantee/",
      "tags": [
        "Engineering",
        "Technical method"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/floating-point-tiny-factorial-effects/",
      "url": "https://www.licklider.ai/engineering/floating-point-tiny-factorial-effects/",
      "title": "When floating-point calculations change a tiny factorial effect",
      "summary": "A 945-case comparison separated effects lost during input rounding from errors introduced by cell means and QR calculations, including cases where centering did not help.",
      "content_text": "A 945-case comparison separated effects lost during input rounding from errors introduced by cell means and QR calculations, including cases where centering did not help.\n\nStatus: Steward-accepted, independently reviewed bounded Release 4 numerical research\n\nPublished September 8, 2026 (UTC). Updated September 9, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/floating-point-tiny-factorial-effects/",
      "tags": [
        "Engineering",
        "Technical method"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/checking-multiple-testing-against-original-papers/",
      "url": "https://www.licklider.ai/engineering/checking-multiple-testing-against-original-papers/",
      "title": "Checking multiple-testing procedures against their original papers",
      "summary": "Reviewing six original papers clarified multiple-testing guarantees and exposed a numerical table entry that disagrees with its defining equation.",
      "content_text": "Reviewing six original papers clarified multiple-testing guarantees and exposed a numerical table entry that disagrees with its defining equation.\n\nStatus: Independently reviewed SR-C source evidence with bounded acceptance recorded; Release 3 preparation\n\nPublished September 7, 2026 (UTC). Updated September 9, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/checking-multiple-testing-against-original-papers/",
      "tags": [
        "Engineering",
        "Technical method"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/scipy-mannwhitneyu-batch-method-selection/",
      "url": "https://www.licklider.ai/engineering/scipy-mannwhitneyu-batch-method-selection/",
      "title": "SciPy’s automatic Mann–Whitney U test can change a result when tests are batched",
      "summary": "An unchanged sample pair crosses the 5% significance threshold when another pair contains repeated values, because SciPy selects one calculation method for the batch.",
      "content_text": "An unchanged sample pair crosses the 5% significance threshold when another pair contains repeated values, because SciPy selects one calculation method for the batch.\n\nStatus: Triaged by a SciPy maintainer into scipy.stats; implementation path confirmed — intended behavior and remedy awaiting decision\n\nPublished September 7, 2026 (UTC). Updated September 7, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/scipy-mannwhitneyu-batch-method-selection/",
      "tags": [
        "Engineering",
        "Upstream report"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/scipy-ttest-scale-range-loss/",
      "url": "https://www.licklider.ai/engineering/scipy-ttest-scale-range-loss/",
      "title": "SciPy t-tests can return p=0 or p=1 after exact rescaling",
      "summary": "SciPy’s one-sample and paired t-tests can reverse a 5% decision after exact power-of-two rescaling because an intermediate variance underflows or overflows.",
      "content_text": "SciPy’s one-sample and paired t-tests can reverse a 5% decision after exact power-of-two rescaling because an intermediate variance underflows or overflows.\n\nStatus: PR #24840 merged with the mparray high-precision backend; NumPy float64 repair PR #26135 remains open and unreviewed; issue #26113 open\n\nPublished September 7, 2026 (UTC). Updated September 13, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/scipy-ttest-scale-range-loss/",
      "date_modified": "2026-09-13T06:25:54Z",
      "tags": [
        "Engineering",
        "Upstream report"
      ]
    },
    {
      "id": "https://www.licklider.ai/research/humane-endpoints-and-the-unmeasured-outcome/",
      "url": "https://www.licklider.ai/research/humane-endpoints-and-the-unmeasured-outcome/",
      "title": "What three welfare guidance documents define, and what they leave to statistics",
      "summary": "Three animal-welfare guidance documents define exactly when a study must end and what is recorded, but none says whether the measurement it prevented still exists.",
      "content_text": "Three animal-welfare guidance documents define exactly when a study must end and what is recorded, but none says whether the measurement it prevented still exists.\n\nStatus: Source-bounded research finding — two independent close reviews; not peer reviewed; FND-1 gate open\n\nPublished September 4, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/research/humane-endpoints-and-the-unmeasured-outcome/",
