All posts

The WordPress Succession Problem

Cloudflare dropped a ‘spiritual successor to WordPress’ on April Fools’ Day. After two decades of watching platforms try to dethrone the most entrenched CMS on the internet, I have thoughts on why this one won’t work either — and what would.

Abstract visualization of CMS platform architecture and ecosystem moats

What would it take for sites to switch?

Cloudflare dropped what it calls "the spiritual successor to WordPress" on April 1st. After two decades of watching platforms try to dethrone the most entrenched CMS on the internet, I have thoughts on why this one won't work either -- and what would.


Cloudflare announced EmDash on April 1, 2026. I want you to sit with that for a second.

A company with the engineering talent and infrastructure muscle to build genuinely interesting technology chose April Fools' Day to unveil what it calls a "spiritual successor to WordPress" -- the platform that powers 43.3% of the entire internet.w3techs The Cloudflare CTO had to get on Hacker News and personally confirm that no, this was not a joke, because enough people assumed it was that the technical merits were getting buried under date-related skepticism.hn-thread

Whether the timing was a miscalculation or deliberate controversy-bait, it set the tone for everything that followed. And the tone is: this is a product that doesn't know what it wants to be yet. It's an MIT-licensed, TypeScript-based, Astro 6.0-powered CMS that runs on Cloudflare Workers, claims portability to any Node.js server, and arrives with a plugin security model that is -- genuinely -- the most interesting thing about it. It also arrives with zero plugins, zero themes, zero agency ecosystem, zero migration path beyond a WordPress XML import tool, and a 483MB install size that makes WordPress's 70MB footprint look quaint.hn-thread

I've been building and deploying platforms for over 20 years. I've watched this movie before. The script is always the same: a technically superior alternative appears, the developer community gets excited, the press writes glowing coverage, and five years later the challenger is a rounding error on a market share chart. The question isn't whether EmDash has good ideas. It does. The question is whether good ideas have ever been enough to displace WordPress. They haven't.

Let me explain why.

The 96% Problem

Here is the stat that Cloudflare leads with, and it's a good one: 96% of WordPress security vulnerabilities originate in plugins.patchstack-2025 Not in WordPress core. Not in themes. In the third-party code that site owners install because WordPress, by itself, doesn't do much. WordPress core had exactly 7 vulnerabilities in all of 2024. The plugin ecosystem had 7,633. That is not a typo.

The full picture is worse. Patchstack's 2026 report -- covering 2025 data -- documents 11,334 new vulnerabilities in the WordPress ecosystem, a 42% increase over the previous year.patchstack-2026 Of those, 1,966 had high severity scores likely to be exploited in automated mass-scale attacks. More high-severity vulnerabilities were discovered in 2025 than in the previous two years combined. Sucuri documented over 500,000 infected WordPress sites in 2024 alone -- SEO spam, malicious redirects, credential theft, the full catalog of things that happen when untrusted code runs with unrestricted access to your database and filesystem.patchstack-2025

Cloudflare's diagnosis is that the problem is a lack of sandboxing. WordPress plugins run with full access to everything -- the database, the filesystem, the network, other plugins' data, user sessions, the entire execution environment. There is no permission model. There is no isolation boundary. A free slider plugin from a developer who abandoned it three years ago has the same access to your payment data as WooCommerce does.

That diagnosis is correct. But the framing matters, because it determines whether the cure makes sense.

Is the problem that WordPress plugins are unsandboxed? Or is the problem that WordPress created a platform architecture where the only way to be useful is to give every piece of third-party code the keys to the entire kingdom? The distinction matters. Because if the problem is architectural, then a sandbox is a legitimate response. But if the problem is that useful plugins need broad access to be useful, then a sandbox doesn't solve anything -- it just moves the trust boundary from "install the plugin" to "approve the permissions."

I've seen this exact pattern in mobile app security. iOS and Android both sandbox apps. Apps are isolated from each other. And yet -- the apps that do anything useful still request access to your contacts, your location, your camera, your photos, your network, and your files. The sandbox didn't eliminate the trust problem. It formalized it. You still have to decide whether to trust the code. You just do it through a permissions dialog instead of through blind faith.

