Looks like anything above ID 8000000 returns a 404 too.

Together, the two of you can probably guess the regex responsible :)

Does this mean that old content will be permanently inaccessible? One of my (rarely exercised) pastimes used to be typing a few random five- or six- digit IDs into the address bar, then following the chain up to the top-level article. I found some really interesting stuff that way.

I suspect that's intentional; older items are less likely to be cached.

Regex pattern gone wrong?