      "tags": [
        "Research",
        "Research note"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/cataloguing-multi-group-procedures/",
      "url": "https://www.licklider.ai/engineering/cataloguing-multi-group-procedures/",
      "title": "Cataloguing every multi-group procedure before proposing any of them",
      "summary": "We catalogued 49 multi-group comparison procedures, gave each an explicit disposition, and found only seven backed by primary text we had actually read.",
      "content_text": "We catalogued 49 multi-group comparison procedures, gave each an explicit disposition, and found only seven backed by primary text we had actually read.\n\nStatus: Reviewed catalogue research; bounded Release 3 public discussion now open\n\nPublished September 4, 2026 (UTC). Updated September 9, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/cataloguing-multi-group-procedures/",
      "tags": [
        "Engineering",
        "Technical method"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/julia-signed-rank-pvalue-above-one/",
      "url": "https://www.licklider.ai/engineering/julia-signed-rank-pvalue-above-one/",
      "title": "A Julia signed-rank p-value above 1, now fixed in a release",
      "summary": "HypothesisTests.jl returned 1.25 for an exact two-sided signed-rank p-value. The matching correction shipped in v0.12.0 and remains in v0.12.2.",
      "content_text": "HypothesisTests.jl returned 1.25 for an exact two-sided signed-rank p-value. The matching correction shipped in v0.12.0 and remains in v0.12.2.\n\nStatus: Matching fix released in v0.12.0; present through v0.12.2; issue open\n\nPublished September 3, 2026 (UTC). Updated September 11, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/julia-signed-rank-pvalue-above-one/",
      "tags": [
        "Engineering",
        "Upstream report"
      ]
    },
    {
      "id": "https://www.licklider.ai/news/nomue-mcp-on-npm/",
      "url": "https://www.licklider.ai/news/nomue-mcp-on-npm/",
      "title": "nomue Record Verifier MCP release candidate is available",
      "summary": "The public @licklider/nomue-verifier-mcp package exposes a method-neutral Record verification tool over local stdio; the current release supports the Release 1 Welch bundle.",
      "content_text": "The public @licklider/nomue-verifier-mcp package exposes a method-neutral Record verification tool over local stdio; the current release supports the Release 1 Welch bundle.\n\nStatus: Public local MCP release candidate — npm rc 0.2.0-rc.1; official MCP Registry\n\nPublished September 2, 2026 (UTC). Updated September 18, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/news/nomue-mcp-on-npm/",
      "date_modified": "2026-09-18T03:16:02Z",
      "tags": [
        "News",
        "Product"
      ]
    },
    {
      "id": "https://www.licklider.ai/research/why-statistical-method-names-are-not-enough/",
      "url": "https://www.licklider.ai/research/why-statistical-method-names-are-not-enough/",
      "title": "Why a statistical method name is not enough",
      "summary": "Primary-source review showed why procedure names alone cannot identify statistical guarantees, assumptions, comparison families, or later variants.",
      "content_text": "Primary-source review showed why procedure names alone cannot identify statistical guarantees, assumptions, comparison families, or later variants.\n\nStatus: Independently checked research note — three source questions closed; wider source review continues; not peer reviewed\n\nPublished September 1, 2026 (UTC). Updated September 4, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/research/why-statistical-method-names-are-not-enough/",
      "tags": [
        "Research",
        "Research note"
      ]
    },
    {
      "id": "https://www.licklider.ai/news/nomue-verifier-on-npm/",
      "url": "https://www.licklider.ai/news/nomue-verifier-on-npm/",
      "title": "nomue verifier Release 1 candidate is available on npm",
      "summary": "The public @licklider/nomue-verifier package provides a direct local command for checking supported nomue Release 1 Records.",
      "content_text": "The public @licklider/nomue-verifier package provides a direct local command for checking supported nomue Release 1 Records.\n\nStatus: Public release candidate — npm rc 0.2.1-rc.1\n\nPublished September 1, 2026 (UTC). Updated September 18, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/news/nomue-verifier-on-npm/",
      "date_modified": "2026-09-18T03:16:02Z",
      "tags": [
        "News",
        "Product"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/bounding-paired-t-arithmetic-and-pvalues/",
      "url": "https://www.licklider.ai/engineering/bounding-paired-t-arithmetic-and-pvalues/",
