The early web was not designed as a collection of private estates. It was designed as a commons of documents that anyone could publish, link, copy, and leave. You did not need permission from a platform to start a site. You did not need a vendor’s blessing to move it. HTML was readable. URLs were portable. If a host failed or a company changed its terms, the content could travel.
That architecture is no longer the default experience of the internet. A large and growing share of human expression now lives inside systems that are easy to enter and hard to leave. Social networks, website builders, design tools, newsletter platforms, course platforms, and “all-in-one” business suites have become the places where people write, photograph, sell, teach, organize, and remember. The interfaces are polished. The onboarding is frictionless. The exit is not.
This is the problem the WordPress Data Liberation project exists to attack. Its premise is simple and radical: your content should be yours in practice, not merely in a terms-of-service clause. You should be able to reclaim it, reshape it, and move it anywhere on the web—without rebuilding years of work by hand, and without needing a new hosting account just to begin. The project’s own slogan is the whole argument in four words: “Your data, your move.”
That sounds like a migration initiative. It is also something larger. If the internet is going to remain an open medium rather than a set of rented rooms, someone has to build the exits. Data liberation is one of the few serious, public, engineering-scale attempts to do that.
The Quiet Capture of the Publishing Layer
The internet’s infrastructure is still comparatively open. TCP/IP, DNS, HTTP, and HTML remain public standards. What closed first was the application layer—the place where ordinary people actually create.
A walled garden is not merely a product with a login. It is a system in which identity, content, distribution, design, analytics, payments, and social graph are bound together so tightly that leaving means abandoning some or all of them. The wall is rarely announced on day one. It appears later, when a user tries to take a decade of posts to another home, reconstruct a store catalog, preserve URL structure for search engines, or keep the visual identity they spent years refining.
This pattern is now ordinary. Many popular website builders do not offer a complete, faithful export of a working site. Wix has long been cited as a platform with no full site export: pages, design, and structure live in a proprietary format on Wix’s servers. Squarespace will export some content as WordPress-compatible XML, but not the design, templates, CSS, store pages, or many block types. Webflow can export code on paid plans, with important exceptions around CMS content and platform features. Framer and similar tools often have no meaningful site export at all. The published page may look like the open web. The source of truth does not.
Social platforms are more extreme. Years of photographs, comments, DMs, and audience relationships sit behind interfaces designed for consumption inside the product. Official “download your data” archives, when they exist, are often incomplete, poorly structured, or useless as a living site. The content is technically “yours.” It is not operationally portable.
The economic logic is obvious. Switching costs protect revenue. If leaving requires weeks of manual copying, broken links, lost rankings, and a visual rebuild, most people stay. Lock-in can look like loyalty. It is often just exhaustion.
That is why data liberation is not a niche developer concern. It is a question of who owns the cultural and commercial memory of the web.
What WordPress Means by Liberation
WordPress announced Data Liberation as a focused initiative in early 2024. Matt Mullenweg framed it as a return to an older promise: a web where people can switch platforms, and where being locked into a system becomes a thing of the past. Migrating a site to WordPress, or exporting everything out of WordPress, should eventually be possible in one click. The WordPress export format, in this vision, should become a kind of lingua franca among content systems—not because WordPress deserves a monopoly, but because an open, widely understood interchange format is how lock-in dies.
The project is organized as five phases, published on wordpress.org/data-liberation:
Phase 1: Migration guides. The first move was unglamorous and necessary. Instead of waiting for perfect software, the community began collecting step-by-step paths off platforms such as Wix, Figma-based workflows, and static HTML. The point was to stop forcing every migrant to invent the same painful process alone.
Phase 2: Importing and exporting structured data. This is the hard engineering layer. Real migrations fail on details: rewritten URLs, missing media, partial imports, huge sites, malformed markup, and the gap between “a page exists” and “the page is editable as structured content.” Phase 2 treats those as first-class problems. The work includes streaming parsers, robust importers, error recovery, and conversion of HTML into WordPress blocks so the result is not a frozen snapshot but living content.
