You are a constitutional council ranking individual git commits for ownership allocation.

Compare these two commits. Decide which contributed more lasting value to the project.

Judge substance, not spectacle:
- Prefer correct, lasting design and real bugfixes over churn, formatting, renames, or generated noise.
- Prefer clarity and necessity over sheer line count. A small precise change can beat a large diffuse one.
- Do not favor a side merely because its patch is longer or noisier.
- Weight what the change does for the project, not the contributor's name.

Return ONLY a JSON object: {"winner": "A" or "B", "ratio": "N:M", "explanation": "..."}
The explanation must cite concrete differences in the patches (1-3 sentences).

Side A — contributor: tommy-mor
Side A — commit message:
[5ca518f6] url refactor

Side A — unified diff (full patch):
diff --git a/Cargo.lock b/Cargo.lock
index 67a09a3b54f778fa7e857fdd589c3ed9c92e1322..ad7e4fe6d4ba2f2b033916194c1ef1ed873f1d46 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -1757,6 +1757,7 @@ name = "slug-types"
 version = "0.1.0"
 dependencies = [
  "serde",
+ "url",
 ]
 
 [[package]]
@@ -2272,6 +2273,7 @@ dependencies = [
  "idna",
  "percent-encoding",
  "serde",
+ "serde_derive",
 ]
 
 [[package]]
diff --git a/server/src/api/ui_html.rs b/server/src/api/ui_html.rs
index 606f7d6f97a4428efb90d1e0d544934c861fc4c5..cd3e0f0afd972d9ad9e7e4b92c5fa4c22bb8f620 100644
--- a/server/src/api/ui_html.rs
+++ b/server/src/api/ui_html.rs
@@ -270,10 +270,10 @@ fn post_redirect_location(room: &str, thread_tag: &str) -> String {
         format!("/t/{tag}")
     } else {
         let room = room.trim();
-        let Some((a, b)) = room.split_once('/') else {
+        let Some(seg) = slug_types::room_route_segment(room) else {
             return "/".to_string();
         };
-        format!("/r/{a}/{b}/t/{tag}")
+        format!("/r/{seg}/t/{tag}")
     }
 }
 
diff --git a/server/src/api/write_actor.rs b/server/src/api/write_actor.rs
index cb78d2f3f95b1bc163c1fb064d1d5f657e18000f..f9c3b8bd3fbf8fcb9c035e1a1572fef0b08fa8a9 100644
--- a/server/src/api/write_actor.rs
+++ b/server/src/api/write_actor.rs
@@ -19,13 +19,15 @@ use crate::{
 use super::auth::{issue_token_for_user, verify_token};
 use super::helpers::{now_ms, resolve_item};
 use super::validate::{normalize_room_and_thread, validate_ingest_document};
-use slug_types::RpcResult;
+use slug_types::{room_route_segment, RpcResult, ROOM_SHORT_ID_LEN};
 
 fn gen_short_id() -> String {
     use rand::Rng;
     const ALPHABET: &[u8] = b"0123456789abcdefghijklmnopqrstuvwxyz";
     let mut rng = rand::thread_rng();
-    (0..7).map(|_| ALPHABET[rng.gen_range(0..ALPHABET.len())] as char).collect()
+    (0..ROOM_SHORT_ID_LEN)
+        .map(|_| ALPHABET[rng.gen_range(0..ALPHABET.len())] as char)
+        .collect()
 }
 
 fn parse_capability(s: &str) -> Result<crate::events::ThreadCapability, String> {
@@ -55,8 +57,8 @@ async fn broadcast_web_refresh(state: &AppState, room_key: &str, thread_id: &str
     let feed_id = if room_key == "public" { "thread-feed" } else { "room-thread-feed" };
     let thread_url = if room_key == "public" {
         format!("/t/{thread_id}")
-    } else if let Some((short, slug)) = room_key.split_once('/') {
-        format!("/r/{short}/{slug}/t/{thread_id}")
+    } else if let Some(seg) = room_route_segment(room_key) {
+        format!("/r/{seg}/t/{thread_id}")
     } else {
         format!("/t/{thread_id}")
     };
@@ -78,8 +80,8 @@ async fn broadcast_web_refresh(state: &AppState, room_key: &str, thread_id: &str
     let js = builder.build();
     let mut path_prefixes = vec![if room_key == "public" {
         "/".to_string()
-    } else if let Some((short, slug)) = room_key.split_once('/') {
-        format!("/r/{short}/{slug}")
+    } else if let Some(seg) = room_route_segment(room_key) {
+        format!("/r/{seg}")
     } else {
         "/".to_string()
     }];
diff --git a/server/src/html/forum/nav.rs b/server/src/html/forum/nav.rs
index 0ee33d91160fc5542817b5e3e9ab4fee1d0e600f..48fe11e46731670874ff8b6b05baa6f09ae0b7e4 100644
--- a/server/src/html/forum/nav.rs
+++ b/server/src/html/forum/nav.rs
@@ -1,7 +1,8 @@
 use crate::canonical_path::canonicalize_item;
 use crate::reducer::ScopeId;
+use slug_types::room_route_segment;
 
-/// URL helpers for public `/t/…` and private room threads `/r/{short}/{slug}/t/…`.
+/// URL helpers for public `/t/…` and private room threads `/r/{short}{slug}/t/…`.
 #[derive(Clone)]
 pub struct ThreadNav {
     pub room_wire: String,
@@ -22,18 +23,15 @@ impl ThreadNav {
         }
     }
 
-    /// `room_id` wire form `shortid/slug`.
+    /// `room_id` wire form `shortid/slug` (HTTP uses [`slug_types::room_route_segment`]).
     pub(crate) fn from_room_id(room_id: &str) -> Option<Self> {
-        let (short, slug) = room_id.split_once('/')?;
-        if short.is_empty() || slug.is_empty() {
-            return None;
-        }
+        let room_seg = room_route_segment(room_id)?;
         Some(Self {
             room_wire: room_id.to_string(),
             scope: ScopeId::Room(room_id.to_string()),
-            room_path: format!("/r/{short}/{slug}"),
-            thread_path_prefix: format!("/r/{short}/{slug}/t"),
-            garden_path_prefix: format!("/r/{short}/{slug}/~"),
+            room_path: format!("/r/{room_seg}"),
+            thread_path_prefix: format!("/r/{room_seg}/t"),
+            garden_path_prefix: format!("/r/{room_seg}/~"),
         })
     }
 
diff --git a/server/src/html/forum/post_single.rs b/server/src/html/forum/post_single.rs
index c316f8f836df9d4ef9c05ebd9e54f699540e6d72..473747b3da3d4d7a54b5e0c63165d2533df643e1 100644
--- a/server/src/html/forum/post_single.rs
+++ b/server/src/html/forum/post_single.rs
@@ -93,12 +93,14 @@ pub async fn thread_post_view(
 
 pub async fn room_thread_post_view(
     State(state): State<AppState>,
-    Path((room_short, room_slug, tag, index_str)): Path<(String, String, String, String)>,
+    Path((room_key, tag, index_str)): Path<(String, String, String)>,
     headers: HeaderMap,
     jar: CookieJar,
     uri: Uri,
 ) -> impl IntoResponse {
-    let room_id = format!("{room_short}/{room_slug}");
+    let Some(room_id) = slug_types::room_id_from_route_segment(&room_key) else {
+        return (StatusCode::NOT_FOUND, "bad room path").into_response();
+    };
     let reduced = state.reduced.read().await;
     let user = optional_principal(&headers, &jar, &reduced);
     if !user_can_view_room(&reduced, &room_id, user.as_deref()) {
diff --git a/server/src/html/forum/views.rs b/server/src/html/forum/views.rs
index be5df1745ef580891a167c23c3dd6c06f804f295..1ec421f84335cbfe7f9db775b73a8ed24b197174 100644
--- a/server/src/html/forum/views.rs
+++ b/server/src/html/forum/views.rs
@@ -183,16 +183,18 @@ pub async fn thread_view(
     thread_view_inner(state, tag, q, ThreadNav::public(), headers, jar, uri).await
 }
 
