Understanding the Web's Missing Structure: A Q&A on the Block Protocol and Semantic Web

<p>Since the 1990s, the web has primarily been a place for human-readable documents. But as we'll explore in this Q&A, the lack of structure in standard web pages limits what computers can do with that information. Let's dive into the challenges and the vision for a more intelligent web.</p> <h2 id="q1">What was the original web like?</h2> <p>The early web, beginning in the 1990s, was designed as a publishing platform for documents intended to be read by people. These documents were written in HTML, which gave a basic level of structure—like marking a paragraph or emphasizing a word. Then, CSS was added to make things look appealing, such as styling paragraphs with tiny gray text. While this worked for human readers, the structure remained superficial. A computer program could see that something was in a paragraph or bolded, but it couldn't understand the <em>meaning</em> behind the content. That limited the web to being just a collection of pretty pages, not a rich source of data that machines could process automatically.</p><figure style="margin:20px 0"><img src="https://www.joelonsoftware.com/wp-content/uploads/2022/12/IMG_0203-scaled.webp" alt="Understanding the Web&#039;s Missing Structure: A Q&amp;A on the Block Protocol and Semantic Web" style="width:100%;height:auto;border-radius:8px" loading="lazy"><figcaption style="font-size:12px;color:#666;margin-top:5px">Source: www.joelonsoftware.com</figcaption></figure> <h2 id="q2">What is the core problem with HTML's limited structure?</h2> <p>HTML tells a browser how to display content, but it doesn't tell a computer what the content means. For instance, you can make text <strong>bold</strong> or turn it into a heading, but there's no way to say “this is a book title” or “this is an author's name.” Without that deeper meaning, automated programs—whether simple bots or advanced AI—struggle to extract and use the information reliably. The problem becomes clear when you want to do things like aggregate data from multiple pages or power intelligent assistants. The web is full of human-friendly formatting, but very little machine-friendly semantics.</p> <h2 id="q3">Can you give an example of a typical book citation lacking structure?</h2> <p>Imagine you write a blog post mentioning <em>Goodnight Moon</em> by Margaret Wise Brown, illustrated by Clement Hurd, published by Harper &amp; Brothers in 1947, with ISBN 0-06-443017-0. On a standard web page, you might just make the title bold. A computer reading that page sees only formatting: a string of text with a bold tag. It cannot automatically identify this as a book, or extract the author, illustrator, publisher, or ISBN—even though that information is right there. In contrast, if the page used semantic markup from <a href="https://schema.org">schema.org</a>, a program could say, “Aha! This is a <em>Book</em> entity with these properties.” Without such structure, the web remains largely blind to the data it contains.</p> <h2 id="q4">What was Tim Berners-Lee's vision for the Semantic Web?</h2> <p>As early as 1999, Tim Berners-Lee dreamed of a web where computers could analyze all data—content, links, and transactions. In his book <em>Weaving the Web</em>, he described a “Semantic Web” that would make it possible for machines to talk to machines, handling trade, bureaucracy, and daily life through intelligent agents. The core idea was to add explicit meaning to web content using standards like RDF and JSON-LD, often based on shared vocabularies such as <strong>schema.org</strong>. That way, a computer could understand that a page mentions a book, a person, a recipe, or an event—not just display it. Berners-Lee believed this would unlock huge potential for automation and interoperability.</p> <h2 id="q5">How would the Semantic Web add structure to a page?</h2> <p>To make a page machine-readable, you would start by looking up a suitable vocabulary, such as schema.org's definition for a <em>Book</em>. Then you would add extra markup to your HTML using formats like <strong>JSON-LD</strong> or <strong>RDFa</strong>. For example, you might embed a snippet of JSON-LD that explicitly says: “This is a Book with title X, author Y, ISBN Z.” When a computer reads the page, it sees not just a paragraph but a structured data object. This approach allows search engines, smart assistants, and other tools to extract precise information. However, implementing it correctly requires extra effort after writing the human-readable content, which has been a major barrier to widespread adoption.</p><figure style="margin:20px 0"><img src="https://www.joelonsoftware.com/wp-content/uploads/2016/12/11969842-1.jpg" alt="Understanding the Web&#039;s Missing Structure: A Q&amp;A on the Block Protocol and Semantic Web" style="width:100%;height:auto;border-radius:8px" loading="lazy"><figcaption style="font-size:12px;color:#666;margin-top:5px">Source: www.joelonsoftware.com</figcaption></figure> <h2 id="q6">Why has adoption of semantic markup been so slow?</h2> <p>Despite the promise of the Semantic Web, adoption remains minimal. The main reason is the effort required: once you've written a beautiful blog post or article for people, adding semantic markup feels like extra homework. It's time-consuming to learn the vocabularies and embed the correct code. Also, without immediate payoff—like a search engine that rewards rich snippets—many authors give up. As a result, even decades after Berners-Lee's dream, very few pages include semantic annotations. The web remains mostly flat data from a computer's perspective, and the envisioned “intelligent agents” have not materialized because there's so little structured information available for them to use.</p> <h2 id="q7">Why is fixing this important for human progress?</h2> <p>Adding semantic structure to web pages would dramatically improve how information is shared and processed. Human progress depends on making data accessible not only to people but also to AI systems and traditional programs. For instance, a structured web would allow smart assistants to answer complex questions, researchers to aggregate data automatically, and businesses to streamline operations. Without this foundation, we limit the potential of machine learning and automation. Making semantic markup easy and rewarding is a crucial step toward a web where both humans and machines can truly understand and use the vast amount of information available.</p> <h2 id="q8">What would encourage people to add semantic markup?</h2> <p>In my view, people will only add semantic markup to their web pages if the process becomes as simple as writing in WordPress or using a simple plugin—essentially, if it's a natural part of their workflow instead of an extra chore. The <strong>Block Protocol</strong> aims to solve exactly this: by standardizing how content blocks carry structured data, authors can enrich their pages without learning complex formats. When semantic richness is built into the tools we already use, adoption can skyrocket. That's the path to finally realizing Berners-Lee's vision: a web that is both human-friendly and machine-readable, unlocking new levels of automation and discovery.</p>