2026-07-12 · 8 min read · Raw Drafts / Open Web / Taiwan
The Island Inside Other People's Gardens
Taiwan built an open digital democracy while much of its everyday public life moved into privately governed platforms.
For about twenty years, I assumed websites were dying.
It seemed like a safe prediction. Blogs moved to Facebook. Photographers moved to Instagram. Restaurants stopped updating their websites and posted menus as Stories instead. Small businesses replaced contact forms with LINE accounts. Publishers became dependent on social feeds, while creators learned to introduce themselves through a row of platform icons.
The website remained, but increasingly as a neglected page containing an address, a logo, and links to the places where the organisation's real life was happening.
People did not abandon the open web because they rejected openness. They left because the alternatives worked better. Instagram offered publishing, distribution, identity, analytics, messaging, and an audience in one application. An independent website offered freedom, followed by the responsibility to assemble everything yourself.
The walled gardens won through convenience.
Taiwan makes this conflict unusually visible. It has produced one of the world's best-known civic technology communities. g0v is built around grassroots participation, open-source collaboration, and releasing work so that others can build on it. It treats digital systems as spaces in which participation should remain open and power should remain contestable.
Everyday Taiwanese internet life follows a different model. According to the 2025 Taiwan Internet Report, 86.59 percent of internet users used social media. When asked to choose their most frequently used social platform, 51.38 percent selected Facebook and 21.07 percent chose Instagram. LINE was the most frequently used messaging service for 90.26 percent of internet users.
Taiwan's digital democracy may be open, but much of its digital public life takes place inside systems owned elsewhere.
Instagram is a country you cannot vote in
Calling Instagram a country sounds dramatic until we consider what governments actually do. They establish rules, recognise identities, determine what may be seen, resolve disputes, and decide who can participate.
Platforms do all of these things.
An Instagram account functions as a form of identity. Its recommendation system determines visibility. Its moderation system establishes acceptable speech and behaviour. Its appeals process acts as a court. A suspension can remove years of work, contacts, and cultural memory.
The difference is that Instagram's residents cannot vote out its leadership, inspect its recommendation system, or move their relationships to another provider. They can download an archive, perhaps, but an archive is not an audience. Ten thousand followers cannot be exported and contacted somewhere else in the way an email list can.
In June 2026, a technical problem caused Meta to suspend between 3,000 and 4,000 Taiwanese accounts after mistakenly identifying users as being under thirteen. The affected accounts included media organisations, public figures, and ordinary users. Taiwan's Ministry of Digital Affairs contacted Meta and asked it to restore them.
Most accounts were eventually returned. The incident still revealed the arrangement clearly. A foreign company's automated system could remove Taiwanese people and institutions from a significant part of their own public sphere. The local government could request assistance, but it could not directly reverse the decision.
This dependence is easy to overlook because the garden is pleasant. People are already inside. The paths are maintained. Publishing takes seconds. Discovery arrives automatically. A restaurant can reach customers without learning web design, and an artist can find an audience without understanding hosting.
The garden provides genuine value. It also has a gate.
The rebellion that became a market
Web3 emerged partly from an awareness of this imbalance.
Its advocates argued that users should own their identities, digital assets, and communities. A person should be able to leave a service without losing everything accumulated there. Creators should have direct relationships with supporters. Platforms should become replaceable interfaces over shared infrastructure.
The diagnosis was often correct.
Then ownership became associated with tokens. Community became associated with speculation. Participation required wallets, seed phrases, transaction fees, and an ability to distinguish technical complexity from financial theatre. The movement attempted to escape the walled garden and constructed a marketplace at the exit.
Web3 was right about the problem. It made the solution feel like a casino.
The open web had already offered a quieter kind of ownership. A domain provides a stable public address. A website can be moved between hosting providers. A URL can be linked from any compatible service. Email lists, feeds, and files can be handled by different software without requiring permission from a single platform.
This form of ownership is incomplete. A domain does not provide an audience, and a website does not automatically create a community. Open protocols often feel less coherent than a single integrated product. Yet they preserve an important possibility: the service provider can change while the underlying identity or material survives.
