Redirects and old links

An address somebody has linked to is an asset. Settings › Redirects is the list of every old address this blog is still answering for, and the place to add more.

What is created for you

You should not have to think about most of these.

Renaming a published post. The old address redirects to the new one with a 301. Rename it twice and the oldest address still points at where the post is now rather than at the middle name.

Renaming a category or a tag. Same, and for a category its feed address moves with it, because a subscription is the link least likely to be fixed by hand.

Merging two categories. The one that went away redirects to the one you merged into.

Renaming the blog's own address. Every old address on the previous hostname redirects to the new one.

An import. Every address a post had on its old platform, including the platform's own URL shapes: WordPress's /?p=123, Blogger's /2019/04/a-post.html, Substack's /p/a-post.

A slug the web cannot carry. An import keeps slugs exactly, except when a slug genuinely cannot be a web address. Then it is transliterated and the original becomes a redirect.

Each row in the list says where it came from, so you can tell your own rules apart from the ones the product wrote.

Adding your own

Three fields: from, to, and what kind.

From is a path on this blog, starting with a slash.

To is where the reader should end up.

Kind is the part people get wrong, so the list spells each one out.

A 301 is moved for good, and it is what you want almost every time. It passes on whatever the old address had earned.

A 302 is moved for now. Use it for something genuinely temporary, like pointing an article at a live event page for a week.

A 307 and a 308 are the same two, for cases where the request method has to survive the hop. A blog rarely needs either.

A 410 is gone on purpose. Not an error: a deliberate statement that the page existed, was removed, and is not coming back. It is the honest answer for a post you have deleted with nothing to replace it, and it gets the page out of search results faster than letting it 404 quietly.

Test an address

Under the list is a box. Paste a path or a whole URL and it says exactly what a reader asking for that address would get: which rule matched, what kind of redirect it is, and where they land. If nothing matches, it says the blog would look for a post at that path instead.

Use it before you go looking for a bug. "The old link is broken" and "the old link redirects to a post that has itself been renamed" look identical from outside and are very different problems.

What cannot be a redirect

A redirect cannot shadow a published post. If you try to send /the-long-way-round somewhere while a live post sits at that address, it is refused, and the message says to change the post's address instead. Allowing it would make the post unreachable with no sign anywhere of why.

Redirects are for this blog only. They cannot affect the rest of your site, even when the blog is served at yourdomain.com/blog, because everything outside that path is answered by your own site rather than by us.

Before you delete a post

Deleting is the one action here that can leave a dead address, because nothing is created for you.

Decide what should happen to the old address first, and add the redirect: a 301 to whatever replaced the post, or a 410 if nothing did. It takes a minute now and is impossible to reconstruct in a year when you no longer remember what was there.