Luke Carbis

Who decides the future of themes in WordPress?

I remember back when Gutenberg was still in beta, we were talking a lot about what the block editor would mean for the future of WordPress.

Would blocks be distributed in themes, or in plugins? What would Gutenberg look like if it were combined with the Customiser? Should widgets be blocks (and if so, what about the header logo, or copyright colophon)?

This week we started to get some early rumblings of answers to these questions. Riad Benguella created a Pull Request to the Gutenberg GitHub repo, proposing an experiment for completely block based themes.

It described something very important! The future of themes in WordPress! Unfortunately, it was published on GitHub, and to read it, you needed to view the changeset and parse the markdown. Of course, the WP community doesn’t read every new PR that comes through the Gutenberg repo – so this could have easily flew under the radar.

But we got lucky – it caught the attention of Justin Tadlock over on WP Tavern, which drew some attention to the PR.

This Pull Request has been met with a considerable degree of concern. I wanted to take the time to write about this from the perspective of:

1. What does this experiment look like?
2. Why is there resistance to the idea?
3. Where do we go from here?

What does this experiment look like?

Sorry Riad! I think a large part of the problem here is that this first documentation on block based themes is actually quite difficult to understand. I spoke to a few different people who couldn’t quite wrap their head around it.

Here’s my interpretation:

Let’s start by imagining that blocks could be used as widgets. You could add, say, a Quote block to any one of your widget areas.

Okay, makes sense so far. Now imagine replacing your footer with a widget area. Chances are, your footer already contains one, so this shouldn’t be too hard. But instead of a © section at the bottom, imagine this is a paragraph block instead, and there’s an interface in the WP Admin to change the text displayed here.

Now, did you know that in the latest version of Gutenberg there’s a “Navigation” block? This is what it looks like:

So, now we could easily replace the navigation in our header with a widget area, and insert the Navigation block. While we’re there, we could insert our logo as an image block, and align it to the left.

At this point, the name “Widget Area” isn’t such a great fit. Let’s call it a “Template Part” instead. We’ll have Template Parts like Header, Footer, and Sidebar.

Hey, now that we’re thinking about it, don’t forget that WordPress 5.3 introduced Group blocks. We could use these to do something pretty cool…

Start with a completely blank page. Nothing on it – no header, no sidebar, nothing. We could add a group block to the top, and make that our Header Template Part. Then below that we could add another Group Block, and expect the Post Title and Post Content to be added there. Maybe another Group Block next, and put Comments in that. Let’s finish off with a Group block for the Sidebar Template Part, and then a Footer Group.

We’ll call it “single”. It will be structured something like…

• Group block
  • Header Template Part
    • “Logo” image block
    • Navigation block
• Group block
  • Post Title block
  • Post Content block
• Group block
  • Comments block
• Group block
  • Sidebar Template Part
    • Search block
    • Categories block
• Group block
  • Footer Template Part
    • “Copyright” paragraph block

(Notice our Template Parts are kind of like reusable block groups.)

This is called a Block Template. We could create it in some sort of Template Editor in the WP-Admin, and export it to be used as a theme.

So themes would be responsible for three things:

Style

This includes the structure of how the Template Parts fit together visually, as well as styles for all blocks (they might even be styled differently in different Template Parts).

Block Templates

Not dissimilar to current theme templates, and maintaining adherence to the Template Hierarchy, these describe the overall structure of the page, in the language of blocks (especially Group blocks) and Template Parts.

Template Parts

The theme should probably add some default blocks: a logo and navigation to the header, and a paragraph with copyright text to the footer. Even though these blocks could be easily changed and removed by the user, the theme should include a starting point. These defaults belong in a Template Part, and the Template Part is included inside a Block Template.

Hopefully that makes things a little clearer! For those of us who have been using traditional WordPress templating for the past decade or more, this is quite a tough change to wrap your head around. 💫😵💫

Why is there resistance to the idea?