      "title": "From numerical bounds to a controlled paired-t execution candidate",
      "summary": "The reviewed Release 2 formal decision packet now assembles the paired-t evidence and required decisions without adopting or issuing Protocol support.",
      "content_text": "The reviewed Release 2 formal decision packet now assembles the paired-t evidence and required decisions without adopting or issuing Protocol support.\n\nStatus: Independently reviewed Release 2 formal decision-readiness package — not adopted, issued, or supported\n\nPublished September 1, 2026 (UTC). Updated September 20, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/bounding-paired-t-arithmetic-and-pvalues/",
      "date_modified": "2026-09-20T09:45:33Z",
      "tags": [
        "Engineering",
        "Implementation note"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/r-wilcoxon-exact-pvalue-out-of-range/",
      "url": "https://www.licklider.ai/engineering/r-wilcoxon-exact-pvalue-out-of-range/",
      "title": "R’s exact Wilcoxon test can return p-values outside the valid range",
      "summary": "R’s exact Wilcoxon test returned negative p-values and a value above 1 on a zero-difference input; we reported it with three independent exact-arithmetic checks.",
      "content_text": "R’s exact Wilcoxon test returned negative p-values and a value above 1 on a zero-difference input; we reported it with three independent exact-arithmetic checks.\n\nStatus: Reported to R — PR#19144 open and unconfirmed\n\nPublished September 1, 2026 (UTC). Updated October 1, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/r-wilcoxon-exact-pvalue-out-of-range/",
      "date_modified": "2026-10-01T03:59:39Z",
      "tags": [
        "Engineering",
        "Upstream report"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/tracing-paired-t-from-observations-to-pvalue/",
      "url": "https://www.licklider.ai/engineering/tracing-paired-t-from-observations-to-pvalue/",
      "title": "Tracing a paired-t calculation from observations to p-value",
      "summary": "We connected paired observations to a p-value in one reviewed trace; later work closed its two numerical error ledgers and added a reviewed interval trace.",
      "content_text": "We connected paired observations to a p-value in one reviewed trace; later work closed its two numerical error ledgers and added a reviewed interval trace.\n\nStatus: Independently reviewed p-value and confidence-interval execution traces; interval proof continues\n\nPublished August 31, 2026 (UTC). Updated September 1, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/tracing-paired-t-from-observations-to-pvalue/",
      "tags": [
        "Engineering",
        "Implementation note"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/building-paired-t-tables-and-error-checks/",
      "url": "https://www.licklider.ai/engineering/building-paired-t-tables-and-error-checks/",
      "title": "Building paired-t numerical tables and input-specific error checks",
      "summary": "We built two reviewed 200-value tables and input-specific error checks; later decisions selected the p-value bound and one table for candidate interval work.",
      "content_text": "We built two reviewed 200-value tables and input-specific error checks; later decisions selected the p-value bound and one table for candidate interval work.\n\nStatus: Two independently reviewed 200-value tables and input-specific error checks — candidate Release 2 work\n\nPublished August 31, 2026 (UTC). Updated September 1, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/building-paired-t-tables-and-error-checks/",
      "tags": [
        "Engineering",
        "Implementation note"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/testing-paired-t-floating-point-boundaries/",
      "url": "https://www.licklider.ai/engineering/testing-paired-t-floating-point-boundaries/",
      "title": "Testing a paired-t evaluator at floating-point boundaries",
      "summary": "We built and independently reviewed a deterministic paired-t probability evaluator and boundary evidence while leaving accuracy bounds, supported inputs, and protocol registration open.",
      "content_text": "We built and independently reviewed a deterministic paired-t probability evaluator and boundary evidence while leaving accuracy bounds, supported inputs, and protocol registration open.\n\nStatus: Independently reviewed deterministic evaluator and floating-point boundary evidence — candidate Release 2 work\n\nPublished August 30, 2026 (UTC). Updated August 31, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/testing-paired-t-floating-point-boundaries/",
      "tags": [
        "Engineering",
        "Implementation note"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/certifying-paired-t-numerical-evidence/",
