bits.

FREE Companies & institutions LESSON · Companies & institutions

1989: Keep the Web one Web

CERN, W3C, Netscape, Microsoft, and the browser boundary

The Web's chronology needs several dates. Tim Berners-Lee's initial proposal belongs to 1989; a later proposal with Robert Cailliau is dated November 1990; CERN's public-domain release document is dated 30April1993. Standards institutions and commercial browser competition developed afterward. The contribution was not one release that instantly supplied every feature of the modern browser. It was a combination of identifiers, document representations, transfer protocols, implementations, and permissions that could support independent participation.

A URL identifies a resource; HTTP specifies request and response semantics; HTML provides document structure interpreted by browsers. Their separation lets a publisher host a page without building every reader's browser. A browser can retrieve pages from independent servers without negotiating a bespoke interface for each publisher. That interoperability is a practical property tested by implementations, not merely the existence of a written specification.

Consider a fictional publisher with 100 articles and three browser families. If each browser requires a distinct representation, the publisher maintains 300 versions. A common representation can reduce the main content to 100 versions plus compatibility testing. But if one browser adds a proprietary layout behavior that attracts most readers, the publisher may begin depending on it. The shared standard still exists while practical portability weakens.

CERN's release document provides evidence about permission to use and redistribute the software. It should not be confused with a guarantee that every later Web technology is patent-free, every service is open source, or every website can be accessed without restriction. The W3C Process Document describes a separate institutional mechanism for developing and reviewing technical work. A standard's governance, patent commitments, implementation quality, and actual deployment each require their own evidence.

Default distribution can influence adoption independently of protocol quality. A browser already installed on many machines is easier to reach than a technically capable alternative requiring discovery and installation. Developers then target the largest reachable audience, which can reinforce the default's position. The United States Microsoft case offers a detailed historical record of such conflicts, but its findings concern specified conduct and evidence; they are not a universal verdict on every integrated product.

A useful interoperability test includes multiple independent implementations, malformed inputs, accessibility behavior, and version changes. Perfectly copying one implementation's accidental bugs can preserve existing pages while making the specification less meaningful. Strictly following an incomplete specification can also break deployed content. Standards work therefore combines technical precision with migration and governance decisions.

The engineering question is how users and creators retain credible alternatives. Open document formats, exportable data, stable identifiers, accessible interfaces, and independent implementations can help. None removes every concentration effect, because hosting, discovery, identity, and distribution remain separate control points. Study the Web as a set of negotiated boundaries whose openness can vary by layer, rather than calling the entire system either open or closed.

Reconstruction exercise

A publisher has 100 articles and maintains three browser-specific versions of each. A shared format reduces this to 100 main versions but requires 20 compatibility fixtures per browser. Compare the number of maintained artifacts, stating why the count is only a rough cost model. Then a proprietary extension is used by 70 articles. Describe an exit plan and a test showing that the replacement preserves more than visual similarity.

Worked solution

The original count is 300 article versions. The new count is 100 articles plus 3×20=60 fixtures, totaling 160 artifacts, though a fixture and an article may require different effort. For the 70 dependent articles, inventory the extension’s behavior, choose a standard representation or fallback, and migrate in batches while testing independent browsers. Verify document structure, keyboard access, link targets, readable ordering, and data meaning as well as appearance. A screenshot match can hide inaccessible controls or broken navigation. The plan also needs durable URLs and redirects where necessary, because interoperable content that destroys existing references still imposes a migration cost on readers and other publishers.

Test the limits and transfer the idea

1. State trace and test design · allow 25 minutes

A page can look correct while its contract fails

A fictional browser extension hides a form's validation error visually but leaves the submit button enabled. Browser A blocks invalid submission internally; Browser B ignores the extension and sends the form. The server trusts client validation. Trace an invalid age=-5 through both browsers and design a correction that preserves usability and data validity. Include a test with scripts disabled and a keyboard-only user. Explain why screenshot comparison is insufficient.

Get a starting hint

The browser can assist entry, but the server must enforce the state invariant independently.

Compare your reasoning with the worked solution

A prevents the request, though its hidden error may leave the user unable to understand why. B sends age=-5, and the trusting server stores an invalid record. The correction validates the allowed age range on the server, returns a structured field error without committing, and renders a visible, keyboard-reachable explanation with the user's input preserved. Client validation can add immediate feedback but cannot be the only enforcement point. Test an ordinary valid request, a direct invalid request, disabled scripts and keyboard submission/recovery; compare durable records as well as pixels. A published Web process is a primary governance document, and the CERN proposal is a primary design proposal. Neither establishes that this fictional implementation follows a correct validation contract.

Check your reasoning

  • Traces different browser behavior to the shared server flaw.
  • Enforces validity at the durable-state boundary.
  • Tests recovery and accessibility beyond visual similarity.

2. Source criticism · allow 25 minutes

A release document is not a claim about every later service

An article reads CERN's 1993 public-domain software release and concludes that all modern websites must expose their source, charge no fees, and allow every browser. Analyze each inference. Then design an evidence table for one actual historical implementation, separating software rights, protocol conformance and operating access policy. Explain how the 1990 proposal, 1993 release and a later W3C process document answer different questions.

Get a starting hint

Permission granted for a named artifact is not an obligation automatically imposed on all systems using an idea.

Compare your reasoning with the worked solution

The release concerns specified CERN software and its rights. It does not require unrelated website implementations to publish source, prohibit charging, or prevent servers from applying access rules. A table should identify the exact implementation and version; attach its license or release instrument to the rights row, conformance specification and test results to the protocol row, and contemporaneous operating terms or access logs to the access-policy row. The 1990 proposal records a design and intended use; the 1993 instrument records a permission decision; a W3C process specifies how later standards work is organized. These are primary records for different assertions. A later summary that calls the Web open is an interpretation whose scope must be tied to the relevant layer.

Check your reasoning

  • Rejects each unsupported universal obligation.
  • Pairs three separate claims with the correct evidence types.
  • Distinguishes design proposal, rights instrument and governance process.

Check your understanding

A website works only because it depends on one browser’s undocumented extension. What has happened despite an open base standard?

  • Practical portability has weakened at an additional implementation boundary.
  • The publisher has eliminated compatibility testing.
  • The base standard has legally disappeared.
  • All browsers must already support the extension.
Explanation

Openness can differ by layer; deployment may depend on behavior outside the shared contract.

A migrated page looks identical in screenshots but keyboard controls and links fail. Which conclusion is warranted?

  • The migration is complete because pixels match.
  • Accessibility is unrelated to interface behavior.
  • Visual similarity did not establish functional or accessible interoperability.
  • Broken links prove the new browser is nonstandard.
Explanation

The required behavior includes navigation, structure, input, and meaning as well as appearance.

Sources and further reading