From 9814741b54c4333f7526390efb3f03e40c5de961 Mon Sep 17 00:00:00 2001
From: Lukas Wallrich <lukas.wallrich@gmail.com>
Date: Mon, 9 Nov 2026 17:53:17 +0000
Subject: [PATCH 1/5] Render italics and small caps in publication titles

Publication titles from Crossref carry <i>, <em> and <scp> markup and HTML
entities, which the cluster cards showed as literal text. A new partial,
clusters/rich_text.html, escapes the title and restores only those tags
(scp as small caps). featured-resources.js does the same for the reference
popup and the reading list, and copies and exports plain text. The popup
and reading-list APA references join the D15 italics exception for
reference lists.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---
 assets/css/doors/base.css                     |  5 ++-
 .../partials/clusters/cluster_section.html    |  4 +--
 layouts/partials/clusters/featured_cards.html |  4 +--
 layouts/partials/clusters/rich_text.html      |  7 ++++
 static/js/featured-resources.js               | 36 ++++++++++++++-----
 5 files changed, 43 insertions(+), 13 deletions(-)
 create mode 100644 layouts/partials/clusters/rich_text.html

diff --git a/assets/css/doors/base.css b/assets/css/doors/base.css
index 4defd74672b..36175115303 100644
--- a/assets/css/doors/base.css
+++ b/assets/css/doors/base.css
@@ -19,10 +19,13 @@ main:lang(zh){--display:'Noto Serif CJK SC','Songti SC','STSong',SimSun,serif;--
 *,*::before,*::after,.font-italic{font-style:normal!important}
 em,i:not([class]),dfn,cite,var{font-weight:600}
 .references-list :is(em,i),.references-page .article-style :is(em,i),
-:is(h2,h3,h4)[id^=references i]~:is(ul,ol,p) :is(em,i),.csl-bib-body :is(em,i){font-style:italic!important;font-weight:400}
+:is(h2,h3,h4)[id^=references i]~:is(ul,ol,p) :is(em,i),.csl-bib-body :is(em,i),
+.ref-apa :is(em,i),.rl-detail-apa :is(em,i){font-style:italic!important;font-weight:400}
 