-/// Room thread — `/r/:short/:slug/t/:tag`
+/// Room thread — `/r/:room_key/t/:tag` (`room_key` = `{short}{slug}`).
 pub async fn room_thread_view(
     State(state): State<AppState>,
-    Path((room_short, room_slug, tag)): Path<(String, String, String)>,
+    Path((room_key, tag)): Path<(String, String)>,
     Query(q): Query<ThreadViewQuery>,
     headers: HeaderMap,
     jar: CookieJar,
     uri: Uri,
 ) -> impl IntoResponse {
-    let room_id = format!("{room_short}/{room_slug}");
+    let Some(room_id) = slug_types::room_id_from_route_segment(&room_key) else {
+        return (StatusCode::NOT_FOUND, "bad room path").into_response();
+    };
     let reduced = state.reduced.read().await;
     let user = optional_principal(&headers, &jar, &reduced);
     if !user_can_view_room(&reduced, &room_id, user.as_deref()) {
@@ -226,15 +228,17 @@ pub(super) fn room_not_found_page(jar: &CookieJar, uri: &Uri) -> impl IntoRespon
     (StatusCode::NOT_FOUND, Html(page.into_string()))
 }
 
-/// Private room index — `/r/:short/:slug`
+/// Private room index — `/r/:room_key`
 pub async fn room_page(
     State(state): State<AppState>,
-    Path((room_short, room_slug)): Path<(String, String)>,
+    Path(room_key): Path<String>,
     headers: HeaderMap,
     jar: CookieJar,
     uri: Uri,
 ) -> impl IntoResponse {
-    let room_id = format!("{room_short}/{room_slug}");
+    let Some(room_id) = slug_types::room_id_from_route_segment(&room_key) else {
+        return (StatusCode::NOT_FOUND, "room not found").into_response();
+    };
     let now = now_ms();
     let reduced = state.reduced.read().await;
     if !reduced.rooms.contains(&room_id) {
@@ -266,7 +270,10 @@ pub async fn room_page(
     let audit_cli = format!("npx slugsocial private {room_id} audit");
     drop(reduced);
 
-    let slug_display = room_slug.as_str();
+    let slug_display = room_id
+        .split_once('/')
+        .map(|(_, slug)| slug)
+        .unwrap_or(room_id.as_str());
     let page = layout(
         &format!("room {slug_display} — slug.social"),
         "view-thread",
diff --git a/server/src/html/garden.rs b/server/src/html/garden.rs
index 423f23fd8c9ad7b7f454d6ea7a9a7607a4c9c5b9..e615dd356bcf634232d85610c0a26235ead125fd 100644
--- a/server/src/html/garden.rs
+++ b/server/src/html/garden.rs
@@ -309,12 +309,14 @@ pub async fn external_ontology_path(
 
 pub async fn room_garden_index(
     State(state): State<AppState>,
-    Path((room_short, room_slug)): Path<(String, String)>,
+    Path(room_key): Path<String>,
     headers: HeaderMap,
     jar: CookieJar,
     uri: Uri,
 ) -> impl IntoResponse {
-    let room_id = format!("{room_short}/{room_slug}");
+    let Some(room_id) = slug_types::room_id_from_route_segment(&room_key) else {
+        return (StatusCode::NOT_FOUND, "bad room path").into_response();
+    };
     let Some(nav) = ThreadNav::from_room_id(&room_id) else {
         return (StatusCode::NOT_FOUND, "bad room path").into_response();
     };
@@ -341,12 +343,14 @@ pub async fn room_garden_index(
 
 pub async fn room_external_garden_index(
     State(state): State<AppState>,
-    Path((room_short, room_slug)): Path<(String, String)>,
+    Path(room_key): Path<String>,
     headers: HeaderMap,
     jar: CookieJar,
     uri: Uri,
 ) -> impl IntoResponse {
-    let room_id = format!("{room_short}/{room_slug}");
+    let Some(room_id) = slug_types::room_id_from_route_segment(&room_key) else {
+        return (StatusCode::NOT_FOUND, "bad room path").into_response();
+    };
     let Some(nav) = ThreadNav::from_room_id(&room_id) else {
         return (StatusCode::NOT_FOUND, "bad room path").into_response();
     };
@@ -416,12 +420,14 @@ pub async fn room_external_garden_index(
 
 pub async fn room_external_ontology_path(
     State(state): State<AppState>,
-    Path((room_short, room_slug, path)): Path<(String, String, String)>,
+    Path((room_key, path)): Path<(String, String)>,
     headers: HeaderMap,
     jar: CookieJar,
     uri: Uri,
 ) -> impl IntoResponse {
-    let room_id = format!("{room_short}/{room_slug}");
+    let Some(room_id) = slug_types::room_id_from_route_segment(&room_key) else {
+        return (StatusCode::NOT_FOUND, "bad room path").into_response();
+    };
     let Some(nav) = ThreadNav::from_room_id(&room_id) else {
         return (StatusCode::NOT_FOUND, "bad room path").into_response();
     };
@@ -442,12 +448,14 @@ pub async fn room_external_ontology_path(
 
 pub async fn room_ontology_path(
     State(state): State<AppState>,
-    Path((room_short, room_slug, path)): Path<(String, String, String)>,
+    Path((room_key, path)): Path<(String, String)>,
     headers: HeaderMap,
     jar: CookieJar,
     uri: Uri,
 ) -> impl IntoResponse {
-    let room_id = format!("{room_short}/{room_slug}");
+    let Some(room_id) = slug_types::room_id_from_route_segment(&room_key) else {
+        return (StatusCode::NOT_FOUND, "bad room path").into_response();
+    };
     let Some(nav) = ThreadNav::from_room_id(&room_id) else {
         return (StatusCode::NOT_FOUND, "bad room path").into_response();
     };
diff --git a/server/src/html/search.rs b/server/src/html/search.rs
index 43e6ebf36cf0fe72c96f0f9d850bea51ac094c43..f01732f7edb32c68fcc10c39545c8f56476adb27 100644
--- a/server/src/html/search.rs
+++ b/server/src/html/search.rs
@@ -351,8 +351,8 @@ fn render_search_results(results: &SearchResults, query: &str) -> Markup {
                     ul class="search-posts" {
                         @for r in &results.posts {
                             @let (post_href, post_label) = if let Some((room, tag)) = r.thread.split_once("/#") {
-                                if let Some((short, slug)) = room.split_once('/') {
-                                    (format!("/r/{short}/{slug}/t/{tag}"), format!("{room}/#{tag}"))
+                                if let Some(seg) = slug_types::room_route_segment(room) {
+                                    (format!("/r/{seg}/t/{tag}"), format!("{room}/#{tag}"))
                                 } else {
                                     ("/".to_string(), r.thread.clone())
                                 }
diff --git a/server/src/lib.rs b/server/src/lib.rs
index 9e69c3f4a153527caa77b00ad101aca94b425995..71032f9a0b556097d90f3b403c677e0c86429ee4 100644
--- a/server/src/lib.rs
+++ b/server/src/lib.rs
@@ -79,30 +79,30 @@ pub fn create_app(state: AppState) -> Router {
         .route("/~/*path", get(crate::html::ontology_path))
         .route("/-", get(crate::html::external_garden_index))
         .route("/-/*path", get(crate::html::external_ontology_path))
-        .route("/r/:room_short/:room_slug/~", get(crate::html::room_garden_index))
+        .route("/r/:room_key/~", get(crate::html::room_garden_index))
         .route(
-            "/r/:room_short/:room_slug/~/*path",
+            "/r/:room_key/~/*path",
             get(crate::html::room_ontology_path),
         )
         .route(
-            "/r/:room_short/:room_slug/-",
+            "/r/:room_key/-",
             get(crate::html::room_external_garden_index),
         )
         .route(
-            "/r/:room_short/:room_slug/-/*path",
+            "/r/:room_key/-/*path",
             get(crate::html::room_external_ontology_path),
         )
         .route("/t/:tag/:index", get(crate::html::thread_post_view))
         .route("/t/:tag", get(crate::html::thread_view))
         .route(
-            "/r/:room_short/:room_slug/t/:thread_tag/:index",
+            "/r/:room_key/t/:thread_tag/:index",
             get(crate::html::room_thread_post_view),
         )
         .route(
-            "/r/:room_short/:room_slug/t/:thread_tag",
+            "/r/:room_key/t/:thread_tag",
             get(crate::html::room_thread_view),
         )
-        .route("/r/:room_short/:room_slug", get(crate::html::room_page))
+        .route("/r/:room_key", get(crate::html::room_page))
         .route("/join/:token", get(api::get_join_invite))
         .route("/auth/login", get(api::get_auth_login))
         .route("/auth/callback", get(api::get_auth_callback))
diff --git a/server/tests/integration.rs b/server/tests/integration.rs
index 515c691d04ef3f1f4bdfa8ee911360a43fd2b26e..feb73d475b3fc635d7f16c329d4600718c7472ba 100644
--- a/server/tests/integration.rs
+++ b/server/tests/integration.rs
@@ -1,4 +1,5 @@
 use sha2::{Digest, Sha256};
+use slug_types::room_route_segment;
 use slugsocial_server::{
     event_log::EventLog,
     events::{Event, TokenIssued, UserRegistered},
@@ -607,7 +608,7 @@ async fn test_private_room_thread_urls_use_t_segment() {
         .as_str()
         .unwrap()
         .to_string();
-    let (room_short, room_slug) = room_id.split_once('/').unwrap();
+    let room_seg = room_route_segment(&room_id).unwrap();
 
