Architecture and Building Things Right

The WordPress Coding Standards PHP_CodeSniffer rules or “sniffs” have been updated to the 3.2 version of the standards for accessibility, PHP, HTML, CSS, and JavaScript.

WordPress VIP has some nice documentation and new sandbox demos for Remote Data Blocks. Remote Data Blocks make it easy to pull remote data into a templated block or inline content. Now you have no-, low-, and vibe-code options for using Remote Data Blocks. Airtable, Google Sheets, and Shopify can be selected as data sources out of the box, or you can code your own API connection. (Docs, code samples, and AI coding prompts are provided.)

As usual, the latest issue of the Gutenberg Times highlights new things coming or already delivered in the WordPress block and site editor, like the evolution of the Site Editor, refined content creation, expansion of the Command Palette, developer updates, and performance improvements. Check out the roadmap to WordPress 6.9 for the full details. As Rae Morey reports in The Repository, “in a first since 2010, there won’t be a new default theme.” (More on that below.)

Lots of people are excited about a new, free Menu Designer plugin coming from Ollie-creator Mike McAllister. It makes even complex (mega-)menu design easy, fun, and attractive in WordPress. Try the demo yourself! Jamie Marsland is right, WordPress Is About to Get a Whole Lot Better with this addition, which is hard to imagine not going into core. It massively improves menu creation and management.

https://twitter.com/mikemcalister/status/1950644457650733295

Ronnie Burt at Gravatar wrote a really good explainer for federated identity management (FIM) covering SAML, OAuth, and OpenID Connect. You have used these protocols for single-sign on (SSO) at some point, maybe every day. FIM is nicely described as “SSO with a passport,” as if logging in with a Google account once would carry over not just to other Google properties but you own website, social media and retail store or financial accounts. Gravatar could work this way and clearly has its future aimed at being the FIM “regular people can actually use.”

Felix Arntz released AI Services 0.7.0 which adds support for text generation, text-to-speech, and more for the major AI platform large language models.

Mahangu Weerasinghe shared an unusual and interesting Data/Analytics Engineering Starter Project: Extracting Data from WordPress.org with Meltano. (GitHub repo) In it, you’ll learn “how to “how to build a complete data pipeline using the WordPress.org API Meltano extractor, tap-wordpress-org.” The .org API gives you access to the 60k+ plugins and 10k+ themes and their metadata.

Learn how to use the WordPress Group block to create many kinds of layouts in this video tutorial.

Good System Design

What are the hallmarks of good system design? Sean Goedecke says it’s not about anything impresessive or complex but simplicity and standards:

good system design is not about clever tricks, it’s about knowing how to use boring, well-tested components in the right place. I’m not a plumber, but I imagine good plumbing is similar: if you’re doing something too exciting, you’re probably going to end up with crap all over yourself.

There’s a lot of detail and discussion of different contexts, from databases to logging — worthwhile reading.

Related: The Periodic Table of Systems Design

Remkus DeVries explains when (and when not) to create additional database tables in WordPress. The matters for performance at scale with large amouts of data, so if that’s not your use case, you don’t need to add tables to the WordPress database. Unfortunately many WordPress plugins do add their own tables (or postmeta fields) and hardly any plugin removes whatever they add. Remkus doesn’t get into that, but it’s an often overlooked drag on performance and potential security issue.

David Lockie has some great thoughts on content-based design: Content for fun vs. content for purpose: designing for two distinct modes of consumption. After discussing some practical examples of experiential and utilitarian design, David connects them with AI and the future of content platforms:

As AI becomes more capable of handling utilitarian content consumption, the experiential mode may become increasingly valuable and distinctly human. This doesn’t mean utilitarian content becomes less important—rather, it is processed more efficiently, freeing us to invest our attention in experiences that truly require human consciousness.

Content creators and platform designers who recognize rapid mode switching will be better positioned to serve their audiences. Instead of fighting against AI summarization or trying to make everything  “engaging,” they can thoughtfully design for the mode their content is primarily meant to serve while still accommodating clear switching points to the inverse use case.

The most successful content strategies of the future may be those that explicitly design for both modes within single experiences, creating clear visual and structural cues that help users navigate between task-focused and experiential consumption. After all, both modes serve essential human needs, and often these needs arise within moments of each other. 

I’m curious what a good example of this might look like for a content platform.

How much energy is wasted by 404 pages?

Jeff Starr considers 404s (and other 400-level) HTTP errors at scale and why we should take care to make them stable for the long-term. Various sources gathered on Wikipedia (including Pew Research) suggest about 25% of all web pages that once existed no longer do — that’s some serious link rot!

(For Pew’s study, “web page” refers to an HTML document that has been indexed by Common Crawl but not dependent assets, like images, that are requested when the primary page URL is requested.)

Citing Jonathan Zittrain, Paolo Belcastro notes:

→ Half of all links in U.S. Supreme Court opinions since 1996 no longer work.

→ Three quarters of links in the Harvard Law Review are dead.

→ In The New York Times archive (≈2 million links since 1996), a quarter of deep links have rotted.