Phase 3: Liberating data from closed platforms. When a platform will not give you a complete export, the project’s answer is not to accept the wall. Phase 3 is a browser extension and related tooling that can pull content from websites that never intended to let it go—social media, closed page builders, design tools. The same idea can double as a personal scratchboard for gathering material from across the web.
Phase 4: Direct WordPress-to-WordPress synchronization. Liberation is not only escape from other vendors. It is also the ability to mirror, back up, clone, and move WordPress sites themselves—across devices, hosts, and even WordPress Playground, the in-browser runtime that can spin up a site without a traditional hosting account.
Phase 5: A content creation powerhouse. The end state is more ambitious than “import from Wix.” WordPress becomes a local-first hub for notes, long-form work, social posts, and books: content stored under the user’s control and synced across devices. The CMS stops being a destination you rent and becomes infrastructure you carry.
The internal principles that emerged from community outreach are equally important. Data liberation starts as practical tooling, not a manifesto. WordPress Playground is treated as a way to preview a migrated site before committing. Transforming messy HTML into blocks is the technical path that serves almost every source platform. And the primary goal is content portability, not pixel-perfect recreation of a proprietary theme. Visual parity is a bonus. Ownership of the words, media, structure, and links is the point.
This is not charity for WordPress’s market share, though WordPress remains the largest CMS in the world, powering roughly two-fifths of all websites and a majority of sites that use a known CMS. It is an attempt to use that installed base as a port of entry back onto the open web.
Why This Ranks Among the Internet’s Most Important Projects
There are many important internet projects: encryption, identity, protocol governance, accessibility, spam defense, AI safety, and the physical build-out of networks. Data liberation belongs in that company because it addresses a failure mode that those other projects cannot fix by themselves. You can have HTTPS, a domain, and a free-speech norm and still lose your work because it is trapped in a format no other system can read.
1. Ownership without an exit is a slogan
Legal ownership of “your content” is nearly worthless if the only usable copy lives in a proprietary renderer. A photograph you uploaded, a product description you wrote, a course you built, a community archive you moderated—these are not abstract rights. They are files, URLs, relationships, and presentation. When a platform withholds the structured form of that work, it has converted your labor into an asset that only it can fully exploit.
Data liberation treats ownership as a technical property: can you leave with something you can keep using? If the answer is no, the platform is not a tool. It is a landlord.
2. Lock-in is a tax on the entire web’s evolution
Markets are supposed to discipline bad products. Users leave. Competitors improve. That mechanism breaks when the cost of leaving exceeds the pain of staying. Then mediocre products persist, prices drift upward, innovation slows, and incumbents spend engineering effort on walls rather than value.
This is an old observation in software. Making it easy for users to leave is, paradoxically, one of the few durable ways to earn trust and force a company to compete on merit. Data liberation is the public-goods version of that idea. If exits become cheap, every platform has to keep deserving its users.
That matters far beyond CMS market share. Website builders, social networks, newsletter tools, and AI writing environments all benefit from the same asymmetry: they ingest open-web content easily and emit it poorly. An effective liberation stack reverses the asymmetry.
3. Cultural memory should not depend on a vendor’s roadmap
The web is now the default archive of public life: local journalism, personal blogs, small-business histories, activist documentation, research notes, recipes, sermons, fan communities, and government-adjacent civic sites. When those artifacts are stored only inside a builder or a social silo, they inherit that company’s mortality.
Companies change pricing. They sunset products. They get acquired. They alter acceptable-use rules. They degrade export tools. They disappear. The open web’s great advantage was that a site could outlive its original host. Walled gardens invert that. The work dies when the account dies, or survives only as a degraded zip file.
A serious data-liberation project is therefore a conservation project. It is about whether the next fifty years of human publishing will be inspectable, movable, and forkable, or whether it will exist at the pleasure of a handful of product teams.