That's not nothing. But it's not what Cloudflare is selling it as, either.

The Sandbox That Doesn't Sandbox

EmDash's security architecture is built on Cloudflare's Dynamic Workers -- a new runtime primitive that gives each plugin its own V8 isolate.dynamic-workers This is technically interesting. V8 isolates are the same sandboxing technology that keeps Chrome tabs from accessing each other's memory. They spin up in milliseconds, use a few megabytes of RAM, and provide genuine memory isolation at the engine level. Cloudflare layers additional protections on top: hardware-level memory protection keys, custom Spectre defenses, and a capability-based permission system where plugins declare what they need in a manifest file -- read:content, email:send, database:write -- and the runtime enforces those boundaries.

On paper, this is a meaningful improvement over WordPress's trust-everything model. A malicious slider plugin genuinely cannot read your payment database if it didn't declare database:read in its manifest. The isolation is real. The enforcement is at the runtime level, not the application level. This isn't a polite suggestion. It's a wall.

Here's the problem: the wall has doors, and the doors have to be open for the plugin to work.

A contact form plugin needs to write to the database. A SEO plugin needs to read and modify content. An analytics plugin needs network access. An e-commerce plugin needs -- well, everything. The capability system doesn't eliminate trust decisions. It enumerates them. And the history of capability-based security systems is that users approve whatever permissions an app requests, because the alternative is the app not working.

There's a deeper issue. The full sandbox isolation -- the V8 isolate model with Dynamic Workers -- is a Cloudflare-specific technology. It runs on Cloudflare's infrastructure. EmDash is MIT-licensed and claims to be portable to any Node.js server, but self-hosted EmDash deployments do not get isolate-based plugin sandboxing.dynamic-workers The core security feature -- the one that justifies the entire project's existence -- evaporates the moment you exercise the freedom that the open-source license supposedly guarantees.

This is the central paradox, and the Hacker News community spotted it immediately.

Open Source, Closed Architecture

Let me be precise about what EmDash is. It's an MIT-licensed codebase that is architecturally optimized for -- and in its most important capability, dependent on -- a single vendor's proprietary infrastructure. The code is open. The architecture is closed.

One HN commenter called it what it is: the security model only works on Cloudflare's runtime, making it "architecturally locked in."hn-thread Another identified the broader pattern -- Cloudflare has done this before with Pages, D1, and R2: open-ish projects that are optimized for Cloudflare's stack and degraded everywhere else.hn-thread This is the "open core" play wearing an MIT license as a costume.

The comparison table tells the story plainly:

DimensionWordPressEmDash
Age24 yearsWeeks old
Market share43.3% of the web0%
Plugins70,000+~0
Themes25,000+~0
Hosting requirement$5/month shared hostingCloudflare account (for full features)
Plugin securityTrust everythingSandboxed isolates (Cloudflare only)
LicenseGPLMIT
Install size70MB483MB
DeploymentFTP / one-click installCLI / serverless deploy
Target userEveryoneDevelopers

WordPress's original genius was deployment simplicity. You uploaded files to a $5/month shared host via FTP, ran a five-minute installer, and you had a website. No server management. No restarts. No CLI. No runtime dependencies beyond PHP and MySQL, which every shared host on the planet already had. As one HN commenter put it: that FTP-to-shared-hosting simplicity is WordPress's killer feature, and EmDash doesn't have an answer for it.hn-thread

EmDash requires Node.js, a build step, and either a Cloudflare account or a server you manage yourself. The 483MB install size -- versus WordPress's 70MB -- is the node_modules problem in microcosm. None of this matters to a developer who spins up infrastructure for a living. It matters enormously to the freelancer in Manila building a restaurant website, the small business owner who just wants a blog, the nonprofit that needs a donation page by Friday. Those people are WordPress's actual user base. And EmDash has nothing to offer them.