Gemini (a less than reliable source) told me (without citing a source) there are trillions of links accessed daily, but how many come from intentional clicks or typed-in URLs that lead to a 404 where a known page once existed? This number should exclude mis-typed URLs in the browser or in a link reference in a source document of any type.

Daniel Stenberg cites Mozilla for a median of nearly 3,000 HTTP requests via Firefox on one day in 2019. From this data, he extrapolates over 13 trillion requests per day globally. Keep in mind this includes all HTTP requests made through a browser, including dependencies referenced within a HTML document (a “web page”) when it is requested. So, based on average page weight in 2019. Stenberg estimated a global average of 15 actual sites being visited via links or direct entry of a URL in the browser’s address bar by humans, not bots.

How many links do I click in a day? Like, real links on the open web that cause my browser to make an HTTP request as opposed to local and SaaS applications that don’t take me anywhere? My browser history varies widely throughout the week. I should check my own average and see how many of my bookmarks are broken.

Key assumption: Let’s make a conservative estimate of just one trillion total requests per day for web pages people expect to exist because they’re coming from a link or keyed-in URL that once pointed to a known web page. (I’d guess most of these URLs lead to traditional, content-rich pages that are intended for reading.)

Here is my back-of-the-napkin math based on this and further assumptions:

• 250 Billion (25% of one trillion) URLs clicked each day get a 404 page.
• Our own fancy custom 404 page has to be loaded just 1,000 times to use 1kWh of energy but still has an “A” rating and is deemed more efficient than 92% of all webpages according to Website Carbon Calculator. This amounts to 1Wh per page load, or the amount of energy an old 60W incandescent lightbulb uses in 1 minute. (Here’s an interesting discussion of the reliability of website carbon calculators continuing on Bluesky.)
• That seems rather high, even for a very old server and desktop computer, especially when the average Google search uses only 0.0003 kWh (0.3 Wh).
  https://fershad.com/notes/sceptical-about-website-emissions/
  Let’s take these two numbers as high and low marks. (Using data.firefox.com we might be able to come up with a more exact figure for the client-side energy expenditure, but servers and networks…???)
• Let’s also say this page load takes only 1 second or so.
• That’s a range from 75 to 250 billion watt (75-250 gigawatt) hours wasted on 404 pages each day.
• For reference:
  • An average lightning strike delivers power in the range of 1-10 GW and possesses about 1GJ (gigajoules) in terms of total energy.
  • New York City uses approximately 50TW (terrawatts) per year (50,000 GWh) which averages out to 137 GWh per day. (137 GWh = 493,200 GJ or 493.2 TJ)
    • While the average demand is around 5,500 MW (5.5 GW), peak demand can significantly exceed the average. The record in peak demand for all of New York State at one moment in time is 33,956 MW (33.956 GW), set in July 2013 during a heat wave.
    • Source: NYC.gov, EIA.gov, NYISO.com)

By this estimate, the daily energy wasted on all 404s globally is somewhere in the realm of the average daily power consumption of the entire New York metro area with a population of approximately 10 million people. 🤯

Now if you really want to blow your mind, read endtimes.dev on Why your website should be under 14kB in size:

What is surprising is that a 14kB page can load much faster than a 15kB page — maybe 612ms faster — while the difference between a 15kB and a 16kB page is trivial.

This is because of the TCP slow start algorithm….

🤯🤯🤯

Performance Gains

Feel better about your fat, wasteful web pages and broken links by making any custom CSS you have in the WordPress Customizer a static file that can be cached instead of dynamically rendered inline code. Matt Cromwell explains how.

Ronald Huereca explains How to Generate a Block Manifest to Improve Block Performance.

David Artiss blogged about all the plugins he uses on his blog. They’re mostly excellent utility plugins that do one thing well for interface enhancements or performance. I love lists like this because most WordPress users won’t have heard of 90% of these great free, community plugins.

Jon Henshaw also shared not just the plugins but the settings he uses for optimal WordPress performance.

Instant Back/Forward Navigations in WordPress is a nice, new utility plugin from Weston Ruter that does just what it says.

WooCommerce 10.1 is coming with performance enhancements and other polishing demonstrated by Rodolfo Melogli in a video tour of the first release candidate.

If you think fatal errors in WordPress shouldn’t spawn email notificatons that go to the site admin, Torsten Landsiedel agrees and explains how to change this.

The Future of Woo and Themes

The default “Purple” theme for WooCommerce is going away in favor of “providing the best commerce-building experience on top of a shared, modern WordPress theme, delivering WooCommerce specific patterns, templates, and style variations directly within WooCommerce itself.”

Discussion on Reddit: Purple’s dead. What’s your default theme for WooCommerce, then?

There’s a larger context for this decision. Curently, new core theme development is on hold, and the existing WordPress core themes could be retired. So don’t expect a Twenty Twenty-Six theme. Instead we might see starter templates, curated style variations, or pattern showcases. This would open space for “a shared, modern WordPress theme” for Woo and all other major plugins and theme systems to adhere to as a de facto compatibility standard.