4. Publishing power is still one of the internet’s equalizing forces
WordPress’s historical importance was not that it was the most elegant editor. It was that it gave millions of people a durable publishing stack they could own: a domain, posts, pages, media, plugins, and the right to change hosts. That is a political fact as much as a technical one. Independent media, niche expertise, and small commerce do not scale if every publisher must rent a closed canvas and accept the platform’s ranking logic as destiny.
Closed builders lowered the barrier to starting and raised the barrier to leaving. That combination produces a web that looks participatory and is structurally dependent. Data liberation tries to keep the low starting cost while restoring the old right of exit.
5. The AI era makes trapped data more valuable—and more dangerous
Large models are trained on the open web and then used to generate more content that platforms want to keep. Recommendation systems prefer material that never leaves the garden. Agents can now browse, scrape, rewrite, and republish at industrial scale. In that environment, the question of who can extract, transform, and re-home content is no longer a migration footnote. It is a question of bargaining power.
If only platforms can fully use the corpus users created inside them, then users become raw material. If users can extract structured content, they can choose new tools, mix sources, run local models against their own archives, or republish under rules they control. Liberation tooling is one of the few practical counters to a future in which AI systems can enter any garden while humans cannot leave with their own work intact.
There is already a research argument that AI agents will puncture walled gardens by translating between interfaces that were never meant to interoperate. That may happen. It is not a substitute for deliberate, user-controlled portability. An agent that can scrape your site for a vendor is not the same as a format you can keep.
6. Law created a right. Software has to make it real.
The European Union’s GDPR established a right to data portability. The Digital Markets Act goes further for designated gatekeepers: portability should be free, effective, and in important cases continuous and real-time, including to authorized third parties. Those rules matter. They also demonstrate the limits of rights without implementation.
GDPR-style takeout files are often dumps, not migrations. They satisfy a compliance checklist without producing a living website, a usable social graph, or a faithful editorial archive. The DMA’s stronger requirements are aimed at the largest platforms and will be fought over in APIs, consent flows, and “quality of data” disputes for years.
A community project that builds importers, HTML-to-block converters, browser extractors, and one-click previews does something regulation cannot do alone. It creates the receiving infrastructure. Rights to leave are empty if there is no good place to land and no tool that understands the mess you bring with you.
7. Open source is the only credible steward for this work
A single company can offer a better export button. It cannot be trusted, by itself, to finish the job. Export is against the short-term interest of almost every closed platform. Even open-adjacent companies will prioritize incoming migrations over outgoing ones.
WordPress is not a perfect steward. It is a large, contested ecosystem with its own commercial gravity. But it has properties that make it unusually suited to this problem: a public codebase, an established interchange format (WXR), a huge destination market, a culture of plugins, and now an in-browser runtime that can demonstrate a migrated site before anyone pays a host. The project’s explicit aim is community-owned guides and tools rather than a pile of duplicated agency scripts.
If this work is not done in the open, it will be done as expensive professional services. That would liberate the clients who can pay and leave everyone else inside the garden.
What Liberation Has to Solve That Slogans Ignore
A serious article should not pretend this is easy.
Faithful export is not the same as useful import. Getting HTML off a page is the beginning. Making that HTML into editable blocks, with media downloaded, internal links rewritten, slugs preserved, and metadata intact, is the project. Closed platforms render in JavaScript, lazy-load media, hide content behind sessions, and mix design with content so thoroughly that a scrape can look complete and still be structurally empty.
Design is the hostage. Users think they own “the site.” They usually own the text and images. The layout system, animations, forms, bookings, and theme logic stay behind. Data Liberation’s own principle—that content portability comes first—is honest. It is also a hard sell to someone who will lose a painstaking visual identity the day they leave.
