6 Protocol Pioneers: Web, Email, and Routing

6.5.1 The Problem They Solved

By the early 1970s, ARPANET connected a few dozen computers using the Network Control Protocol (NCP). But NCP had critical limitations:

• Only worked on ARPANET — couldn’t connect to other networks (satellite, radio, Ethernet)
• No error recovery — if a packet was lost, applications had to detect and retransmit
• Assumed reliable links — didn’t handle packet reordering or corruption

ARPA wanted to connect heterogeneous networks — ARPANET, satellite links (SATNET), and packet radio networks (PRNET) — into a single inter-network. No existing protocol could do this.

6.5.2 Their Breakthrough: TCP/IP and the Open Architecture Principle

Baran showed that packet networks could be resilient. Cerf and Kahn’s next question was broader: how can different packet networks become one inter-network?

In 1974, Vint Cerf and Bob Kahn published “A Protocol for Packet Network Intercommunication” in IEEE Transactions on Communications. Their protocol had two revolutionary ideas:

1. The Open Architecture Principle (Kahn’s Insight) Any network — wired, wireless, satellite, smoke signals — can join the Internet if it can carry IP packets. The Internet doesn’t care how individual networks work internally. This is called network layer independence.

2. End-to-End Reliability (Cerf’s Insight) Don’t trust the network to be reliable. Instead, put reliability logic in the endpoints — the computers sending and receiving data. If a packet is lost, the receiving computer detects the gap (via sequence numbers) and asks the sender to retransmit.

This became the end-to-end principle: intelligence belongs at the edges, not in the network core. The network’s job is simple: forward packets. The endpoints handle everything else.

6.5.3 Cerf’s Personal Motivation

Vint Cerf is partially deaf from birth. This personal experience gave him a unique perspective on communication:

“I’ve always been interested in communication because I understand what it’s like when communication fails. The Internet is fundamentally about connecting people and ideas — making sure messages get through despite imperfect infrastructure.”

His passion for robust communication influenced TCP’s design. TCP doesn’t just detect errors — it corrects them, automatically retransmits lost packets, and reorders out-of-sequence packets. This robustness reflects Cerf’s personal understanding that communication cannot be taken for granted.

6.5.4 The Impact

TCP/IP became mandatory on ARPANET in 1983 (“Flag Day”). By the 1990s, it was the foundation of the global Internet. Today, every device — from supercomputers to $2 ESP32 microcontrollers — speaks TCP/IP.

Checkpoint: Internet Architecture

You now know:

• TCP/IP, published in 1974, made heterogeneous networks interoperate instead of forcing every link to behave the same way.
• IP keeps the network core simple: forward packets across whatever link is available.
• TCP puts reliability at the endpoints, which is why MQTT over TCP can survive Wi-Fi, cellular, and satellite variation.

6.5.5 Cerf Today: Chief Internet Evangelist at Google

Vint Cerf continues to advocate for Internet expansion. As Google’s “Chief Internet Evangelist,” he works on:

• Interplanetary Internet: Extending TCP/IP to work over multi-minute delays between Earth and Mars
• Internet access as a human right: Advocating for global connectivity
• IPv6 adoption: Ensuring enough addresses for billions of IoT devices ($2^{128}$ addresses, vs. IPv4’s $2^{32}$)

6.6.1 The Problem He Solved

In 1989, CERN (the European physics lab) had thousands of researchers generating papers, datasets, and software. But there was no easy way to link related documents across different computers. If you read a paper that referenced another paper, you had to manually search for it on a different system. Knowledge was siloed.

Berners-Lee, a British software engineer at CERN, had a vision: What if documents could link to each other across the network? Click a word, jump to a related document, regardless of which computer hosts it.

6.6.2 His Breakthrough: The World Wide Web

Between 1989 and 1991, Berners-Lee invented the three core technologies that make the Web work:

1. HTML (HyperText Markup Language) A simple way to format documents with hyperlinks — clickable text that references other documents by URL.

<a href="https://www.w3.org/History/1989/proposal.html">Original Web Proposal</a>

2. HTTP (HyperText Transfer Protocol) A lightweight protocol for requesting and serving documents over TCP/IP. HTTP is stateless — each request is independent, making it simple and scalable.

GET /related-paper.html HTTP/1.0
Host: example.com

3. URLs (Uniform Resource Locators) A standard way to address any resource on the Internet: http://server/path/to/file.

4. The First Web Browser (WorldWideWeb) Berners-Lee wrote the first browser and web server in 1990 on a NeXT computer. Both were free and open-source.