Aah… politics. Unsurprisingly, there’s some #wpdrama going down because of this proposal. Some of it is based on simple misunderstanding (a.k.a. poor communication), but there’s more to it than that.

The subtext here is that Riad works for Automattic. I won’t dive into the history of WordPress and Automattic, suffice to say that their CEO leads the WordPress project¹, and as such, there is a constant concern from the community when Automattic approaches conflict-of-interest boundaries.

So, when an Automattician makes a giant proposal on the future of theming in WordPress, which gets merged into the Gutenberg project the very next day², I think it’s very reasonable that people are concerned.

Here’s a timeline of how it went down³:

September 5: Automattician Matías Ventura publishes a Make WordPress post describing full site editing with blocks. It’s met with a lot of positivity, but doesn’t include much detail on technical implementation.

September 23: Google employee Felix Arntz created an Issue in the Gutenberg Github repo describing Matías’ post. The 15 comments on that issue (before the PR on December 4) were all made by Automattic or Google employees.

December 4: Automattician Riad Benguella creates a PR for documenting the experiment. There is some early concern from Joy Reynolds (a well-known WordPress core contributor), but there’s no time to discuss it further, because…

December 5: Automattician Enrique Piqueras approves the PR just 24 hours later!

December 5: Riad merges the PR (the day after it’s proposed).

Notice a pattern?

The people with power all work for Automattic and Google.

It seems fairly clear to me why there is so much resistance to the idea. I believe it’s because:

1. The experiment was poorly explained, and widely misunderstood.
2. There wasn’t any opportunity to understand the technical implementation.
3. There wasn’t any opportunity to discuss it either.
4. The whole project is being rushed through by Automattic (with some help from Google).
5. Feedback / concern is being brushed aside or hand-waved away, instead of properly discussed.

December 5: Hours after this PR is approved, Derek Hermann chimes in. Derek has been contributing to WordPress for years. Over that time he was a feature lead on the Customiser, worked with me to build Tide, and has been very active in Gutenberg’s development. He’s also the CTO of XWP (one of the largest WordPress VIP agencies in the world).

Derek has some real concerns, and lots of questions, which Matías tries his best to answer (although to be honest, Matías’ answers left me even more confused). Derek’s Github comments get lots of 👍 and ❤️ support. Even so, nothing changes.

Derek’s most important question is this one:

You have to research what the users want, not what the team wants to build. Where is your research?

Which brings me to…

Where do we go from here?

Alright, now we start to get into the territory of my opinion. Of course, you’re entitled to agree or disagree – take it up with me on Twitter if you like.

So, in my opinion, the entire experiment should be reverted.

The perception is that Automattic is working on Block Template themes in the shadows. Why? Do they see themselves as pioneering a future that we don’t yet know we need?

The rest of us need to be looped in to the decision making process.

Reword the document in this PR as a proposal, and post it on Make WordPress for feedback. The first Make WordPress post from Matías contained very few technical details.

Allow time for discussion! Put it on the agenda for the Editor Chats that happen every week.

Help the community to understand what this looks like – because clearly nobody does right now.

Most importantly, research how Users, Agencies, and Theme Designers want to engage with themes in WordPress. Make a decision based on that research, not based on a handful developers best guesses at how a block based themes should be implemented technically.

Merely asking for feedback is just not good enough. If nobody responds, that doesn’t mean you get to decide for yourself. That’s not how research works.

Maybe a developer tool isn’t the best place to ask for feedback on the future of WordPress. We can’t expect our community to learn GitHub, and follow the discussions happening across nearly 20,000 issues.

Maybe, if we facilitate a friendly, open, and wide-reaching dialog, we might arrive at different conclusions.

---

Important note: Nobody involved here is being intentionally malicious or hostile. I believe we have a problem with structure and power dynamics, which is what I’m trying to call attention to (not the individual developers).

I reached out to Matías and Riad and get their feedback before publishing this article.