WordPress has lock-in of its own. Page builders, page-builder HTML, plugin-specific data, and custom tables can trap users inside WordPress as surely as Wix traps them outside it. Several people in the WordPress community made this point as soon as the initiative was announced: liberation has to happen within WordPress too, from classic editor to blocks, from proprietary builders to core, from plugin silos to standard data. If the project only makes it easy to enter WordPress and not to move around inside it or leave it, it will be accused—fairly—of being a growth program dressed as ethics.
Progress has been slower than the rhetoric. That is not surprising. The announcement created a destination page and a moral frame. The work is parsers, edge cases, platform-specific quirks, legal gray areas around scraping logged-in content, and the unending boredom of migration guides. Important internet projects often look underwhelming in their third year because the remaining work is all residue.
Extraction can collide with terms of service, privacy, and other people’s data. A browser extension that “liberates data from any website” is powerful. It can also pull comments by other users, private resources, or material the source platform has a legitimate reason to gate. Portability for the account holder is not a blank check to vacuum a network.
These constraints do not shrink the importance of the project. They explain why it has to be treated as infrastructure, not a campaign.
A Longer Lineage Than a 2024 Brand Name
WordPress did not invent the underlying idea. Tim Berners-Lee has spent years warning that the web was being broken into fragmented islands and has backed Solid, a model in which personal data lives in user-controlled pods and applications come to the data rather than the reverse. Decentralized social protocols, RSS, ActivityPub, the IndieWeb’s POSSE pattern (“publish on your own site, syndicate elsewhere”), and even the humble XML sitemap are all attempts to keep the unit of value on the open web.
What Data Liberation adds is not a new philosophy. It is an implementation path aimed at the actual places people are stuck right now: Wix sites, Figma files treated as websites, social profiles, page-builder lock-in, and the grinding last mile of “I have an export and it does not work.”
That practicality is why the project punches above its current code volume. The internet does not lack essays about openness. It lacks boring tools that turn a trapped site into a portable one on a Tuesday afternoon.
What Success Would Change
If Data Liberation works, several things follow.
Switching costs fall. A small business can leave a builder without a five-figure rebuild. A writer can take a decade of posts to a new stack without ritual humiliation. A community can fork itself when a platform becomes hostile.
Competition rises. Closed platforms would still win on convenience, design, and distribution. They would have to win on those terms, not on the impossibility of exit.
The unit of the web becomes the site and the document again, rather than the account. WordPress Playground already hints at a world where a site is a package you can run in a browser, clone, snapshot, and move, instead of a rental on a particular host.
New publishing patterns become normal. A person could draft locally, publish to their own domain, syndicate to networks, and bring responses back. Social platforms would become distribution channels rather than the canonical store of record.
The AI stack would have a user-side corpus. Instead of asking a platform for permission to use your own archive, you would already possess it in structured form.
None of that requires WordPress to “win” every future CMS comparison. It requires that some widely deployed, open system act as a solvent. WordPress is the solvent that already exists at planetary scale.
The Stakes, Stated Directly
The internet’s first generation of institutions assumed that content wanted to be free in the old sense: copyable, linkable, movable. The second generation of consumer platforms discovered that content is more profitable when it is free to enter and expensive to extract. We are now in a third phase, in which AI systems intensify the value of corpora and the incentive to enclose them.
Against that backdrop, a public project whose job is to pick the locks—legally, technically, and socially—is not a side quest for CMS enthusiasts. It is maintenance of the web’s original political settlement: that a person can publish under their own name, on their own site, and leave without being erased.
Data liberation will not, by itself, fix concentration in app stores, advertising markets, or model training. It addresses a more intimate failure. It says that the words and images people make should not become a hostage. If that principle fails, the open internet remains a substrate while the human layer moves permanently indoors.
That is why the unglamorous work on wordpress.org/data-liberation—guides, importers, parsers, preview environments, and extractors—deserves to be treated as one of the singular projects of the contemporary internet. Not because it is complete. Because the alternative is a web where everything you make is easy to start and almost impossible to keep.