The WordPress Moat

Here's what I know to be true after watching platform succession attempts for two decades: ecosystem network effects defeat technical superiority every time. Every. Single. Time.

The WordPress ecosystem is valued at $596.7 billion.wp-engine-study That's not a number I'm emphasizing for shock value. That's the economic reality of 70,000+ plugins, 25,000+ themes, thousands of specialized agencies, an entire hosting industry, a global freelancer market, and WooCommerce powering 28% of all online stores globally.colorlib There are more people making a living from WordPress than there are people who have heard of Cloudflare Workers.

The moat isn't technical. It's economic and social. Consider what happens when an agency with 200 WordPress clients evaluates EmDash:

  • Their team has years of WordPress expertise. Retraining costs real money.
  • Their clients have WordPress plugins that are core to their business operations.
  • Their hosting partnerships are WordPress-specific.
  • Their project templates, their workflows, their documentation, their hiring pipeline -- all WordPress.
  • EmDash has zero plugins. Zero themes. Zero community. Zero StackOverflow answers. Zero freelancers who know how to use it.

The chicken-and-egg problem is lethal. Developers won't build EmDash plugins without users. Users won't migrate to EmDash without plugins. And the window for bootstrapping an ecosystem from zero is closing, not opening, because the managed SaaS platforms -- Wix, Squarespace, Shopify -- are eating the low end of WordPress's market with offerings that require no technical expertise at all.

Let's look at the track record of "WordPress killers":

PlatformLaunchedCurrent CMS Market ShareWhat Happened
Ghost20130.1%Niche publishing tool. $305K Kickstarter. 13 years later: a rounding error.ghost
Strapi2015<0.5%Developer-only headless CMS. No visual editor for non-technical users.
Contentful2013~1% (headless segment)Enterprise pricing ($300+/mo). Different market entirely.
Sanity2017<0.5%Requires custom frontend. Steep learning curve. Developer-only.
EmDash20260%You are here.

Ghost is the most instructive case. It launched as an explicit WordPress alternative, raised money on Kickstarter from people who wanted exactly that, shipped a technically elegant product, grew at 15% annually -- and after 13 years, holds 0.1% of the CMS market. Joost de Valk -- the founder of Yoast SEO, the most popular WordPress plugin ever built -- looked at this history and said the quiet part out loud: Ghost and Craft are "technically excellent," but none of them have built the ecosystem needed for broad adoption.hn-thread

One HN commenter captured it perfectly: the valuable IP is the moat, but "that's not an engineering problem."hn-thread You cannot engineer your way past a $596.7 billion ecosystem. You cannot sandbox your way past 70,000 plugins. You cannot MIT-license your way past the fact that EmDash developers don't exist yet and WordPress developers are everywhere.

The Vibe-Coded CMS

Here's where it gets interesting -- and where this story connects to something I've been thinking about for a while.

Multiple HN commenters noted that EmDash appears to be substantially AI-generated code.hn-thread The timeline supports this: Cloudflare reportedly built EmDash in approximately two months, a pace that's only plausible with heavy AI assistance. One commenter objected to Cloudflare "using an LLM to build something and try to pass it off as something built from the ground up." Another was blunter: this is vibe-coded infrastructure.

I wrote about the concept of vibe coding as a quality signal back in February -- the idea that benchmarks don't capture what actually matters about a tool, and that the felt experience of working with something is a more honest measurement than any score on a leaderboard. I stand by that. But the concept cuts both ways.

If the vibe of a tool matters -- if the felt experience of using it tells you more than its spec sheet -- then the vibe of EmDash is a product that was generated fast, shipped fast, and announced on a day that made people question whether it was real. The vibe is a 483MB install competing with a 70MB install. The vibe is "open source" software whose headline feature only works on one vendor's platform. The vibe is a two-month-old codebase claiming to be the successor to a 24-year-old ecosystem.

