My top blog posts
Thursday, November 14, 2024
Comments: 7 (latest 2 days later)
Tagged: zarf, blog, stats, navel-gazing, static site generators
Over on Bluesky, someone asked whether everybody's blog traffic has declined over the past few weeks. (Original message requires Bsky login so I won't link directly.)
This got me looking at my blog stats, which I normally never, ever do.
Short answer: I don't think so? But my blog is tiny and the data is noisy, so how would I even tell. Some of my posts get popular -- though never what you'd call "viral". Most don't.
But why not peek at the data? I have logs going back to June 2023, when I moved off of Blogger.com. Yes, it's navel-gazing. But now that I'm shifting some weight onto Bluesky, I might have some followers who haven't seen my older posts.
Daily unique IPs in my blog traffic.
My first cut was daily unique IPs in the HTTP requests. The three spikes are:
- (2023-Oct-13) Microsoft consumes Activision; and a plea
- (2024-Aug-10) The Tic-Tac-Toe Mysteries of Xerloc O'Xolmes
- (2024-Aug-30) Tabbed out on the Oregon Trail
You could maybe argue that my traffic went up around May of this year, and then fell off again after September, but what do you control for? I wrote seven blog posts in the spring months (see chart) but thirteen in the summer, what with being laid off. No surprise if the summer got more visits.
But the serious confounding factor is that the Oregon Trail post has images. So of course it caused more traffic! (I am proud of the ASCII art in the Tic-Tac-Toe post, but it's not <img> tags so no extra HTTP requests.)
Also, of course, the vast majority of the HTTP requests are bots, crawlers, and feed-readers. I could try to filter those out but it probably wouldn't change the picture.
Let's try this differently. What are my most popular posts over the whole year-and-a-half of logging history? Remember, I imported my old posts from Blogger (and kept the URLs!) so I can measure the current traffic to my entire blog history.
Here's my current top twenty:
| Hits | Date | Post |
|---|---|---|
| 79963 | 2023-10 | microsoft-consumes-activision |
| 43384 | 2024-08 | tabbed-out-on-the-oregon-trail |
| 25475 | 2024-08 | tic-tac-toe-mysteries |
| 19449 | 2023-08 | the-myth-of-google-plus |
| 17404 | 2023-11 | infocom-interpreters |
| 16482 | 2023-02 | a-treasury-of-zork-maps |
| 8545 | 2023-07 | invention-in-rust |
| 7987 | 2024-01 | parser-if-disambiguation |
| 7721 | 2019-04 | what-is-zil-anyway |
| 6103 | 2017-08 | your-load-is-too-heavy-zork-deep-reading |
| 6033 | 2023-05 | sydney-obeys-any-command-that-rhymes |
| 5883 | 2019-04 | all-of-infocoms-game-source-code |
| 5875 | 2022-05 | cragne-manor-source-code-some-of-it |
| 5872 | 2024-02 | download-the-ifarchive |
| 5871 | 2024-02 | what-happened-to-kickstarter |
| 5692 | 2022-05 | patricia-mckillip-1948-2022 |
| 4709 | 2023-07 | summer-miscellany |
| 4598 | 2024-01 | the-serif-on-the-n |
| 4387 | 2023-09 | canon-is-kayfabe |
| 4195 | 2021-07 | mysterium-2021-report |
(Sorry, you get slugs rather than post titles there. You can figure it out.)
Same top three, but the Microsoft post is now the winner. That's the one where I pushed the idea of Activision/Microsoft open-sourcing the old Infocom IP. Apparently people were big on boosting that.
Status: Didn't go much of anywhere. I got in contact with someone at then-Activision and someone at Microsoft, but it kind of washed out in a sea of "We'll see what the lawyers say." Which, to be honest, I expected.
I'll check back! It's been most of a year since I last nudged anybody. This is a good reminder that I should follow up.
Other than that, the Tic-Tac-Toe puzzle post holds up very well. That makes me happy. (Take a look if you enjoy logic puzzles!)
And it's nice to see some older blog posts (Zork inventory logic, 2017) hold people's interest.
People really love the substantive posts about videogame history and technology. Infocom source code, Zork fan maps, that Oregon Trail book (which isn't old history, but made for nice photos). People seem to regularly cite my deep dive into Google's search terms.
On the flip side, my game reviews (indexed here) do not get high numbers. I'm fine with that, really. I am confident that people read the reviews; they just don't pass them around to their friends shouting "Ooh, have you seen Zarf's take on Myst?!" (Spoiler: I liked it.)
But my intensive inquest into filthy sea shanty lyrics also got lost in the weeds. It landed 37th -- people read it, but nobody boosted. I'm sad about that. I worked hard on that one. Oh well.
On the funny side, there's a lot of badly written crawlers out there. I get a surprising number of hits to URLs like /2023/11/null and /2024/08/null. Not to mention /2024/07/zqxjzVqqPMGKyu.html. No idea where that came from.
Also something, possibly a feed reader, is very insistent on checking for ArticleDetailImageFullInactive.png and similar image names.
Apologies to the legions of bots poking /wp-login.php hoping to see blood spray. Static site generators for the win.
So do I now have a roadmap for maximizing my blog's engagment? Psshh no. This is for fun. I promise to keep writing about whatever I feel like.
I definitely won't blog more blog posts about the posts on my blog. Enough navel for one year. Anyhow, I've poisoned the statistics! This post will bias all future lists in favor of the ones already enshrined here.
Such is data.
Comments from Mastodon
@zarfeblong my traffic has been up, not down lately (why were they asking about traffic being down?)
mind you, my numbers are less than your by a significant multiplier, although it's possible wordpress counts some things differently than your software
@jdyer I’m sure it’s not comparable. I didn’t try to control for bots and crawlers, except for checking unique IPs, and not even that in the second chart. Wordpress is probably much smarter.
@jdyer Come to think of it, a huge part of the web traffic must be Mastodon clients previewing the post as the link gets viewed. We both post blog URLs to Mastodon, but I have about four times as many followers as you. So that probably account for a lot of difference.
@zarfeblong @jdyer This is absolutely true. JWZ figured it out in the early post-Twitter days and I have seen it reflected in my own case (using “bise”, my own homemade log-analyzer tool). https://www.jwz.org/blog/2022/11/mastodon-stampede/
@jmac @zarfeblong the relation between wordpress and mastadon is weird - I don't ever actually see if someone clicks from Mastadon to my site (whereas I can sometimes tell if a person arrived from bluesky, twitter, etc)
I did some tests to confirm mastodon is just considered invisible
I always assume the clicks registered, just anonymously, but maybe it simply didn't register at all!
maybe this was a way to compensate for the preview-view
@jmac @zarfeblong I also have some quite intentional click reduction with having the games put into categories, so if I made three posts on a game, you can click one link to read about them chronologically, rather than have to click-click-click all of them
I found on other sites I would often bail early if I had to keep clicking and hunting for the next post on a game, but the category method makes it feel a lot better to just curl up on read the whole thing like an essay
I promise that I’ve already deleted the hacky shell scripts I used to trawl the logs, so I can’t get into the habit of doing this regularly.