     let rpc = ui_post_ingest_rpc(&room_id, "main-thread", "private post via web");
     let post = client
@@ -626,7 +627,7 @@ async fn test_private_room_thread_urls_use_t_segment() {
         Some("text/javascript; charset=utf-8")
     );
     let post_js = post.text().await.unwrap();
-    let location = format!("/r/{room_short}/{room_slug}/t/main-thread");
+    let location = format!("/r/{room_seg}/t/main-thread");
     assert!(post_js.contains(&format!("window.location = {:?};", location)));
     assert!(post_js.contains("#room-thread-feed"));
     assert!(post_js.contains("#thread-feed-region"));
@@ -665,7 +666,7 @@ async fn test_private_room_post_links_use_private_garden_routes() {
         .as_str()
         .unwrap()
         .to_string();
-    let (room_short, room_slug) = room_id.split_once('/').unwrap();
+    let room_seg = room_route_segment(&room_id).unwrap();
 
     let rpc = ui_post_ingest_rpc(
         &room_id,
@@ -682,18 +683,18 @@ async fn test_private_room_post_links_use_private_garden_routes() {
     assert_eq!(post.status(), reqwest::StatusCode::OK);
 
     let thread_page = client
-        .get(format!("http://{addr}/r/{room_short}/{room_slug}/t/garden-thread"))
+        .get(format!("http://{addr}/r/{room_seg}/t/garden-thread"))
         .header("Authorization", format!("Bearer {bearer}"))
         .send()
         .await
         .unwrap();
     assert!(thread_page.status().is_success());
     let body = thread_page.text().await.unwrap();
-    assert!(body.contains(&format!("/r/{room_short}/{room_slug}/~/secret/item")));
+    assert!(body.contains(&format!("/r/{room_seg}/~/secret/item")));
     assert!(!body.contains("href=\"/~/secret/item\""));
 
     let garden_page = client
-        .get(format!("http://{addr}/r/{room_short}/{room_slug}/~/secret/item"))
+        .get(format!("http://{addr}/r/{room_seg}/~/secret/item"))
         .header("Authorization", format!("Bearer {bearer}"))
         .send()
         .await
@@ -701,7 +702,7 @@ async fn test_private_room_post_links_use_private_garden_routes() {
     assert!(garden_page.status().is_success());
     let garden_body = garden_page.text().await.unwrap();
     assert!(garden_body.contains("classified"));
-    assert!(garden_body.contains(&format!("/r/{room_short}/{room_slug}/t/garden-thread")));
+    assert!(garden_body.contains(&format!("/r/{room_seg}/t/garden-thread")));
 }
 
 #[tokio::test]
@@ -726,7 +727,7 @@ async fn test_private_room_garden_root_lists_top_level_tilde_children() {
         .as_str()
         .unwrap()
         .to_string();
-    let (room_short, room_slug) = room_id.split_once('/').unwrap();
+    let room_seg = room_route_segment(&room_id).unwrap();
 
     let rpc = ui_post_ingest_rpc(
         &room_id,
@@ -743,7 +744,7 @@ async fn test_private_room_garden_root_lists_top_level_tilde_children() {
     assert_eq!(post.status(), reqwest::StatusCode::OK);
 
     let root_page = client
-        .get(format!("http://{addr}/r/{room_short}/{room_slug}/~"))
+        .get(format!("http://{addr}/r/{room_seg}/~"))
         .header("Authorization", format!("Bearer {bearer}"))
         .send()
         .await
@@ -782,10 +783,10 @@ async fn test_empty_private_room_garden_returns_404() {
         .as_str()
         .unwrap()
         .to_string();
-    let (room_short, room_slug) = room_id.split_once('/').unwrap();
+    let room_seg = room_route_segment(&room_id).unwrap();
 
     let root = client
-        .get(format!("http://{addr}/r/{room_short}/{room_slug}/~"))
+        .get(format!("http://{addr}/r/{room_seg}/~"))
         .header("Authorization", format!("Bearer {bearer}"))
         .send()
         .await
@@ -896,8 +897,8 @@ async fn test_sse_stream_emits_evalable_js_after_post() {
         .as_str()
         .unwrap()
         .to_string();
-    let (room_short, room_slug) = room_id.split_once('/').unwrap();
-    let room_path = format!("/r/{room_short}/{room_slug}");
+    let room_seg = room_route_segment(&room_id).unwrap();
+    let room_path = format!("/r/{room_seg}");
 
     let sse_resp = client
         .get(format!("http://{addr}/sse?path={}", urlencoding::encode(&room_path)))
diff --git a/test/browser_room_delete.clj b/test/browser_room_delete.clj
index 833e3f8aa7f850b60ee59742736143cbf101a07d..764ecaa674798a61c1ffca3859f2a56a91a0e97e 100644
--- a/test/browser_room_delete.clj
+++ b/test/browser_room_delete.clj
@@ -54,7 +54,7 @@
            room-id (get-in create-json ["results" 0 "result" "RoomCreated" "room_id"])
            _ (is (string? room-id) "room id present")
            [room-short room-slug] (str/split room-id #"/" 2)
-           room-path (str "/r/" room-short "/" room-slug)]
+           room-path (str "/r/" room-short room-slug)]
        (core/with-playwright [pw]
          (core/with-browser [browser (core/launch-chromium pw {:headless true :channel "chrome"})]
            (core/with-context [ctx (core/new-context browser)]
diff --git a/test/browser_sse.clj b/test/browser_sse.clj
index 8bb064dbc2dc0feb6caaa67f520a461645958277..678399b79f78ff48b5e77342fa21e69992393834 100644
--- a/test/browser_sse.clj
+++ b/test/browser_sse.clj
@@ -77,7 +77,7 @@
                   (login-user! bob-pg base-url "bob")
 
                    (let [[room-short room-slug] (str/split room-id #"/" 2)
-                         room-url (str base-url "/r/" room-short "/" room-slug)
+                         room-url (str base-url "/r/" room-short room-slug)
                          thread-url (str room-url "/t/sse-thread")]
                     ;; Object under test: slug_ui.js intercepts POST /ui, evals JS, morphs
                     ;; #new-thread-ui-slot (expand compose), then post_ingest redirects to thread.
diff --git a/test/walkthrough_fixture.clj b/test/walkthrough_fixture.clj
index a487ff1ef8548580577914fb33f321586859c9af..f7eb1cacca0317129e9baf68fa4f4e33356ad70f 100644
--- a/test/walkthrough_fixture.clj
+++ b/test/walkthrough_fixture.clj
@@ -88,9 +88,9 @@
      :room {:id room-id
             :short room-short
             :slug room-slug
-            :url (str base-url "/r/" room-short "/" room-slug)
-            :thread_url (str base-url "/r/" room-short "/" room-slug "/t/walkthrough-thread")
-            :garden_url (str base-url "/r/" room-short "/" room-slug "/~/secret/item")}}))
+            :url (str base-url "/r/" room-short room-slug)
+            :thread_url (str base-url "/r/" room-short room-slug "/t/walkthrough-thread")
+            :garden_url (str base-url "/r/" room-short room-slug "/~/secret/item")}}))
 
 (defn- rebase-fixture-summary [saved current-base-url current-google-url current-data-dir]
   (let [inner (:summary saved)
@@ -103,9 +103,9 @@
            :data_dir (str current-data-dir)
            :summary (assoc inner
                            :room (assoc room
-                                         :url (str current-base-url "/r/" rs "/" lg)
-                                         :thread_url (str current-base-url "/r/" rs "/" lg "/t/walkthrough-thread")
-                                         :garden_url (str current-base-url "/r/" rs "/" lg "/~/secret/item"))))))
+                                         :url (str current-base-url "/r/" rs lg)
+                                         :thread_url (str current-base-url "/r/" rs lg "/t/walkthrough-thread")
+                                         :garden_url (str current-base-url "/r/" rs lg "/~/secret/item"))))))
 