6.6.3 The Decision That Changed the World

In 1993, CERN faced a choice: patent the Web or release it freely?

Berners-Lee convinced CERN to release the Web into the public domain with no royalties, no restrictions. This decision made the Web universal. If CERN had patented it, the Web might have become a proprietary system like AOL or CompuServe — walled gardens that eventually died.

6.6.4 The Impact

By 2026, over 5 billion people use the Web. It has become the primary interface to the Internet. Berners-Lee was knighted by Queen Elizabeth II in 2004 for his contribution to humanity.

6.6.5 Berners-Lee Today: Fighting for an Open Web

Berners-Lee founded the World Wide Web Consortium (W3C) to maintain open web standards. Today, he’s working on Solid — a project to give users control over their personal data, addressing privacy concerns that have emerged 30+ years after the Web’s creation.

6.7.1 The Problem He Solved

In 1971, computers could run local messaging programs — users on the same computer could leave messages for each other. But ARPANET had connected computers at different sites. Tomlinson, working at BBN Technologies (the company building ARPANET), asked: What if people could send messages across the network to users on different computers?

6.7.2 His Breakthrough: Networked Email with the @ Symbol

Tomlinson wrote the first networked email program in 1971. His key innovation was the user@host addressing format:

tomlinson@bbn-tenexa

Why the @ symbol? Tomlinson needed a character that:

• Wasn’t already used in usernames (so fred@host wouldn’t be confused with fred)
• Indicated “at” in natural language
• Was available on the Model 33 Teletype keyboard

The @ symbol was perfect. It was rarely used for anything else (outside accounting contexts), and it naturally read as “user at host.”

6.7.3 What Was the First Email?

Tomlinson sent the first networked email to himself — from one computer to another computer sitting right next to it. What did it say?

He doesn’t remember. In later interviews, he said it was probably something like:

QWERTYUIOP

— just testing the keyboard. The content didn’t matter. What mattered was that it worked.

6.7.4 The Impact

By 1973, email was 75% of ARPANET traffic. Today, billions of people use email, all following Tomlinson’s user@host format. The @ symbol has become one of the most recognized symbols in the world.

Checkpoint: Human-Readable Network Use

You now know:

• Berners-Lee’s 1989 Web work made networked documents usable through HTTP, HTML, and URLs.
• Tomlinson’s email work made user@host addressing memorable enough to survive into everyday life.
• Both stories matter to IoT because dashboards, REST APIs, webhooks, device IDs, and topic names need human operators to understand them.

6.8.1 The Problem She Solved

When you connect Ethernet switches together, you create a network graph. But graphs can have loops — paths that circle back on themselves. In a loop, a broadcast packet circulates forever, consuming all bandwidth. Within seconds, a single packet can multiply into a broadcast storm that crashes the entire network.

In the early 1980s, network engineers had to carefully plan network topologies to avoid loops. Redundancy was risky — if you added a backup link for resilience, you might accidentally create a loop that crashed everything.

6.8.2 Her Breakthrough: Spanning Tree Protocol (STP)

In 1985, while working at Digital Equipment Corporation (DEC), Radia Perlman invented the Spanning Tree Protocol. STP automatically detects and blocks loops in Ethernet networks:

How STP Works:

1. Switches exchange messages to discover the network topology
2. They elect a “root bridge” (the switch with the lowest ID)
3. Each switch calculates its shortest path to the root
4. Any link that would create a loop is automatically blocked (disabled)
5. If a link fails, STP recalculates and unblocks backup paths

This happens automatically. Network engineers can connect switches with redundant links for resilience, and STP ensures the active topology is always loop-free.

6.8.3 Why She Dislikes the “Mother of the Internet” Title

Media often calls Perlman the “Mother of the Internet” to parallel Vint Cerf’s “Father of the Internet” title. She dislikes this for two reasons:

1. Many people built the Internet — singling out individuals ignores collaborative effort
2. Gender-specific titles create false parallels — her contributions stand on their own merit, not as a female counterpart to a male pioneer

Perlman prefers to be known for her work: STP, TRILL (modern Ethernet routing), and network security research.

6.8.4 The Impact

Every Ethernet switch in every building runs Perlman’s algorithm. STP is standardized as IEEE 802.1D and has evolved into Rapid Spanning Tree (RSTP, 802.1w) and Multiple Spanning Tree (MSTP, 802.1s), but the core algorithm is Perlman’s 1985 invention.

6.8.5 Perlman Today

Perlman holds over 100 patents and has taught at MIT and the University of Washington. Her textbook Interconnections is considered the definitive guide to network protocol design. She continues to work on network security and routing innovations.