 .wrap{width:min(1280px,calc(100% - 96px));margin-inline:auto}
 .visually-hidden{position:absolute;width:1px;height:1px;overflow:hidden;clip-path:inset(50%);white-space:nowrap}
+/* Small capitals in publication titles from Crossref (<scp>), see clusters/rich_text.html */
+.scp{font-variant:small-caps}
 .skip{position:fixed;top:-100px;inset-inline-start:12px;z-index:1100;background:var(--ink);color:#fff;padding:12px 16px}
 .skip:focus,.skip:hover{top:8px;color:#fff}
 .crumbs{font-size:15px;line-height:1.4;color:var(--ink)}
diff --git a/layouts/partials/clusters/cluster_section.html b/layouts/partials/clusters/cluster_section.html
index 8fae4c85ebc..232b5678be9 100644
--- a/layouts/partials/clusters/cluster_section.html
+++ b/layouts/partials/clusters/cluster_section.html
@@ -91,7 +91,7 @@ <h3 class="acc-label"><button type="button" class="acc-trigger" aria-expanded="f
                 {{ with $card.focus }}<span class="tag tag-focus">{{ . }}</span>{{ end }}
                 {{ with $card.resource_type }}<span class="tag tag-type">{{ . }}</span>{{ end }}
               </div>
-              <div class="fr-title">{{ $card.title }}</div>
+              <div class="fr-title">{{ partial "clusters/rich_text.html" $card.title }}</div>
               {{ with $card.summary }}<div class="fr-summary">{{ . }}</div>{{ end }}
               <div class="fr-actions">
                 <button class="fr-ref-trigger" data-doi="{{ .doi }}">
@@ -152,7 +152,7 @@ <h3 class="acc-label"><button type="button" class="acc-trigger" aria-expanded="f
                   {{ with $card.focus }}<span class="tag tag-focus">{{ . }}</span>{{ end }}
                   {{ with $card.resource_type }}<span class="tag tag-type">{{ . }}</span>{{ end }}
                 </div>
-                <div class="fc-title">{{ $card.title }}</div>
+                <div class="fc-title">{{ partial "clusters/rich_text.html" $card.title }}</div>
               </div>
               {{ with $card.summary }}
               <div class="fc-summary">{{ . }}</div>
diff --git a/layouts/partials/clusters/featured_cards.html b/layouts/partials/clusters/featured_cards.html
index 6abb75d26a6..6734505d63c 100644
--- a/layouts/partials/clusters/featured_cards.html
+++ b/layouts/partials/clusters/featured_cards.html
@@ -24,7 +24,7 @@
           {{ if .focus }}<span class="tag tag-focus">{{ .focus }}</span>{{ end }}
           {{ if .resource_type }}<span class="tag tag-type">{{ .resource_type }}</span>{{ end }}
         </div>
-        <div class="fr-title">{{ .title }}</div>
+        <div class="fr-title">{{ partial "clusters/rich_text.html" .title }}</div>
         <div class="fr-summary">{{ .summary }}</div>
         <div class="fr-actions">
           <button class="fr-ref-trigger" data-doi="{{ .doi }}">
@@ -94,7 +94,7 @@
             {{ if .focus }}<span class="tag tag-focus">{{ .focus }}</span>{{ end }}
             {{ if .resource_type }}<span class="tag tag-type">{{ .resource_type }}</span>{{ end }}
           </div>
-          <div class="fc-title">{{ .title }}</div>
+          <div class="fc-title">{{ partial "clusters/rich_text.html" .title }}</div>
         </div>
         {{ if .summary }}
         <div class="fc-summary">{{ .summary }}</div>
diff --git a/layouts/partials/clusters/rich_text.html b/layouts/partials/clusters/rich_text.html
new file mode 100644
index 00000000000..0c43e016597
--- /dev/null
+++ b/layouts/partials/clusters/rich_text.html
@@ -0,0 +1,7 @@
+{{- /* Publication titles from Crossref: plain text plus <i>, <em> and <scp> markup, sometimes
+       with HTML entities. Escape everything, then restore only those tags (scp as small caps).
+       static/js/featured-resources.js has the same rule (richText). */ -}}
+{{- $s := . | default "" | htmlUnescape | replaceRE `\s+` " " | strings.TrimSpace | htmlEscape -}}
+{{- $s = replaceRE `&lt;(?:i|em)&gt;(.*?)&lt;/(?:i|em)&gt;` "<i>$1</i>" $s -}}
+{{- $s = replaceRE `&lt;scp&gt;(.*?)&lt;/scp&gt;` `<span class="scp">$1</span>` $s -}}
+{{- return ($s | safeHTML) -}}
diff --git a/static/js/featured-resources.js b/static/js/featured-resources.js
index da29b2eb0a8..131f0cc400b 100644
--- a/static/js/featured-resources.js
+++ b/static/js/featured-resources.js
@@ -22,6 +22,26 @@
     return d.innerHTML;
   }
 
+  /** Decode HTML entities without parsing markup (a textarea's content is text only). */
+  function decodeEntities(s) {
+    var t = document.createElement('textarea');
+    t.innerHTML = s || '';
+    return t.value;
+  }
+
+  /** Titles and references from Crossref carry <i>, <em> and <scp>: escape, then restore those.
+      Same rule as layouts/partials/clusters/rich_text.html. */
+  function richText(s) {
+    return escHtml(decodeEntities(s).replace(/\s+/g, ' ').trim())
+      .replace(/&lt;(i|em)&gt;([\s\S]*?)&lt;\/\1&gt;/g, '<i>$2</i>')
+      .replace(/&lt;scp&gt;([\s\S]*?)&lt;\/scp&gt;/g, '<span class="scp">$1</span>');
+  }
+
+  /** The same text without markup, for copying and the PDF. */
+  function plainText(s) {
+    return decodeEntities(s).replace(/<\/?(?:i|em|scp)>/gi, '').replace(/\s+/g, ' ').trim();
+  }
+
   /** Flat lookup: DOI → resource object across all clusters. */
   function buildResourceIndex(data) {
     var idx = {};
@@ -85,13 +105,13 @@
       if (r.focus) html += '<span class="tag tag-focus">' + escHtml(r.focus) + '</span>';
       if (r.resource_type) html += '<span class="tag tag-type">' + escHtml(r.resource_type) + '</span>';
       html += '</div>';
-      html += '<div class="ref-popup-title">' + escHtml(r.title) + '</div>';
+      html += '<div class="ref-popup-title">' + richText(r.title) + '</div>';
 