There's an irony here that I find hard to ignore. We're in an era where AI-generated code is flooding the ecosystem -- where 45% of AI-generated code introduces security vulnerabilities, according to Veracode -- and the response is to use AI to generate a platform whose entire selling point is security. The tool being used to solve the security problem is the same class of tool that's creating security problems elsewhere. That doesn't mean EmDash's code is insecure. But it means the "built in two months with AI" narrative undermines rather than supports the confidence you'd want in a security-focused platform.

If the vibe is the benchmark, the vibe of EmDash is: interesting ideas, wrong vehicle, unfortunate timing.

What Would Actually Work

So if EmDash won't displace WordPress -- and it won't -- what would?

The answer isn't a better CMS. It's never been a better CMS. Every challenger for two decades has been a better CMS in at least some dimension, and none of them have moved the needle. Ghost is better for publishing. Strapi is better for APIs. Sanity is better for structured content. They're all sitting below 1% market share. Technical superiority, applied to the same problem category, does not overcome ecosystem lock-in. I've watched this play out across enough technology cycles to call it a law.

WordPress won't be displaced by something that does what WordPress does, but better. It will be displaced by something that makes what WordPress does irrelevant.

Think about how smartphones didn't fix feature phones. They didn't ship a better T9 keyboard or a higher-resolution 2-inch screen. They changed what a phone was. The feature phone's capabilities didn't get outperformed. They got subsumed into a device that operated on a different paradigm entirely. Feature phone makers couldn't compete by making better feature phones, because the category itself stopped mattering.

The paradigm shift for content on the web isn't a new CMS. It might be AI agents that generate, deploy, and maintain web presences without a CMS layer at all -- where "building a website" becomes a conversation, not a software installation. It might be a world where content is stored as structured data and rendered on-demand by whatever interface is consuming it, making the CMS-as-rendering-engine model obsolete. It might be something I can't predict, because paradigm shifts are by definition unpredictable from inside the current paradigm.

What I know is this: the WordPress succession problem isn't a technology problem. It's a category problem. EmDash is a very good answer to the question "what would WordPress look like if we designed it today?" But that's the wrong question. The right question is: "what makes WordPress unnecessary?" And that's not a question a CMS can answer.

Cloudflare has the infrastructure, the engineering talent, and the distribution to build something genuinely paradigm-shifting. Instead, they built a CMS. A good one, with real ideas about security, but a CMS nonetheless. And on April 1, 2026 -- whether they meant it as a joke or not -- the timing told the truth that the technology couldn't: this is a product out of time, solving yesterday's problem with yesterday's architecture, wearing a new vendor's logo.

The next WordPress won't call itself a WordPress successor. It won't be a CMS at all. And when it arrives, we probably won't recognize it -- because the whole point of a paradigm shift is that it makes the old categories unrecognizable.

I'm still waiting.

References

  • w3techsW3Techs. WordPress usage statistics, 2025-2026. WordPress powers 43.3% of all websites. W3Techs.
  • patchstack-2025Patchstack. "State of WordPress Security in 2025." 96% of vulnerabilities originate in plugins; 7,966 total vulnerabilities in 2024. Patchstack.
  • patchstack-2026Patchstack. "State of WordPress Security in 2026." 11,334 vulnerabilities in 2025; 42% increase over 2024; 1,966 high-severity. Patchstack.
  • dynamic-workersCloudflare. "Dynamic Workers." V8 isolate-based plugin sandboxing; isolate startup in milliseconds. Self-hosted EmDash deployments do not receive isolate-based sandboxing. Cloudflare Blog.
  • hn-threadHacker News discussion on EmDash announcement. 364 upvotes, 261 comments. Themes: vendor lock-in, April Fools suspicion, AI-generated code criticism, ecosystem moat analysis. Hacker News.
  • wp-engine-studyWP Engine. "The Value of WordPress: World's First Study of the WordPress Economy." WordPress economy valued at $596.7 billion. WP Engine.
  • colorlibColorlib. "WordPress Statistics 2025-2026." 70,000+ plugins; WooCommerce powers 28% of online stores globally. Colorlib.
  • ghostW3Techs. Ghost CMS usage statistics. 0.1% of CMS-using websites after 13 years. W3Techs.