      "url": "https://www.licklider.ai/engineering/certifying-paired-t-numerical-evidence/",
      "title": "Certifying paired-t numerical evidence before protocol support",
      "summary": "We built and independently reviewed a proof pipeline for paired-t p-values and critical values before deciding what nomue Protocol will support.",
      "content_text": "We built and independently reviewed a proof pipeline for paired-t p-values and critical values before deciding what nomue Protocol will support.\n\nStatus: Independently reviewed proof pipeline for paired-t p-values and critical values — candidate Release 2 work\n\nPublished August 28, 2026 (UTC). Updated August 31, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/certifying-paired-t-numerical-evidence/",
      "tags": [
        "Engineering",
        "Technical method"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/scipy-wilcoxon-exact-pvalue/",
      "url": "https://www.licklider.ai/engineering/scipy-wilcoxon-exact-pvalue/",
      "title": "A SciPy exact Wilcoxon p-value error, fixed upstream",
      "summary": "SciPy’s exact Wilcoxon path could return zero for a positive p-value. SciPy diagnosed the cause and merged a fix the same day it was reported.",
      "content_text": "SciPy’s exact Wilcoxon path could return zero for a positive p-value. SciPy diagnosed the cause and merged a fix the same day it was reported.\n\nStatus: Fix merged in SciPy — awaiting a SciPy release\n\nPublished August 26, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/scipy-wilcoxon-exact-pvalue/",
      "tags": [
        "Engineering",
        "Upstream report"
      ]
    },
    {
      "id": "https://www.licklider.ai/news/nomue-protocol-release-1/",
      "url": "https://www.licklider.ai/news/nomue-protocol-release-1/",
      "title": "nomue Protocol Release 1 Public Draft is now available",
      "summary": "The first public draft of the nomue Protocol is available with local verification, public fixtures, and signed release artifacts.",
      "content_text": "The first public draft of the nomue Protocol is available with local verification, public fixtures, and signed release artifacts.\n\nStatus: Public Draft — experimental\n\nPublished August 24, 2026 (UTC). Updated September 18, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/news/nomue-protocol-release-1/",
      "date_modified": "2026-09-18T03:16:02Z",
      "tags": [
        "News",
        "Product"
      ]
    },
    {
      "id": "https://www.licklider.ai/research/correctly-rounded-or-refused/",
      "url": "https://www.licklider.ai/research/correctly-rounded-or-refused/",
      "title": "Correctly Rounded or Refused — preprint v0.2",
      "summary": "A preprint on checking the numerical accuracy of paired Student-t test results before software returns them.",
      "content_text": "A preprint on checking the numerical accuracy of paired Student-t test results before software returns them.\n\nStatus: Preprint v0.2 — not peer reviewed\n\nPublished August 23, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/research/correctly-rounded-or-refused/",
      "tags": [
        "Research",
        "Publication"
      ]
    },
    {
      "id": "https://www.licklider.ai/engineering/scipy-student-t-extreme-tail/",
      "url": "https://www.licklider.ai/engineering/scipy-student-t-extreme-tail/",
      "title": "An extreme-tail sign error in SciPy’s Student-t quantile",
      "summary": "A SciPy bug can return positive infinity instead of a large negative value for an extreme-tail Student-t quantile.",
      "content_text": "A SciPy bug can return positive infinity instead of a large negative value for an extreme-tail Student-t quantile.\n\nStatus: Fix merged in Boost.Math — awaiting a Boost release\n\nPublished August 23, 2026 (UTC). Updated September 9, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/engineering/scipy-student-t-extreme-tail/",
      "tags": [
        "Engineering",
        "Bug report"
      ]
    },
    {
      "id": "https://www.licklider.ai/news/nomue-verifier-release-1-candidate/",
      "url": "https://www.licklider.ai/news/nomue-verifier-release-1-candidate/",
      "title": "nomue verifier Release 1 candidate is now public",
      "summary": "The experimental local verifier is available for anyone to inspect and test and is now pinned to nomue Protocol Release 1.",
      "content_text": "The experimental local verifier is available for anyone to inspect and test and is now pinned to nomue Protocol Release 1.\n\nStatus: Experimental verifier — 0.2.1-rc.1; pinned to Protocol Release 1\n\nPublished August 16, 2026 (UTC). Updated September 18, 2026 (UTC).\n\nRead the full article: https://www.licklider.ai/news/nomue-verifier-release-1-candidate/",
      "date_modified": "2026-09-18T03:16:02Z",
      "tags": [
        "News",
        "Product preview"
      ]
    }
  ]
}