Riad also offered these thoughts:

The reality is that I’d love more contributors that are non-Automatticians, but the reality is that a big part of WordPress contributors today are Automatticians. If you’d like to join our team and work with us to improve the proposal or change it you’re welcome.

There’s no shadow here . This was discussed for more than a year now in different Github issues by different people and regularly on #core-editor weekly meetings (which I invite you to join to discuss the proposal and help).

In chatting with Matías, he noted that the proposal for block templates has been discussed for a really long time in multiple places — there are GitHub issues from all the way back to 2017.

---

1. Matt Mullenweg, CEO of Automattic and co-founder of WordPress, is known as the “Benevolent Dictator for Life” of the WordPress project.

2. Actually, the experiment itself was merged a while ago – what we’re talking about here is only the documentation about the experiment. To be aware that any of this was happening, you’d have to be following along on GitHub.

The reason this PR was noticed was because the part that got merged was Documentation – thankfully a bit easier to understand than trying to string together multiple conversations between PRs, issues, and code. Even as it was, though, this could have easily gone under the radar if it weren’t for Justin picking it up on the Tavern.

3. There was also a call for feedback published on the Make WordPress blog, on November 3. I didn’t include it in the timeline because block based themes are just one line item among many – easily lost on a Make post that seems to have gone mostly unnoticed.

I just learned about a proposal from Jonathan Bossenger to implement better notifications in WordPress.

You can read up on it for yourself, but I thought it would be helpful (for myself as much as anyone) to add some context, and explain why it’s so important.

• Make WordPress Feature Proposal
• WP Notify Requirements Documentation

What would WP Notify do?

If added to core, WP Notify would:

1. Provide developers with a Notifications API to surface new Notifications to the user, and handle their state (e.g. “dismissed”)

2. Introduce a Unified Notification Format (fixing the current visual mess of having boxes everywhere)

3. Group notifications together in a Notifications Hub (such as a menu item in the Admin Bar)¹

---

Technical Part.

The current notifications in WordPress are a hack at best. You need to write all the HTML for the notification wrapper, and use a specific CSS class to get it to show up.

Something like:

add_action( 'admin_notices', function() {
	<div class="notice is-dismissible">
		<h2><?php esc_html_e( 'We come in peace.', 'my-plugin' ); ?></h2>
		<p>🖖 <?php esc_html_e( 'Live long and prosper.', 'my-plugin' ); ?></p>
	</div>
} );

That’s just to show the notice. You’ve also got to save whether the notice has been dismissed.

---

I’ve got some personal history tied up in WordPress notifications. Joshua Wold and I spent a lot of time designing a notifications solution for Stream (which never came to be).

Then, in 2016, I gave a talk at WordCamp Sydney containing this screenshot.

What we’re looking at here is a brand new WordPress install after installing just 5 plugins. It’s a mess!

Now, a common Notifications API in WordPress won’t fix this entirely, but it could improve it a lot. The most interesting idea is to group notifications together in the Admin Bar.

This could produce a UI similar to the way Yoast handles notifications (see the screenshot above). It would also restrict notifications to:

• A limited amount of text (280 characters, for example)
• An icon

---

Notifications in WordPress can be grouped into a few categories:

Action

The plugin wants the user to take an action (often required before the plugin can be fully functional). For example, adding an API Key to Akisment, or connecting Jetpack. Sometimes it’s an optional action, such as taking a survey.

Onboarding

The plugin wants to teach the user about a feature, and so it shows a notice (usually contextual to where the feature exists) with some help text.

Informative

The plugin has detected information about WordPress which it wants to surface to the user. For example, Yoast detecting that search engine indexing is disallowed.

Results

The user has taken an action, and the plugin wants to notify them about the result (e.g. Post Saved, or Error Saving Post).

Advertisements

The plugin wants to show the user information about a paid upgrade or a current sale (please, no!).

---

While grouping notifications together into a single Notifications Hub would help reduce a lot of visual clutter, not all of these notification types belong in a Notifications Hub.

