diff --git a/.github/scripts/backfill_blog_posts.py b/.github/scripts/backfill_blog_posts.py
index 1349d853..a655dc83 100644
--- a/.github/scripts/backfill_blog_posts.py
+++ b/.github/scripts/backfill_blog_posts.py
@@ -19,8 +19,9 @@
 
 GitHub's release list can lag a just-published Release by several seconds,
 so with --wait-for-prs it lists again until every named PR has a Release,
-for up to a minute. A PR whose release failed never appears; after the last
-attempt the run writes what is listed, and the next run picks up the rest.
+every 2s for up to half a minute. A PR whose release failed never appears;
+after the last attempt the run writes what is listed, and the next run
+picks up the rest.
 """
 
 import argparse
@@ -42,8 +43,8 @@ def parse_pr_numbers(text):
     return {int(n) for n in numbers}
 
 
-WAIT_ATTEMPTS = 12
-WAIT_DELAY = 5
+WAIT_ATTEMPTS = 15
+WAIT_DELAY = 2
 
 
 def listed_releases(gh, wait_for=None, attempts=WAIT_ATTEMPTS, delay=WAIT_DELAY, sleep=time.sleep):
diff --git a/.github/workflows/release.yml b/.github/workflows/release.yml
index badc39a2..88c51a5f 100644
--- a/.github/workflows/release.yml
+++ b/.github/workflows/release.yml
@@ -392,7 +392,7 @@ jobs:
         run: |
           # The release list can lag a Release published seconds ago, so name
           # the PRs this run released: the script lists again until each has
-          # one, for up to a minute.
+          # one, for up to half a minute.
           python3 .github/scripts/backfill_blog_posts.py --wait-for-prs "$(jq -r 'join(",")' <<< "${QUEUE:-[]}")"
 
           # git status (not git diff) so brand-new, still-untracked files count.