       // APA
       if (r.apa) {
         html += '<div class="ref-section">';
         html += '<div class="ref-section-label">APA Reference</div>';
-        html += '<div class="ref-apa">' + escHtml(r.apa) + '</div>';
+        html += '<div class="ref-apa">' + richText(r.apa) + '</div>';
         html += '<div class="ref-copy-row">';
         html += '<button class="ref-copy-btn" data-copy="apa" data-doi="' + escHtml(doi) + '">' +
           '<svg viewBox="0 0 16 16" fill="none" stroke="currentColor" stroke-width="1.5"><rect x="3" y="3" width="8" height="10" rx="1"/><path d="M5 3V2a1 1 0 011-1h5a1 1 0 011 1v9a1 1 0 01-1 1h-1"/></svg>' +
@@ -154,7 +174,7 @@
         var d = copyBtn.getAttribute('data-doi');
         var res = resourceIdx[d];
         if (!res) return;
-        var text = type === 'bibtex' ? res.bibtex : res.apa;
+        var text = type === 'bibtex' ? res.bibtex : plainText(res.apa);
         navigator.clipboard.writeText(text).then(function () {
           copyBtn.classList.add('copied');
           var orig = copyBtn.innerHTML;
@@ -258,7 +278,7 @@
         html += '<svg viewBox="0 0 16 16" fill="none" stroke="currentColor" stroke-width="2"><path d="M4 6l4 4 4-4"/></svg>';
         html += '</button>';
         html += '<div class="rl-panel-item-info">';
-        html += '<div class="rl-panel-item-title">' + escHtml(r.title || doi) + '</div>';
+        html += '<div class="rl-panel-item-title">' + (r.title ? richText(r.title) : escHtml(doi)) + '</div>';
         html += '<div class="rl-panel-item-ref">' + escHtml(r.short_ref || '') + '</div>';
         html += '</div>';
         html += '<button class="rl-panel-item-remove" data-doi="' + escHtml(doi) + '" aria-label="Remove">&times;</button>';
@@ -268,7 +288,7 @@
         html += '<div class="rl-panel-item-details">';
         if (r.apa) {
           html += '<div class="rl-detail-ref-row">';
-          html += '<div class="rl-detail-apa">' + escHtml(r.apa) + '</div>';
+          html += '<div class="rl-detail-apa">' + richText(r.apa) + '</div>';
           html += '<button class="rl-detail-copy" data-copy-apa="' + escHtml(doi) + '" title="Copy APA reference">';
           html += '<svg viewBox="0 0 16 16" fill="none" stroke="currentColor" stroke-width="1.5"><rect x="3" y="3" width="8" height="10" rx="1"/><path d="M5 3V2a1 1 0 011-1h5a1 1 0 011 1v9a1 1 0 01-1 1h-1"/></svg>';
           html += '</button>';
@@ -365,7 +385,7 @@
         var d = copyBtn.getAttribute('data-copy-apa');
         var res = resourceIdx[d];
         if (res && res.apa) {
-          navigator.clipboard.writeText(res.apa).then(function () {
+          navigator.clipboard.writeText(plainText(res.apa)).then(function () {
             copyBtn.classList.add('copied');
             setTimeout(function () { copyBtn.classList.remove('copied'); }, 1500);
           });
@@ -467,7 +487,7 @@
           doc.setFont('helvetica', 'bold');
           doc.setFontSize(11);
           doc.setTextColor(34, 34, 34);
-          var titleLines = doc.splitTextToSize(r.title || doi, contentW);
+          var titleLines = doc.splitTextToSize(r.title ? plainText(r.title) : doi, contentW);
           doc.text(titleLines, margin, y);
           y += titleLines.length * 5 + 2;
 
@@ -477,7 +497,7 @@
             doc.setFont('times', 'normal');
             doc.setFontSize(9);
             doc.setTextColor(68, 68, 68);
-            var apaLines = doc.splitTextToSize(r.apa, contentW - 10);
+            var apaLines = doc.splitTextToSize(plainText(r.apa), contentW - 10);
             // Hanging indent: first line at margin, rest indented
             apaLines.forEach(function (line, li) {
               checkPage(4);