Hub	Notification Type
✅	Action
❌	Onboarding
✅	Informative
❌	Results
❌	Advertisements

Onboarding notifications nearly always need to be shown in context. These are typically inserted into the UI, intentionally interrupting the user journey.

Results are a little different to other notifications in that they’re not usually persistent. They only need to be shown once, and automatically dismissed. Gutenberg’s notifications implementation does this really nicely.

Advertisements could exist in a Notifications Hub, but probably won’t. These notifications are all about grabbing a user’s attention, and grouping with other notifications isn’t a great way of doing that.

---

Given that only a few types of Notifications that would make use of a Hub (and even then, only at the plugin developers discretion), is WP Notify worth pursuing?

To answer this, I installed the 20 most popular plugins and catalogued the notification types I encountered. By collecting this data, we can see what percentage of notifications would be “captured” in a Notifications Hub area.

Notification Type	Count	%
✅ Action	10	33%
❌ Onboarding	6	20%
✅ Informative	9	30%
❌ Results	2	7%
❌ Advertisements	3	10%

Of all notifications that I encountered, 19 / 30 (63%) could belong in a Notifications Hub.²

---

One interesting side note: Due to the brief usage of each plugin, I’m sure I missed many “Results” notifications, such as notifications shown when saving settings.

Should WP Notify provide a method for developers to show this style of transient notification, separately to the “grouped” Notifications Hub?

---

If you’d like to be involved in making this proposal a reality, you can get involved by joining the #feature-notifications channel in the WordPress Slack.

This is a great opportunity to get involved in WordPress core development, especially for UX / UI designers who want to have a hugely positive impact toward a distributed and open web.

---

1. There is some concern that the admin bar might not be the best place to show notifications. It’s already crowded, hard to use on mobile, and sometimes hidden on the frontend.

2. In these tests, the average amount of notifications displayed (after only a few minutes of usage) is 1.5 notifications per plugin.

Yoast’s Black Friday Fumble

It’s Thursday afternoon. I’m having a quiet beer with Rob (cofounder of Block Lab) and Michael (founder of Visser Labs). We’re chatting about our WordPress plugins.

We all laugh.

Imagine our surprise when the next morning, that’s exactly what Yoast, and some others, did in their plugins.

The Fallout

The reaction from their users, and the WordPress community in general, was swift (and pretty nasty!).

Are you kidding me @yoast? An ad on literally every page of the WordPress dashboard? pic.twitter.com/rJDcHyt33H

— Colin Newcomer (@ColinNewcomer) November 28, 2019

Shitty practices from plugin authors like this Dashboard ad from @yoast, will ensure that I never spend money in your store. This sort of crap shouldn’t be rewarded! pic.twitter.com/rBmMHn69u4

— Anthony Hortin ︽✵︽ (@maddisondesigns) November 29, 2019

Many others took to WordPress.org to express their outrage, leaving Yoast SEO 75+ 1 star reviews over 2 days.

In the end, Marieke van de Rakt (CEO at Yoast) set things right. She issued an apology, and Yoast released an update removing the banner ad.

That BlackFridayBanner was not the best idea. We’re truly sorry for the annoyance and difficulties it may have caused. We did not think this through properly. If you want, you can update to a new version of our plugin without that banner. #blackfriday #neveragain #apologies

— Marieke van de Rakt (@MariekeRakt) November 28, 2019

The question has to be asked: Who thought this was a good idea? And who approved it? A business the size of Yoast would have likely had a process something like this:

1. A Marketer suggested the idea
2. A Product Manager approved it
3. A Business Analyst defined how it should work
4. A Designer designed it (questionable in this case)
5. A Developer implemented it
   • In this case, it was actually developed twice
6. QA reviewed it

That’s six different people who all just went along with an objectively awful idea. Was there any internal conflict or discussion? We’ll never know.

I don’t think there’s any long-term repercussions here for Yoast. They made a mistake, they fixed it, they promised not to do it again. But with such a violent reaction from their users, it’s worth diving into why this hit a nerve.

