But in its simplest form, you had just a small sliver of code that allowed you to edit text files that then were rendered as HTML to the user dynamically, the most important "markup" being a translation of WikiWords to hrefs, allowing you to easily link together PageAboutFoo to PageAboutBar.
That's it. There were some very minimal implementations that did this with e.g. a few lines of Perl.
Everything else is either luxury or bloat.
It's based on a ~400 line Python script that extracts mails from specified folders in my mailbox and a ~100 line Bash script that turns it into a website. It's not user-friendly really... but I like the fact that the system is rather simple (bias mine).
The Python script takes a mailbox folder, concatenates the bodies of all the messages and stores the result in a single file derived from the mailbox-folder's name. The attachments for a message are stored in a dedicated folder for each message, where the names of those folder are derived from the subjects of the messages, and the attachments' filenames are simplified.
The Bash script then assumes that the resulting collection of concatenated files are in markdown format and runs Pandoc on them to generate the HTML files. Then, the entire result is copied to document-root.
As it is, it is mainly useful for (micro)blogging... it's a bit tedious to edit markdown on a smart-phone, especially linking to files. On the other hand, it's basically Twitter in 500 LOC. There's got to be some value in that.
I recently kind of did a 2019 version of this. I built a site with gatsbyjs, and users can submit corrections via a form that POSTs to a google survey. I look at the row in the spreadsheet the survey created, if it's good I copy it to another spreadsheet and redeploy. I haven't automated that part as I get submissions very rarely, but I could.
Not saying it's a good way to build a website. But it is interesting exploring ways to build a site without a backend that probably should have one :)