From c0191757f6f0e7b249a55291cdcdf8c40c11ddf5 Mon Sep 17 00:00:00 2001
From: Lukas Wallrich <lukas.wallrich@gmail.com>
Date: Mon, 9 Nov 2026 17:59:27 +0000
Subject: [PATCH 2/5] Show italics in publication titles and check cluster
 pages for literal markup

Card, popup and reading-list titles join the D15 exception for reference
lists, so journal names and species set in <i> are italic.
check_seo_html.py check 17 fails a build whose cluster pages show
&lt;i&gt;, &lt;em&gt; or &lt;scp&gt; as text or leave <scp> unrendered.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---
 assets/css/doors/base.css            |  2 ++
 scripts/check_seo_html.py            | 16 ++++++++++++++++
 scripts/tests/test_check_seo_html.py | 16 ++++++++++++++++
 3 files changed, 34 insertions(+)

diff --git a/assets/css/doors/base.css b/assets/css/doors/base.css
index 36175115303..cfbe2714faf 100644
--- a/assets/css/doors/base.css
+++ b/assets/css/doors/base.css
@@ -21,6 +21,8 @@ em,i:not([class]),dfn,cite,var{font-weight:600}
 .references-list :is(em,i),.references-page .article-style :is(em,i),
 :is(h2,h3,h4)[id^=references i]~:is(ul,ol,p) :is(em,i),.csl-bib-body :is(em,i),
 .ref-apa :is(em,i),.rl-detail-apa :is(em,i){font-style:italic!important;font-weight:400}
+/* Publication titles from those lists keep their journal names and species in italics, at the title's weight. */
+:is(.fr-title,.fc-title,.ref-popup-title,.rl-panel-item-title) i{font-style:italic!important;font-weight:inherit}
 
 .wrap{width:min(1280px,calc(100% - 96px));margin-inline:auto}
 .visually-hidden{position:absolute;width:1px;height:1px;overflow:hidden;clip-path:inset(50%);white-space:nowrap}
diff --git a/scripts/check_seo_html.py b/scripts/check_seo_html.py
index 41ce3f7d0cb..0d3ea51153c 100644
--- a/scripts/check_seo_html.py
+++ b/scripts/check_seo_html.py
@@ -309,6 +309,21 @@ def check_img_alt(pages):
             for p in pages if p.imgs_without_alt]
 
 
+LITERAL_TITLE_MARKUP = re.compile(r"&lt;/?(?:i|em|scp)&gt;|</?scp>", re.IGNORECASE)
+
+
+def check_cluster_title_markup(public_dir):
+    """Cluster pages render publication-title markup (clusters/rich_text.html), never show it."""
+    found = []
+    for root, _dirs, files in os.walk(os.path.join(public_dir, "clusters")):
+        if "index.html" in files:
+            full = os.path.join(root, "index.html")
+            with open(full, encoding="utf-8", errors="replace") as handle:
+                if LITERAL_TITLE_MARKUP.search(handle.read()):
+                    found.append(os.path.relpath(full, public_dir))
+    return sorted(found)
+
+
 def check_citation_targets(pages):
     return ["%s (%s)" % (p.path, href or "missing href")
             for p in pages for href in p.citations
@@ -564,6 +579,7 @@ def main(public_dir):
         "14. no inline <script> over 200 KB": check_inline_scripts(pages),
         "15. every citation has a local reference target": check_citation_targets(pages),
         "16. glossary links reach existing English terms": check_glossary_links(pages),
+        "17. cluster pages show no literal <i>/<em>/<scp> title markup": check_cluster_title_markup(public_dir),
     }
     conditional = check_conditional_assets(pages)
     results["3. conditional highlight.js / mermaid loading"] = [
diff --git a/scripts/tests/test_check_seo_html.py b/scripts/tests/test_check_seo_html.py
index 51fbbc6dfdd..c6bc02a5e09 100644
--- a/scripts/tests/test_check_seo_html.py
+++ b/scripts/tests/test_check_seo_html.py
@@ -217,5 +217,21 @@ def test_missing_non_english_external_and_empty_links_fail(self):
         self.assertEqual(len(self.check_links(links)), len(links))
 