Is This Allowed?

The free version of Yoast is run on 15% of the top 1 million websites. Not 15% of WordPress – 15% of the web. If the success of this ad outweighs the negative reaction, is there anything stopping Yoast from doing this again?

From the WordPress.org plugin guidelines:

Advertising within the WordPress dashboard should be avoided, as it is generally ineffective. Users normally only visit settings pages when they’re trying to solve a problem. Making it harder to use a plugin does not generally encourage a good review, and we recommend limiting any ads placed therein.

Detailed Plugin Guidelines

So… not recommended, but not disallowed. Yoast couldn’t be delisted from WordPress.org as a result of ads.

Yoast’s Black-Friday-Faux-Pas is an exercise in extremes. Their bright yellow and cyan full-width banner, animated like it’s 1999, was displayed on every – single – page. Freemium plugins including a small up-sell ad on their settings page don’t cause this kind of uproar.

Is there a place for ads in WordPress? Maybe there is. Certainly a sustainable plugin ecosystem is good for everyone, and the Freemium model seems to be an effective one.

Ads Done Well

Of course, the offensiveness of advertising prominence will always be subjective. Here’s a few plugins that I believe do a great job of letting their free users know about a premium option.

Sidebar Ads

Advanced Custom Fields – everybody’s favourite custom meta plugin takes a very understated approach. You can’t even really call this one an ad, it’s so subtle (maybe a little too subtle?).

WP Migrate DB has a Pro ad in the sidebar. It only shows when you’re using the migration tool. Importantly, it outlines precisely how Pro improves on the Free version.

WP Migrate DB uses a formula that keeps things simple and unobtrusive: Call to Action > Value Statement > Social Proof. Mix and match to taste.

Sub-Menu Ads

Some plugins add an item to the menu. In these cases, a sub-menu item showing the availability of Pro might be acceptable.

The advantage here is that visually there’s no “Ad” in the plugin (similarly to ACF above), but information about Pro is available for users who might be interested.

Ads Done Poorly

Contextual Ads

Contextual ads sound like a good idea in theory, but in practice can be borderline annoying. This is where you show users the features they could be using, if only they upgrade to Pro!

I imagine that it is possible for this type of advertising to be done tastefully, but it’s hard to find an example that doesn’t leave me feeling aggravated.

Yoast is famous for this brand of advertising. Who wants to use a tool that’s littered with bright yellow Calls to Action?

Onboarding Ads

This is the worst. I haven’t even had a chance to try your product yet – at least let me see what it’s all about before hitting me with an upgrade prompt!

• 

• 

• 

• 

WooCommerce literally tries to have me install 7 additional plugins during the onboarding process! Some of them may be free, but all of them lead to paid services. I haven’t even seen WooCommerce yet.

This is extremely heavy-handed, and in my opinion, a textbook example of advertising gone wrong.

But even Woo is not quite as bad as WordFence, which, upon activation, creates a takeover model which requires an email address, and then promotes Pro! Again, before anyone has had a chance to use their plugin.

Advertising Guidelines for the WordPress Admin

There’s clearly a right way and a wrong way to advertise in the WordPress admin. If you, like Yoast, can’t figure out what counts as acceptable, here’s a handy guide:

1. Don’t advertise after plugin activation

Even if you’ve chosen to avoid the full-screen modal for your onboarding (👍), let your users at least interact with your product before hitting them with a Pro prompt.

2. Only advertise on pages created by your plugin

This includes post meta boxes, the Dashboard, and the Plugins list page.

3. No animations

I mean, that really should go without saying.

4. No teasers

“More features available in Pro” is fine.

Feature X is only available in Pro. Upgrade Now!
Feature Y is only available in Pro. Upgrade Now!
Feature Z is only available in Pro. Upgrade Now!

… is not fine.

5. Say it once

Don’t include more than one ad per page.

6. Advertise the product, not the sale

Introducing ads for limited time events is tacky. It’s also code bloat and will last forever in your version control.