 (defn- fixture-log-present? [data-dir]
   (let [p (fs/path data-dir "events.jsonl")]
diff --git a/types/Cargo.toml b/types/Cargo.toml
index 1ed06d483ac7c126516bce27f90e7639c2dfe370..5dc3becf238f23243313dff0484c682a7b225737 100644
--- a/types/Cargo.toml
+++ b/types/Cargo.toml
@@ -5,3 +5,4 @@ edition = "2021"
 
 [dependencies]
 serde = { version = "1.0", features = ["derive"] }
+url = { version = "2.5", features = ["serde"] }
diff --git a/types/src/lib.rs b/types/src/lib.rs
index fce1b5fa5e259a4b86a11cb4d4f7cc9bec3a0aa1..516cf935f15fca97081b39b988da5be894c67725 100644
--- a/types/src/lib.rs
+++ b/types/src/lib.rs
@@ -1,5 +1,7 @@
 use serde::{Deserialize, Serialize};
 
+pub mod room_route;
+pub mod url_normalize;
 pub mod paths;
 pub mod timeago;
 
@@ -8,6 +10,8 @@ pub use paths::{
     CanonicalItemUrl, ForumThreadUrl, GardenItemUrl, RelativePath, SLUG_TILDE_ONTOLOGY_ROOT,
     TildeHttpPathTail, TildeOntologyPath, TildePath, tilde_http_path_to_canonical,
 };
+pub use room_route::{room_id_from_route_segment, room_route_segment, ROOM_SHORT_ID_LEN};
+pub use url_normalize::normalize_http_identity_url;
 
 /// Max characters returned for a garden item body unless `full=true` / `--full` (API + CLI).
 pub const MAX_ITEM_BODY_PREVIEW_CHARS: usize = 100_000;
@@ -657,3 +661,6 @@ pub struct VoteResponse {
     pub ranking: Vec<RankRow>,
     pub next: NextMoves,
 }
+
+#[cfg(test)]
+mod url_identity_tests;
diff --git a/types/src/paths.rs b/types/src/paths.rs
index 8971f599a7ebcf399f60c306d35ee28d0f581194..98787a5fb481a1599556564bbbbd1d54dc693fc3 100644
--- a/types/src/paths.rs
+++ b/types/src/paths.rs
@@ -5,7 +5,7 @@
 //!
 //! - **[`canonicalize_item`] / [`CanonicalItemUrl`]** — graph storage key and DSL form; tilde
 //!   ontology root is always [`SLUG_TILDE_ONTOLOGY_ROOT`] (no `…/~/` trailing slash only).
-//! - **[`TildeHttpPathTail`]** — capture from `GET /~/*path` or `…/r/…/~/…` (the `*path` segment).
+//! - **[`TildeHttpPathTail`]** — capture from `GET /~/*path` or `…/r/{short}{slug}/~/…` (the `*path` segment).
 //! - **`-/…` wire form** — external items; see [`canonicalize_item`] dash branch.
 //! - **[`GardenItemUrl`], [`ForumThreadUrl`]** — JSON / browser href surfaces.
 
@@ -15,6 +15,9 @@ use std::ops::Deref;
 
 use serde::{Deserialize, Serialize};
 
+use crate::room_route::room_route_segment;
+use crate::url_normalize::{host_preserves_dash_path_case, normalize_http_identity_url};
+
 // ---------------------------------------------------------------------------
 // Slug tilde ontology (single storage form for `~/`)
 // ---------------------------------------------------------------------------
@@ -41,6 +44,29 @@ pub fn canonicalize_tag(input: &str) -> String {
     input.trim().trim_start_matches('#').to_lowercase()
 }
 
+fn finalize_external_identity_url(s: String) -> String {
+    if s.starts_with("https://slug.social/") {
+        return s;
+    }
+    let normalized = normalize_http_identity_url(&s).unwrap_or_else(|| s.clone());
+    strip_redundant_root_slash(&normalized).unwrap_or(normalized)
+}
+
+/// `url::Url` serializes bare hosts with a `/` path; we keep host-only items slash-free for stable
+/// keys matching the pre-normalizer spellings.
+fn strip_redundant_root_slash(s: &str) -> Option<String> {
+    let u = url::Url::parse(s).ok()?;
+    if u.path() == "/" && u.query().is_none() && u.fragment().is_none() {
+        let scheme = u.scheme();
+        let host = u.host_str()?;
+        return Some(match u.port() {
+            Some(p) => format!("{scheme}://{host}:{p}"),
+            None => format!("{scheme}://{host}"),
+        });
+    }
+    None
+}
+
 /// Ontology item reference → canonical absolute URL on the slug host.
 pub fn canonicalize_item(input: &str) -> String {
     let s = input.trim();
@@ -57,8 +83,9 @@ pub fn canonicalize_item(input: &str) -> String {
         if host.is_empty() {
             return String::new();
         }
+        let preserve_case = host_preserves_dash_path_case(&host);
         return if tail.is_empty() {
-            format!("https://{}", host)
+            finalize_external_identity_url(format!("https://{}", host))
         } else {
             let path = tail
                 .trim_start_matches('/')
@@ -68,33 +95,35 @@ pub fn canonicalize_item(input: &str) -> String {
                     let t = seg.trim();
                     if t.is_empty() {
                         None
+                    } else if preserve_case {
+                        Some(t.to_string())
                     } else {
                         Some(t.to_lowercase())
                     }
                 })
                 .collect::<Vec<_>>()
                 .join("/");
-            format!("https://{}/{}", host, path)
+            finalize_external_identity_url(format!("https://{}/{}", host, path))
         };
     }
 
     if let Some(rest) = s.strip_prefix("https://") {
         let (host, tail) = rest.split_once('/').map_or((rest, ""), |(h, t)| (h, t));
         let host = host.trim().to_lowercase();
-        if tail.is_empty() {
-            return format!("https://{}", host);
+        return finalize_external_identity_url(if tail.is_empty() {
+            format!("https://{}", host)
         } else {
-            return format!("https://{}/{}", host, tail);
-        }
+            format!("https://{}/{}", host, tail)
+        });
     }
     if let Some(rest) = s.strip_prefix("http://") {
         let (host, tail) = rest.split_once('/').map_or((rest, ""), |(h, t)| (h, t));
         let host = host.trim().to_lowercase();
-        if tail.is_empty() {
-            return format!("http://{}", host);
+        return finalize_external_identity_url(if tail.is_empty() {
+            format!("http://{}", host)
         } else {
-            return format!("http://{}/{}", host, tail);
-        }
+            format!("http://{}/{}", host, tail)
+        });
     }
 
     let is_tilde = s.starts_with("~/");
@@ -319,7 +348,7 @@ impl CanonicalItemUrl {
     }
 }
 
-/// HTTP route capture: path segment after `~/` in `GET /~/*path` or `…/r/…/~/…` (empty = ontology root).
+/// HTTP route capture: path segment after `~/` in `GET /~/*path` or `…/r/{short}{slug}/~/…` (empty = ontology root).
 #[derive(Debug, Clone, PartialEq, Eq, Hash)]
 pub struct TildeHttpPathTail(pub String);
 
@@ -506,12 +535,9 @@ fn garden_href_string(item: &str, room_wire: &str) -> String {
     if room.is_empty() || room == "public" {
         return api_path_or_url(item);
     }
-    let Some((short, slug)) = room.split_once('/') else {
+    let Some(room_seg) = room_route_segment(room) else {
         return api_path_or_url(item);
     };
-    if short.is_empty() || slug.is_empty() {
-        return api_path_or_url(item);
-    }
     let Some(c) = CanonicalItemUrl::parse(item) else {
         return api_path_or_url(item);
     };
@@ -520,19 +546,19 @@ fn garden_href_string(item: &str, room_wire: &str) -> String {
     let root_norm = root.as_str().trim_end_matches('/');
     if let Some(tail) = c.tilde_tail() {
         return if tail.is_empty() {
-            format!("https://slug.social/r/{short}/{slug}/~")
+            format!("https://slug.social/r/{room_seg}/~")
         } else {
-            format!("https://slug.social/r/{short}/{slug}/~/{}", tail)
+            format!("https://slug.social/r/{room_seg}/~/{}", tail)
         };
     }
     if item_norm == root_norm {
-        return format!("https://slug.social/r/{short}/{slug}/~");
+        return format!("https://slug.social/r/{room_seg}/~");
     }
     // External http(s) items use the same `/-/…` namespace under the room garden.
     if c.as_str().starts_with("https://") || c.as_str().starts_with("http://") {
         let tail = c.display_path();
         let tail = tail.strip_prefix("-/").unwrap_or(tail.as_str());
-        return format!("https://slug.social/r/{short}/{slug}/-/{tail}");
+        return format!("https://slug.social/r/{room_seg}/-/{tail}");
     }
     api_path_or_url(item)
 }
@@ -556,12 +582,8 @@ impl ForumThreadUrl {
         let tag = thread_tag.trim().trim_start_matches('#');
         Self(if room.is_empty() || room == "public" {
             format!("https://slug.social/t/{tag}")
-        } else if let Some((short, slug)) = room.split_once('/') {
-            if short.is_empty() || slug.is_empty() {
-                format!("https://slug.social/t/{tag}")
-            } else {
-                format!("https://slug.social/r/{short}/{slug}/t/{tag}")
-            }
+        } else if let Some(room_seg) = room_route_segment(room) {
+            format!("https://slug.social/r/{room_seg}/t/{tag}")
         } else {
             format!("https://slug.social/t/{tag}")
         })
@@ -706,7 +728,7 @@ mod tests {
     fn garden_private_room_prefixes_ontology() {
         assert_eq!(
             GardenItemUrl::from_storage_str("https://slug.social/~/topic/x", "9ab12cd/my-room").as_str(),
-            "https://slug.social/r/9ab12cd/my-room/~/topic/x"
+            "https://slug.social/r/9ab12cdmy-room/~/topic/x"
         );
     }
 