 
+class ClusterTitleMarkupCheck(unittest.TestCase):
+    def check(self, body):
+        with tempfile.TemporaryDirectory() as public:
+            page = Path(public) / "clusters" / "cluster-1" / "index.html"
+            page.parent.mkdir(parents=True)
+            page.write_text("<div class=fc-title>%s</div>" % body, encoding="utf-8")
+            return seo.check_cluster_title_markup(public)
+
+    def test_rendered_markup_passes(self):
+        self.assertEqual(self.check("In <span class=scp>A</span>frica at <i>Cognition</i> (p &lt; .05)"), [])
+
+    def test_literal_or_unrendered_markup_fails(self):
+        for body in ("at &lt;i&gt;Cognition&lt;/i&gt;", "in <scp>A</scp>frica", "&lt;em&gt;AIMS&lt;/em&gt;"):
+            self.assertEqual(self.check(body), ["clusters/cluster-1/index.html"], body)
+
+
 if __name__ == "__main__":
     unittest.main()

From 38dcfaf53dcd6ddc6e4431f97504b8acc6cec108 Mon Sep 17 00:00:00 2001
From: Lukas Wallrich <lukas.wallrich@gmail.com>
Date: Mon, 9 Nov 2026 18:00:04 +0000
Subject: [PATCH 3/5] Match <i> and <em> closing tags separately in
 rich_text.html

RE2 has no backreferences, so <i> could close with </em>; the JS pairs them.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---
 layouts/partials/clusters/rich_text.html | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/layouts/partials/clusters/rich_text.html b/layouts/partials/clusters/rich_text.html
index 0c43e016597..84f8922fca2 100644
--- a/layouts/partials/clusters/rich_text.html
+++ b/layouts/partials/clusters/rich_text.html
@@ -2,6 +2,7 @@
        with HTML entities. Escape everything, then restore only those tags (scp as small caps).
        static/js/featured-resources.js has the same rule (richText). */ -}}
 {{- $s := . | default "" | htmlUnescape | replaceRE `\s+` " " | strings.TrimSpace | htmlEscape -}}
-{{- $s = replaceRE `&lt;(?:i|em)&gt;(.*?)&lt;/(?:i|em)&gt;` "<i>$1</i>" $s -}}
+{{- $s = replaceRE `&lt;i&gt;(.*?)&lt;/i&gt;` "<i>$1</i>" $s -}}
+{{- $s = replaceRE `&lt;em&gt;(.*?)&lt;/em&gt;` "<i>$1</i>" $s -}}
 {{- $s = replaceRE `&lt;scp&gt;(.*?)&lt;/scp&gt;` `<span class="scp">$1</span>` $s -}}
 {{- return ($s | safeHTML) -}}

From 6a81aa590ef37f849f19951f1e499379fc8a8f29 Mon Sep 17 00:00:00 2001
From: Lukas Wallrich <lukas.wallrich@gmail.com>
Date: Mon, 9 Nov 2026 18:13:19 +0000
Subject: [PATCH 4/5] Catch double-escaped title markup and a build with no
 cluster pages

Check 17 now also matches &amp;lt;i&amp;gt; and fails when it scans no cluster
page, as the homepage checks do when there is no homepage. The minimal fixture
build in the tests gains a cluster page.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---
 scripts/check_seo_html.py            | 11 ++++++++---
 scripts/tests/test_check_seo_html.py | 10 +++++++++-
 2 files changed, 17 insertions(+), 4 deletions(-)

diff --git a/scripts/check_seo_html.py b/scripts/check_seo_html.py
index 0d3ea51153c..81f812baaa2 100644
--- a/scripts/check_seo_html.py
+++ b/scripts/check_seo_html.py
@@ -309,18 +309,23 @@ def check_img_alt(pages):
             for p in pages if p.imgs_without_alt]
 