References
WordPress.org. “Data Liberation.” https://wordpress.org/data-liberation/
WordPress.org. “Guides.” https://wordpress.org/data-liberation/guides/
Mullenweg, Matt. “Data Liberation in 2024.” WordPress News, January 12, 2024. https://wordpress.org/news/2024/01/data-liberation-in-2024/
WordPress/data-liberation. GitHub repository. https://github.com/WordPress/data-liberation
WordPress/data-liberation. “Locked vs Liberated data.” Discussion #82. https://github.com/WordPress/data-liberation/discussions/82
jordesign. “Data Liberation Next Steps.” Make WordPress Core, February 19, 2024. https://make.wordpress.org/core/2024/02/19/data-liberation-next-steps/
jordesign. “Data Liberation: Outreach recap and updates.” Make WordPress Core, May 3, 2024. https://make.wordpress.org/core/2024/05/03/data-liberation-outreach-recap-and-updates/
jordesign. “Data Liberation: Project plan.” Make WordPress Core, May 3, 2024. https://make.wordpress.org/core/2024/05/03/data-liberation-project-plan/
WordPress/try-wordpress. GitHub repository. https://github.com/WordPress/try-wordpress/
Zieliński, Adam. “Using Playground for data liberation, site synchronization, and building streaming parsers.” Make WordPress Playground, November 6, 2024. https://make.wordpress.org/playground/2024/11/06/using-playground-for-data-liberation-site-synchronization-and-building-streaming-parsers/
Mullenweg, Matt. “WordPress Everywhere.” ma.tt, March 2026. https://ma.tt/2026/03/wordpress-everywhere/
Karkovack, Eric. “Data Liberation Needs to Happen within WordPress, Too.” The WP Minute, 2024. https://thewpminute.com/data-liberation-needs-to-happen-within-wordpress-too/
W3Techs. “Usage Statistics and Market Share of WordPress, August 2026.” https://w3techs.com/technologies/details/cm-wordpress
W3Techs. “Market share trends for content management systems, August 2026.” https://w3techs.com/technologies/history_overview/content_management
Squarespace Help Center. “Exporting your site.” https://support.squarespace.com/hc/en-us/articles/206566687-Exporting-your-site
Stacktree. “Can you export a site from a website builder? Mostly no.” August 16, 2026. https://stacktr.ee/website-builder-migration
Chillybin. “The Real Downsides of Using Wix for Your Business Website.” https://www.chillybin.co/downsides-of-using-wix-website-builder/
Forged Digital Design. “The Hidden Cost of Squarespace and Wix.” May 2026. https://forgeddigitaldesign.com/blog/hidden-cost-squarespace-wix-platform-lockin/
European Commission. “End user data portability — Digital Markets Act (DMA).” https://digital-markets-act.ec.europa.eu/developer-portal/end-user-data-portability_en
Tech Policy Press. “Can the Digital Markets Act Free Users’ Data in the AI Age?” December 2025. https://techpolicy.press/can-the-digital-markets-act-free-users-data-in-the-ai-age
Berners-Lee, Tim. “Let’s knock down social media’s walled gardens.” Financial Times (archived discussion of Solid and user-controlled data). https://www.ft.com/content/79d2d19a-08df-48fc-9a6f-a9dbef58f642
Sanger, Larry. “Proposing a ‘Declaration of Digital Independence.’” WIRED, March 12, 2019. https://www.wired.com/story/larry-sanger-declaration-of-digital-independence/
Marro, Samuele, and Philip Torr. “LLM Agents Are the Antidote to Walled Gardens.” arXiv:2506.23978, 2025. https://arxiv.org/abs/2506.23978
ACM Queue. “The Case Against Data Lock-in.” https://queue.acm.org/ (reprint commonly circulated as “The Case Against Data Lock-in”)
Noema. “We Need To Rewild The Internet.” April 15, 2024. https://www.noemamag.com/we-need-to-rewild-the-internet