@@ -714,11 +736,11 @@ mod tests {
     fn garden_private_room_ontology_root() {
         assert_eq!(
             GardenItemUrl::from_storage_str("https://slug.social/~", "9ab12cd/my-room").as_str(),
-            "https://slug.social/r/9ab12cd/my-room/~"
+            "https://slug.social/r/9ab12cdmy-room/~"
         );
         assert_eq!(
             GardenItemUrl::from_storage_str("https://slug.social/~/", "9ab12cd/my-room").as_str(),
-            "https://slug.social/r/9ab12cd/my-room/~"
+            "https://slug.social/r/9ab12cdmy-room/~"
         );
     }
 
@@ -727,7 +749,7 @@ mod tests {
         let u = "https://example.com/z";
         assert_eq!(
             GardenItemUrl::from_storage_str(u, "9ab12cd/my-room").as_str(),
-            "https://slug.social/r/9ab12cd/my-room/-/example.com/z"
+            "https://slug.social/r/9ab12cdmy-room/-/example.com/z"
         );
     }
 
@@ -739,7 +761,19 @@ mod tests {
         );
         assert_eq!(
             ForumThreadUrl::from_room_tag("9ab12cd/my-room", "#debate").as_str(),
-            "https://slug.social/r/9ab12cd/my-room/t/debate"
+            "https://slug.social/r/9ab12cdmy-room/t/debate"
+        );
+    }
+
+    #[test]
+    fn canonicalize_youtube_short_links() {
+        assert_eq!(
+            canonicalize_item("https://youtu.be/dQw4w9WgXcQ"),
+            "https://www.youtube.com/watch?v=dQw4w9WgXcQ"
+        );
+        assert_eq!(
+            canonicalize_item("-/youtu.be/dQw4w9WgXcQ"),
+            "https://www.youtube.com/watch?v=dQw4w9WgXcQ"
         );
     }
 