 
-LITERAL_TITLE_MARKUP = re.compile(r"&lt;/?(?:i|em|scp)&gt;|</?scp>", re.IGNORECASE)
+# Escaped once (&lt;i&gt;) or twice (&amp;lt;i&amp;gt;), or a raw <scp> no browser styles.
+LITERAL_TITLE_MARKUP = re.compile(r"&(?:amp;)?lt;/?(?:i|em|scp)&(?:amp;)?gt;|</?scp>", re.IGNORECASE)
 
 
 def check_cluster_title_markup(public_dir):
     """Cluster pages render publication-title markup (clusters/rich_text.html), never show it."""
-    found = []
-    for root, _dirs, files in os.walk(os.path.join(public_dir, "clusters")):
+    clusters_dir = os.path.join(public_dir, "clusters")
+    found, scanned = [], 0
+    for root, _dirs, files in os.walk(clusters_dir):
         if "index.html" in files:
+            scanned += 1
             full = os.path.join(root, "index.html")
             with open(full, encoding="utf-8", errors="replace") as handle:
                 if LITERAL_TITLE_MARKUP.search(handle.read()):
                     found.append(os.path.relpath(full, public_dir))
+    if not scanned:
+        return ["clusters/ (no cluster pages found)"]
     return sorted(found)
 
 
diff --git a/scripts/tests/test_check_seo_html.py b/scripts/tests/test_check_seo_html.py
index c6bc02a5e09..2320ed42f7a 100644
--- a/scripts/tests/test_check_seo_html.py
+++ b/scripts/tests/test_check_seo_html.py
@@ -68,6 +68,9 @@ def test_raw_url_report_does_not_fail_otherwise_valid_build(self):
                     "<meta property='og:image:height' content='630'>"
                     "<meta name='twitter:card' content='summary_large_image'>")
             (Path(public) / "index.html").write_text(html, encoding="utf-8")
+            cluster = Path(public) / "clusters" / "cluster-1" / "index.html"
+            cluster.parent.mkdir(parents=True)
+            cluster.write_text("<meta name='description' content='Cluster 1'><h1>Cluster 1</h1>", encoding="utf-8")
             output = StringIO()
             with redirect_stdout(output):
                 self.assertEqual(seo.main(public), 0)
@@ -229,9 +232,14 @@ def test_rendered_markup_passes(self):
         self.assertEqual(self.check("In <span class=scp>A</span>frica at <i>Cognition</i> (p &lt; .05)"), [])
 
     def test_literal_or_unrendered_markup_fails(self):
-        for body in ("at &lt;i&gt;Cognition&lt;/i&gt;", "in <scp>A</scp>frica", "&lt;em&gt;AIMS&lt;/em&gt;"):
+        for body in ("at &lt;i&gt;Cognition&lt;/i&gt;", "in <scp>A</scp>frica", "&lt;em&gt;AIMS&lt;/em&gt;",
+                     "at &amp;lt;i&amp;gt;Cognition&amp;lt;/i&amp;gt;"):
             self.assertEqual(self.check(body), ["clusters/cluster-1/index.html"], body)
 
+    def test_missing_cluster_pages_fail(self):
+        with tempfile.TemporaryDirectory() as public:
+            self.assertEqual(seo.check_cluster_title_markup(public), ["clusters/ (no cluster pages found)"])
+
 
 if __name__ == "__main__":
     unittest.main()

From 37ce6306e57bf4fa4cdb2898fc584560e3574cb1 Mon Sep 17 00:00:00 2001
From: Lukas Wallrich <lukas.wallrich@gmail.com>
Date: Wed, 11 Nov 2026 16:58:49 +0000
Subject: [PATCH 5/5] List checks 19 and 20 in the check_seo_html docstring

---
 scripts/check_seo_html.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/scripts/check_seo_html.py b/scripts/check_seo_html.py
index aef0e08e216..6bd68112836 100644
--- a/scripts/check_seo_html.py
+++ b/scripts/check_seo_html.py
@@ -40,7 +40,8 @@
   17. Every link in a visible breadcrumb trail (nav.crumbs) has a non-empty href,
      so a section that is not rendered cannot leave an empty link in the trail.
   18. No glossary page's <title> or <h1> shows Markdown emphasis as typed, e.g. "(*p*)".
-  19. Cluster pages render publication-title markup (<i>, <em>, <scp>) instead of
+  19. No page loads EmailOctopus or reCAPTCHA before a visitor subscribes.
+  20. Cluster pages render publication-title markup (<i>, <em>, <scp>) instead of
      showing it as literal or escaped text, and at least one cluster page exists.
 
 Report-only sections (never fail): pages with no <h1>, the