Newer decentralised social protocols are revisiting this problem without making financial assets the centre of every interaction. ActivityPub is a W3C standard that allows independently operated social services to communicate. The AT Protocol is designed around interoperable providers, algorithmic choice, and account portability. Its architecture assumes that a hosting provider may fail and tries to make it possible for the user to move without that provider's cooperation.
These systems remain smaller and rougher than Instagram. Their relevance lies in what they demonstrate. Social networking does not inherently require one company to own the identities, relationships, content, and interface at the same time.
Google's open web
There is no clean hero in this story.
Google is one of the largest supporters of the open web, largely because its business depends on the web continuing to exist. In 2024, Google made more than 100,000 commits to Chromium, around 94 percent of all contributions. The company says maintaining Chromium's shared infrastructure costs hundreds of millions of US dollars annually before the development of new features is considered.
Google is also preparing websites for an internet increasingly visited by software agents. At Google I/O 2026, Chrome described an agentic web in which AI systems can interact with websites through proposed open standards such as WebMCP. Instead of forcing an agent to imitate a human clicking through a page, a website could expose structured tools and forms that the agent can use directly.
The incentive is straightforward. Search needs public information to index. AI systems need sources they can retrieve, compare, and cite. Agents need places where they can find products, book services, and complete transactions.
Google therefore has a reason to keep the web open while positioning Chrome, Search, and Gemini as its dominant interfaces. Anyone may build a road, while Google would prefer to operate the map, the main junction, and the vehicle travelling across it.
This is still different from a platform that wants every meaningful interaction to occur within its own application. Google's interests overlap with openness, even when they do not overlap with decentralised power.
Websites after browsing
The survival of the website may depend on accepting that fewer people will experience it in the traditional way.
A customer might discover a business through Instagram. A search engine may extract its opening hours. An AI assistant might compare its services with competitors and make a booking. A person may never visit the homepage or admire its navigation.
The website still matters because it acts as the underlying source. It establishes a public identity that exists beyond a social profile. It provides material that search engines and agents can access. It gives an organisation somewhere to preserve its archive, explain itself on its own terms, and conduct transactions without placing every relationship inside a platform.
The website is becoming infrastructure.
That description may sound less exciting than calling it a destination, but infrastructure has greater consequences. Roads, addresses, and public records are rarely entertaining. Their value comes from allowing many different activities to happen without requiring approval from the same private gatekeeper.
For a Taiwanese restaurant, artist, or independent organisation, the practical response does not involve leaving Instagram. The audience is there, and refusing to participate would often amount to voluntary invisibility. The more durable strategy is to treat the platform as an outpost. Discovery may happen there, while the identity, archive, and source of truth remain somewhere under the organisation's control.
This principle also applies at the level of a society.
Taiwan is unusually conscious of sovereignty, resilience, and dependence on systems controlled elsewhere. It has invested in open government, civic participation, and digital tools that allow citizens to inspect and influence public institutions. The same attention has rarely been applied to the privately governed spaces where culture, business, and political conversation increasingly occur.
The question is not whether Taiwanese people care about the open web. Most people will never use that phrase. They may still care whether a business can reach its customers after an account suspension, whether an artist can retain a decade of work, whether a newspaper can be removed by an age-detection error, and whether Taiwan's public conversation should depend on rules written in California.
The walled gardens won because they made belonging effortless. Web3 struggled because it made ownership exhausting. The open web's next task is to make independence feel ordinary.
A website may look modest beside an endless feed. It may receive fewer direct visitors than it once did. Yet it continues to represent something increasingly rare online: a place where a person or organisation can exist without requiring a platform's permission.
AI authorship note
This essay began with Kris Haamer's long-running assumption that websites were dying, followed by a discussion about Google's investment in the open web, Taiwan's dependence on Instagram and LINE, and Web3 as a failed rebellion against platform ownership. AI supported research, synthesis, drafting, editing, and implementation. Kris is responsible for the editorial direction, the Taiwan framing, and the decision to publish it.