diff --git a/types/src/room_route.rs b/types/src/room_route.rs
new file mode 100644
index 0000000000000000000000000000000000000000..4f4780c88e30f2b28e2cfcd7aee713d39dbd1c66
--- /dev/null
+++ b/types/src/room_route.rs
@@ -0,0 +1,56 @@
+//! HTTP path encoding for private rooms: `/r/{short}{slug}` (short is fixed width).
+
+/// Byte length of the random `short` segment in `short/slug` room ids.
+/// Must match room creation (`gen_short_id`) and [`super::paths`][] URL builders.
+pub const ROOM_SHORT_ID_LEN: usize = 7;
+
+/// `ab12cde/my-room` → `ab12cdemy-room` for a single `/r/…` path segment.
+pub fn room_route_segment(room_id: &str) -> Option<String> {
+    let (short, slug) = room_id.split_once('/')?;
+    if short.len() != ROOM_SHORT_ID_LEN || short.is_empty() || slug.is_empty() {
+        return None;
+    }
+    if !short
+        .bytes()
+        .all(|b| matches!(b, b'0'..=b'9' | b'a'..=b'z'))
+    {
+        return None;
+    }
+    Some(format!("{short}{slug}"))
+}
+
+/// `/r/{short}{slug}` path segment → `short/slug` wire id (inverse of [`room_route_segment`]).
+pub fn room_id_from_route_segment(seg: &str) -> Option<String> {
+    if seg.len() <= ROOM_SHORT_ID_LEN {
+        return None;
+    }
+    let (short, slug) = seg.split_at(ROOM_SHORT_ID_LEN);
+    if short.is_empty() || slug.is_empty() {
+        return None;
+    }
+    if !short
+        .bytes()
+        .all(|b| matches!(b, b'0'..=b'9' | b'a'..=b'z'))
+    {
+        return None;
+    }
+    Some(format!("{short}/{slug}"))
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn round_trip_room_segment() {
+        let id = "9ab12cd/my-room";
+        let seg = room_route_segment(id).unwrap();
+        assert_eq!(seg, "9ab12cdmy-room");
+        assert_eq!(room_id_from_route_segment(&seg).as_deref(), Some(id));
+    }
+
+    #[test]
+    fn too_short_segment_rejected() {
+        assert!(room_id_from_route_segment("9ab12cd").is_none());
+    }
+}
diff --git a/types/src/url_identity_tests.rs b/types/src/url_identity_tests.rs
new file mode 100644
index 0000000000000000000000000000000000000000..e2ff0a9a64a88049e31f20979c703777df6a9b65
--- /dev/null
+++ b/types/src/url_identity_tests.rs
@@ -0,0 +1,126 @@
+//! How `url::Url` behaves as `HashMap` keys (`Eq` + `Hash`).
+//!
+//! If `ItemId::External` stores `Url`, these tests are the contract you are buying into
+//! (or the baseline before you add a custom normalization layer).
+
+use std::collections::HashMap;
+use std::hash::{Hash, Hasher};
+use url::Url;
+
+fn hash_one(url: &Url) -> u64 {
+    let mut h = std::collections::hash_map::DefaultHasher::new();
+    url.hash(&mut h);
+    h.finish()
+}
+
+#[test]
+fn identical_parse_strings_are_eq_and_share_hash_bucket() {
+    let a = Url::parse("https://example.com/path").unwrap();
+    let b = Url::parse("https://example.com/path").unwrap();
+    assert_eq!(a, b);
+    assert_eq!(hash_one(&a), hash_one(&b));
+
+    let mut m: HashMap<Url, u32> = HashMap::new();
+    m.insert(a, 1);
+    *m.entry(b).or_default() += 10;
+    assert_eq!(m.len(), 1);
+    assert_eq!(m[&Url::parse("https://example.com/path").unwrap()], 11);
+}
+
+#[test]
+fn host_is_ascii_lowercase_in_eq() {
+    let lower = Url::parse("https://examplE.com/").unwrap();
+    let upper = Url::parse("https://EXAMPLE.com/").unwrap();
+    assert_eq!(lower, upper);
+    assert_eq!(hash_one(&lower), hash_one(&upper));
+}
+
+#[test]
+fn path_space_normalizes_to_percent_encoding_so_forms_merge() {
+    let encoded = Url::parse("https://example.com/a%20b").unwrap();
+    let decoded = Url::parse("https://example.com/a b").unwrap();
+    // Parser normalizes both to the same internal path (`/a%20b`).
+    assert_eq!(encoded, decoded);
+    assert_eq!(hash_one(&encoded), hash_one(&decoded));
+
+    let mut m: HashMap<Url, &str> = HashMap::new();
+    m.insert(encoded, "first");
+    assert_eq!(m.insert(decoded, "second"), Some("first"));
+    assert_eq!(m.len(), 1);
+    assert_eq!(m.values().next().copied(), Some("second"));
+}
+
+#[test]
+fn encoded_slash_in_segment_stays_distinct_from_real_path_separator() {
+    let encoded = Url::parse("https://example.com/a%2Fb").unwrap();
+    let real_slash = Url::parse("https://example.com/a/b").unwrap();
+    assert_ne!(encoded, real_slash);
+    assert_ne!(hash_one(&encoded), hash_one(&real_slash));
+}
+
+#[test]
+fn trailing_slash_on_path_is_significant_for_eq() {
+    let with_slash = Url::parse("https://example.com/foo/").unwrap();
+    let no_slash = Url::parse("https://example.com/foo").unwrap();
+    assert_ne!(with_slash, no_slash);
+    assert_ne!(hash_one(&with_slash), hash_one(&no_slash));
+}
+
+#[test]
+fn default_http_port_80_is_normalized_in_representation() {
+    let explicit = Url::parse("http://example.com:80/").unwrap();
+    let implicit = Url::parse("http://example.com/").unwrap();
+    assert_eq!(explicit, implicit);
+    assert_eq!(hash_one(&explicit), hash_one(&implicit));
+}
+
+#[test]
+fn default_https_port_443_is_normalized() {
+    let explicit = Url::parse("https://example.com:443/foo").unwrap();
+    let implicit = Url::parse("https://example.com/foo").unwrap();
+    assert_eq!(explicit, implicit);
+}
+
+#[test]
+fn non_default_port_is_part_of_identity() {
+    let a = Url::parse("https://example.com:444/").unwrap();
+    let b = Url::parse("https://example.com:445/").unwrap();
+    assert_ne!(a, b);
+}
+
+#[test]
+fn empty_path_vs_slash_only_path_may_differ() {
+    let root = Url::parse("https://example.com").unwrap();
+    let slash = Url::parse("https://example.com/").unwrap();
+    // Both serialize to `https://example.com/` in practice for this crate — verify.
+    assert_eq!(root, slash, "document: root and trailing-slash-only merge for this parser");
+}
+
+#[test]
+fn scheme_case_is_normalized_to_lowercase() {
+    let lower = Url::parse("https://example.com/").unwrap();
+    let upper = Url::parse("HTTPS://example.com/").unwrap();
+    assert_eq!(lower, upper);
+}
+
+#[test]
+fn fragment_is_part_of_eq_and_hash() {
+    let no_frag = Url::parse("https://example.com/a").unwrap();
+    let frag = Url::parse("https://example.com/a#section").unwrap();
+    assert_ne!(
+        no_frag, frag,
+        "#fragment is included in PartialEq — anchors are different HashMap keys"
+    );
+    assert_ne!(hash_one(&no_frag), hash_one(&frag));
+}
+
+#[test]
+fn query_order_and_encoding_can_split_identity() {
+    let a = Url::parse("https://example.com/?b=2&a=1").unwrap();
+    let b = Url::parse("https://example.com/?a=1&b=2").unwrap();
+    assert_ne!(a, b, "query pairs order is preserved in serialization");
+
+    let plus = Url::parse("https://example.com/?q=a+b").unwrap();
+    let encoded = Url::parse("https://example.com/?q=a%20b").unwrap();
+    assert_ne!(plus, encoded, "space as + vs %20 — different keys unless normalized");
+}
diff --git a/types/src/url_normalize.rs b/types/src/url_normalize.rs
new file mode 100644
index 0000000000000000000000000000000000000000..d5bc913f33ad29cd0cd1c7158e28e4fefdbf5f5d
--- /dev/null
+++ b/types/src/url_normalize.rs
@@ -0,0 +1,234 @@
+//! Normalization for external `http(s)://` item identity (not slug tilde ontology).
+//!
+//! Policy (intentional, extend here as new domains need treatment):
+//! - Query pairs sorted lexicographically by **lowercased** key, then value.
+//! - YouTube family → stable `www.youtube.com` shapes where possible.
+
+use url::Url;
+
+/// Whether `-/` → `https://…` path segments should keep original casing (YouTube video IDs are
+/// case-sensitive).
+pub(crate) fn host_preserves_dash_path_case(host: &str) -> bool {
+    let h = host.trim().to_ascii_lowercase();
+    let h = h.strip_prefix("www.").unwrap_or(h.as_str());
+    matches!(
+        h,
+        "youtu.be" | "youtube.com" | "m.youtube.com" | "music.youtube.com"
+    )
+}
+
+/// Normalize external http(s) URLs for stable [`super::paths::canonicalize_item`] output.
+pub fn normalize_http_identity_url(s: &str) -> Option<String> {
+    let mut u = Url::parse(s).ok()?;
+    if !matches!(u.scheme(), "http" | "https") {
+        return None;
+    }
+    rewrite_youtube(&mut u);
+    sort_query_pairs(&mut u);
+    Some(u.to_string())
+}
+
+fn base_host(host: &str) -> String {
+    let lower = host.to_ascii_lowercase();
+    lower
+        .strip_prefix("www.")
+        .unwrap_or(lower.as_str())
+        .to_string()
+}
+
+fn rewrite_youtube(u: &mut Url) {
+    let Some(host_raw) = u.host_str() else {
+        return;
+    };
+    let base = base_host(host_raw);
+    let path = u.path().to_string();
+
+    match base.as_str() {
+        "youtu.be" => {
+            let id = path.trim_start_matches('/').split('/').next().unwrap_or("").to_string();
+            if id.is_empty() {
+                return;
+            }
+            let saved: Vec<(String, String)> = u.query_pairs().into_owned().collect();
+            let Ok(mut out) = Url::parse(&format!("https://www.youtube.com/watch?v={id}")) else {
+                return;
+            };
+            {
+                let mut q = out.query_pairs_mut();
+                for (k, v) in saved {
+                    if k.eq_ignore_ascii_case("v") {
+                        continue;
+                    }
+                    q.append_pair(&k, &v);
+                }
+            }
+            *u = out;
+        }
+        "youtube.com" | "m.youtube.com" => {
+            if base == "m.youtube.com" {
+                let _ = u.set_host(Some("www.youtube.com"));
+            }
+            if path.starts_with("/embed/") {
+                let id = path
+                    .strip_prefix("/embed/")
+                    .unwrap_or("")
+                    .trim_matches('/')
+                    .split('/')
+                    .next()
+                    .unwrap_or("")
+                    .to_string();
+                if id.is_empty() {
+                    return;
+                }
+                let saved: Vec<(String, String)> = u.query_pairs().into_owned().collect();
+                let Ok(mut out) = Url::parse(&format!("https://www.youtube.com/watch?v={id}"))
+                else {
+                    return;
+                };
+                {
+                    let mut q = out.query_pairs_mut();
+                    for (k, v) in saved {
+                        if k.eq_ignore_ascii_case("v") {
+                            continue;
+                        }
+                        q.append_pair(&k, &v);
+                    }
+                }
+                *u = out;
+                return;
+            }
+            if path.starts_with("/v/") {
+                let id = path
+                    .strip_prefix("/v/")
+                    .unwrap_or("")
+                    .trim_matches('/')
+                    .split('/')
+                    .next()
+                    .unwrap_or("")
+                    .to_string();
+                if id.is_empty() {
+                    return;
+                }
+                let saved: Vec<(String, String)> = u.query_pairs().into_owned().collect();
+                let Ok(mut out) = Url::parse(&format!("https://www.youtube.com/watch?v={id}"))
+                else {
+                    return;
+                };
+                {
+                    let mut q = out.query_pairs_mut();
+                    for (k, v) in saved {
+                        if k.eq_ignore_ascii_case("v") {
+                            continue;
+                        }
+                        q.append_pair(&k, &v);
+                    }
+                }
+                *u = out;
+                return;
+            }
+            if path.starts_with("/watch") {
+                let _ = u.set_host(Some("www.youtube.com"));
+                return;
+            }
+            if path.starts_with("/shorts/") {
+                let id = path
+                    .strip_prefix("/shorts/")
+                    .unwrap_or("")
+                    .trim_matches('/')
+                    .split('/')
+                    .next()
+                    .unwrap_or("")
+                    .to_string();
+                if id.is_empty() {
+                    return;
+                }
+                let saved: Vec<(String, String)> = u.query_pairs().into_owned().collect();
+                let Ok(mut out) = Url::parse(&format!("https://www.youtube.com/shorts/{id}"))
+                else {
+                    return;
+                };
+                {
+                    let mut q = out.query_pairs_mut();
+                    for (k, v) in saved {
+                        q.append_pair(&k, &v);
+                    }
+                }
+                *u = out;
+                return;
+            }
+            let _ = u.set_host(Some("www.youtube.com"));
+        }
+        "music.youtube.com" => {}
+        _ => {}
+    }
+}
+
+fn sort_query_pairs(u: &mut Url) {
+    let pairs: Vec<(String, String)> = u.query_pairs().into_owned().collect();
+    if pairs.is_empty() {
+        u.set_query(None);
+        return;
+    }
+    let mut pairs = pairs;
+    pairs.sort_by(|a, b| {
+        a.0.to_ascii_lowercase()
+            .cmp(&b.0.to_ascii_lowercase())
+            .then_with(|| a.1.cmp(&b.1))
+    });
+    u.set_query(None);
+    {
+        let mut q = u.query_pairs_mut();
+        for (k, v) in pairs {
+            q.append_pair(&k, &v);
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn youtube_youtu_be_to_watch() {
+        assert_eq!(
+            normalize_http_identity_url("https://youtu.be/dQw4w9WgXcQ").as_deref(),
+            Some("https://www.youtube.com/watch?v=dQw4w9WgXcQ")
+        );
+    }
+
+    #[test]
+    fn youtube_watch_query_sorted() {
+        assert_eq!(
+            normalize_http_identity_url("https://youtube.com/watch?v=Z&a=1&b=2").as_deref(),
+            Some("https://www.youtube.com/watch?a=1&b=2&v=Z")
+        );
+        assert_eq!(
+            normalize_http_identity_url("https://youtube.com/watch?b=2&a=1&v=Z").as_deref(),
+            Some("https://www.youtube.com/watch?a=1&b=2&v=Z")
+        );
+    }
+
+    #[test]
+    fn youtube_embed_to_watch() {
+        assert_eq!(
+            normalize_http_identity_url("https://www.youtube.com/embed/dQw4w9WgXcQ").as_deref(),
+            Some("https://www.youtube.com/watch?v=dQw4w9WgXcQ")
+        );
+    }
+
+    #[test]
+    fn youtube_shorts_host() {
+        assert_eq!(
+            normalize_http_identity_url("https://youtube.com/shorts/AbCdEfGhIjK").as_deref(),
+            Some("https://www.youtube.com/shorts/AbCdEfGhIjK")
+        );
+    }
+
+    #[test]
+    fn arbitrary_query_sorted() {
+        assert_eq!(
+            normalize_http_identity_url("https://example.com/x?z=1&a=2").as_deref(),
+            Some("https://example.com/x?a=2&z=1")
+        );
+    }
+}


Side B — contributor: tommy-mor
Side B — commit message:
[5db58b98] Improve vote pair selection for spanning trees and rank refinement.

Prefer attaching unranked items to established components before comparing
isolates, then zip down adjacent rank-centrality pairs once the pool is fully
connected, skipping pairs that already have votes.

Co-authored-by: Cursor <cursoragent@cursor.com>

Side B — unified diff (full patch):
diff --git a/server/src/pair.rs b/server/src/pair.rs
index 54b5d2417e9dba04ed8df422156e274c2b2f76b2..c14de4b0502c8b5a17cddf3746077739d56e03e0 100644
--- a/server/src/pair.rs
+++ b/server/src/pair.rs
@@ -3,13 +3,21 @@
 //! Pair selection prefers **bridge** votes — comparisons between items in
 //! different connected components of the voted-pairs graph — so the pool
 //! merges into one ranking group before refining within it.
+//!
+//! Among unvoted bridges, prefer merging established voted components, then
+//! attaching a never-voted child to an established component, and only then
+//! comparing two never-voted children (so the voted graph grows as one tree).
+//!
+//! Once every pool child sits in one voted component, refinement **zips** down
+//! the rank-centrality order: prefer 1 vs 2, then 2 vs 3, and so on, skipping
+//! pairs that already have a vote.
 
 use rand::seq::SliceRandom;
 use std::collections::{HashMap, HashSet};
 
 use crate::{
     path_types::ItemId,
-    ranking::connected_components_from_voted_pairs,
+    ranking::{connected_components_from_voted_pairs, ranked_items},
     reducer::{GlobalTree, GroupState},
 };
 
@@ -28,36 +36,77 @@ fn pair_is_voted(group: &GroupState, a: &ItemId, b: &ItemId) -> bool {
     group.voted_pairs.contains(&(i, j))
 }
 
-/// Component id per pool item: voted-pairs graph components plus one id per
-/// never-voted child.
-fn component_ids(group: &GroupState, pool: &[ItemId]) -> HashMap<ItemId, usize> {
+/// Voted-pairs layout for pool items: component id per item plus which ids are
+/// multi-node voted components (ranked groups in the UI).
+struct ComponentLayout {
+    ids: HashMap<ItemId, usize>,
+    established: HashSet<usize>,
+}
+
+fn component_layout(group: &GroupState, pool: &[ItemId]) -> ComponentLayout {
     let n = group.idx_to_item.len();
     let (comps, isolates) =
         connected_components_from_voted_pairs(n, group.voted_pairs.iter().copied());
 
-    let mut out: HashMap<ItemId, usize> = HashMap::new();
+    let mut established = HashSet::new();
+    let mut ids: HashMap<ItemId, usize> = HashMap::new();
     for (comp_idx, comp) in comps.iter().enumerate() {
+        if comp.len() >= 2 {
+            established.insert(comp_idx);
+        }
         for &idx in comp {
             if idx < n {
-                out.insert(group.idx_to_item[idx].clone(), comp_idx);
+                ids.insert(group.idx_to_item[idx].clone(), comp_idx);
             }
         }
     }
     let mut next = comps.len();
     for &idx in &isolates {
         if idx < n {
-            out.insert(group.idx_to_item[idx].clone(), next);
+            ids.insert(group.idx_to_item[idx].clone(), next);
             next += 1;
         }
     }
     for item in pool {
-        out.entry(item.clone()).or_insert_with(|| {
+        ids.entry(item.clone()).or_insert_with(|| {
             let id = next;
             next += 1;
             id
         });
     }
-    out
+    ComponentLayout { ids, established }
+}
+
+/// Every pool child shares one multi-node voted component (spanning tree phase done).
+fn pool_fully_connected(layout: &ComponentLayout, pool: &[ItemId]) -> bool {
+    if pool.len() < 2 {
+        return false;
+    }
+    let mut comp_id = None;
+    for item in pool {
+        let Some(id) = layout.ids.get(item) else {
+            return false;
+        };
+        if !layout.established.contains(id) {
+            return false;
+        }
+        match comp_id {
+            None => comp_id = Some(*id),
+            Some(expected) if expected == *id => {}
+            _ => return false,
+        }
+    }
+    comp_id.is_some()
+}
+
+/// Pool children that appear in `group`, sorted best rank first.
+fn ranked_pool_order(group: &GroupState, pool: &[ItemId]) -> Vec<ItemId> {
+    let pool_set: HashSet<_> = pool.iter().collect();
+    ranked_items(group)
+        .into_iter()
+        .map(|r| r.item)
+        .filter(|id| pool_set.contains(id))
+        .collect()
 }
 
 #[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord)]
@@ -72,19 +121,108 @@ enum PairPriority {
     WithinVoted = 3,
 }
 
-fn pair_priority(
+/// Tie-break among unvoted bridge pairs.
+#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord)]
+enum BridgeSubPriority {
+    /// Both endpoints lie in established (multi-node) voted components.
+    MergeEstablished = 0,
+    /// One established component member and one never-voted child.
+    AttachIsolate = 1,
+    /// Two never-voted children (separate singleton components).
+    IsolatePair = 2,
+}
+
+/// Tie-break among within-component pairs once the pool is one connected group.
+#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord)]
+struct WithinSubPriority {
+    /// 1 = adjacent ranks (i vs i+1); larger = farther apart in the order.
+    rank_gap: usize,
+    /// min rank index of the two — zip from the top (1 vs 2 before 2 vs 3).
+    zip_index: usize,
+}
+
+const WITHIN_SUB_WORST: WithinSubPriority = WithinSubPriority {
+    rank_gap: usize::MAX,
+    zip_index: usize::MAX,
+};
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord)]
+struct PairSortKey {
+    priority: PairPriority,
+    bridge_sub: BridgeSubPriority,
+    within_sub: WithinSubPriority,
+}
+
+fn item_in_established(layout: &ComponentLayout, item: &ItemId) -> bool {
+    layout
+        .ids
+        .get(item)
+        .is_some_and(|id| layout.established.contains(id))
+}
+
+fn bridge_sub_priority(layout: &ComponentLayout, a: &ItemId, b: &ItemId) -> BridgeSubPriority {
+    let a_est = item_in_established(layout, a);
+    let b_est = item_in_established(layout, b);
+    match (a_est, b_est) {
+        (true, true) => BridgeSubPriority::MergeEstablished,
+        (true, false) | (false, true) => BridgeSubPriority::AttachIsolate,
+        (false, false) => BridgeSubPriority::IsolatePair,
+    }
+}
+
+fn within_sub_priority(
+    group: &GroupState,
+    pool: &[ItemId],
+    layout: &ComponentLayout,
+    a: &ItemId,
+    b: &ItemId,
+) -> WithinSubPriority {
+    if !pool_fully_connected(layout, pool) {
+        return WITHIN_SUB_WORST;
+    }
+    let order = ranked_pool_order(group, pool);
+    let (Some(i), Some(j)) = (order.iter().position(|x| x == a), order.iter().position(|x| x == b))
+    else {
+        return WITHIN_SUB_WORST;
+    };
+    WithinSubPriority {
+        rank_gap: i.abs_diff(j),
+        zip_index: i.min(j),
+    }
+}
+
+fn pair_sort_key(
     group: &GroupState,
-    components: &HashMap<ItemId, usize>,
+    pool: &[ItemId],
+    layout: &ComponentLayout,
     a: &ItemId,
     b: &ItemId,
-) -> PairPriority {
+) -> PairSortKey {
     let voted = pair_is_voted(group, a, b);
-    let bridge = components.get(a) != components.get(b);
-    match (bridge, voted) {
+    let bridge = layout.ids.get(a) != layout.ids.get(b);
+    let priority = match (bridge, voted) {
         (true, false) => PairPriority::BridgeUnvoted,
         (false, false) => PairPriority::WithinUnvoted,
         (true, true) => PairPriority::BridgeVoted,
         (false, true) => PairPriority::WithinVoted,
+    };
+    let bridge_sub = if priority == PairPriority::BridgeUnvoted {
+        bridge_sub_priority(layout, a, b)
+    } else {
+        BridgeSubPriority::MergeEstablished
+    };
+    let within_sub = if matches!(
+        priority,
+        PairPriority::WithinUnvoted | PairPriority::WithinVoted
+    ) {
+        within_sub_priority(group, pool, layout, a, b)
+    } else {
+        WITHIN_SUB_WORST
+    };
+    PairSortKey {
+        priority,
+        bridge_sub,
+        within_sub,
     }
 }
 
@@ -109,9 +247,12 @@ fn candidate_pairs(pool: &[ItemId], exclude: Option<(&ItemId, &ItemId)>) -> Vec<
 
 /// Pick the next pair to vote on within `pool`.
 ///
-/// 1. Prefer unvoted **bridge** pairs (connect separate ranking components).
-/// 2. Then unvoted within-component pairs (refinement).
-/// 3. Then already-voted pairs (re-compare).
+/// 1. Prefer unvoted **bridge** pairs (connect separate ranking components),
+///    with sub-priority: merge established components, attach an isolate to
+///    established, then compare two isolates.
+/// 2. Then unvoted within-component pairs; when the pool is one connected group,
+///    prefer adjacent ranks (1 vs 2, 2 vs 3, …) in order, skipping voted pairs.
+/// 3. Then already-voted pairs (re-compare), with the same zip ordering.
 pub fn suggest_next_pair_in_pool(
     group: &GroupState,
     pool: &[ItemId],
@@ -121,15 +262,15 @@ pub fn suggest_next_pair_in_pool(
     if candidates.is_empty() {
         return None;
     }
-    let components = component_ids(group, pool);
+    let layout = component_layout(group, pool);
     let best = candidates
         .iter()
-        .map(|(a, b)| (pair_priority(group, &components, a, b), (a, b)))
-        .min_by_key(|(p, _)| *p)?
+        .map(|(a, b)| (pair_sort_key(group, pool, &layout, a, b), (a, b)))
+        .min_by_key(|(k, _)| *k)?
         .0;
     let best_pairs: Vec<(ItemId, ItemId)> = candidates
         .into_iter()
-        .filter(|(a, b)| pair_priority(group, &components, a, b) == best)
+        .filter(|(a, b)| pair_sort_key(group, pool, &layout, a, b) == best)
         .collect();
     best_pairs.choose(&mut rand::thread_rng()).cloned()
 }
@@ -303,6 +444,38 @@ mod tests {
         assert!(from_ab && from_cd, "expected bridge pair, got {:?}", chosen);
     }
 
+    #[test]
+    fn suggest_prefers_attach_over_isolate_pair_among_many_unranked() {
+        let parent = ItemId::parse("reddit.com/r/rust").unwrap();
+        let mut tree = seed_children(
+            &parent,
+            &[
+                "reddit.com/r/rust/a",
+                "reddit.com/r/rust/b",
+                "reddit.com/r/rust/c",
+                "reddit.com/r/rust/d",
+                "reddit.com/r/rust/e",
+            ],
+        );
+        let ab =
+            VoteData::from_recorded(1, "reddit.com/r/rust/a", "reddit.com/r/rust/b", 2, 1).unwrap();
+        tree.apply_vote(&parent, ab);
+        let group = tree.get(&parent).unwrap().local_ranking.clone();
+        let pool = children_of(&tree, &parent);
+        let pair = suggest_next_pair_in_pool(&group, &pool, None).unwrap();
+        let chosen = pair_set(&pair);
+        let from_ab =
+            chosen.contains("reddit.com/r/rust/a") || chosen.contains("reddit.com/r/rust/b");
+        let from_cde = chosen.contains("reddit.com/r/rust/c")
+            || chosen.contains("reddit.com/r/rust/d")
+            || chosen.contains("reddit.com/r/rust/e");
+        assert!(
+            from_ab && from_cde,
+            "expected ranked+unranked attach, got {:?}",
+            chosen
+        );
+    }
+
     #[test]
     fn suggest_connects_isolate_to_existing_component() {
         let parent = ItemId::parse("reddit.com/r/rust").unwrap();
@@ -325,6 +498,65 @@ mod tests {
         assert!(chosen.contains("reddit.com/r/rust/a") || chosen.contains("reddit.com/r/rust/b"));
     }
 
+    #[test]
+    fn suggest_zips_adjacent_ranks_when_tree_complete() {
+        let parent = ItemId::parse("reddit.com/r/rust").unwrap();
+        let mut tree = seed_children(
+            &parent,
+            &[
+                "reddit.com/r/rust/a",
+                "reddit.com/r/rust/b",
+                "reddit.com/r/rust/c",
+            ],
+        );
+        // Star at a connects all three; b-c is the only unvoted adjacent pair left.
+        for (a, b, l, r) in [
+            ("reddit.com/r/rust/a", "reddit.com/r/rust/b", 3, 1),
+            ("reddit.com/r/rust/a", "reddit.com/r/rust/c", 2, 1),
+        ] {
+            let v = VoteData::from_recorded(1, a, b, l, r).unwrap();
+            tree.apply_vote(&parent, v);
+        }
+        let group = tree.get(&parent).unwrap().local_ranking.clone();
+        let pool = children_of(&tree, &parent);
+        let pair = suggest_next_pair_in_pool(&group, &pool, None).unwrap();
+        let chosen = pair_set(&pair);
+        // a-b and a-c voted; b-c is the only unvoted adjacent pair in rank order.
+        assert!(chosen.contains("reddit.com/r/rust/b"));
+        assert!(chosen.contains("reddit.com/r/rust/c"));
+    }
+
+    #[test]
+    fn suggest_zip_prefers_1v2_before_2v3_when_both_unvoted() {
+        let parent = ItemId::parse("reddit.com/r/rust").unwrap();
+        let mut tree = seed_children(
+            &parent,
+            &[
+                "reddit.com/r/rust/a",
+                "reddit.com/r/rust/b",
+                "reddit.com/r/rust/c",
+                "reddit.com/r/rust/d",
+            ],
+        );
+        // Hub at c connects all four; leave rank-adjacent a-b and b-c unvoted.
+        for (a, b, l, r) in [
+            ("reddit.com/r/rust/c", "reddit.com/r/rust/d", 3, 1),
+            ("reddit.com/r/rust/b", "reddit.com/r/rust/c", 2, 1),
+            ("reddit.com/r/rust/a", "reddit.com/r/rust/c", 2, 1),
+        ] {
+            let v = VoteData::from_recorded(1, a, b, l, r).unwrap();
+            tree.apply_vote(&parent, v);
+        }
+        let group = tree.get(&parent).unwrap().local_ranking.clone();
+        let pool = children_of(&tree, &parent);
+        assert!(pool_fully_connected(&component_layout(&group, &pool), &pool));
+        let pair = suggest_next_pair_in_pool(&group, &pool, None).unwrap();
+        let chosen = pair_set(&pair);
+        // Top adjacent unvoted edge should be a-b (zip index 0), not b-c (index 1).
+        assert!(chosen.contains("reddit.com/r/rust/a"));
+        assert!(chosen.contains("reddit.com/r/rust/b"));
+    }
+
     #[test]
     fn resolve_pair_picks_from_pool() {
         let parent = ItemId::parse("reddit.com/r/rust").unwrap();