Chapters

4 Internet History: Web, Mobile, and IoT

fundamentals
history
internet

4.1 Start With the Decision

The web made linked pages easy to reach. Mobile, cloud, and IoT then moved that reach into daily devices and services.

4.2 Route Overview

This is part 2 of 2. Review Internet History: ARPANET and TCP/IP for the preceding evidence.

4.3 Learning Objectives

  • Connect DNS and the web to human use of the Internet.
  • Compare the web, mobile-cloud, and IoT eras.

4.4 Chapter Roadmap

  • Checkpoint: TCP/IP and DNS
  • The World Wide Web (1991-1999)
  • The Mobile and Cloud Era (2000-2015)
  • The IoT Era (2015-Present)
  • Checkpoint: Web to IoT
  • Timeline: 55 Years of Networking
  • The Unsung Heroes
  • See Also
  • Common Pitfalls
  • Knowledge Check
  • Deep Dive: Packet Switching, Layers, and the IP Narrow Waist
  • Summary
  • What’s Next
  • Key Takeaway
Physics PhoebeCheckpoint: TCP/IP and DNS

You now know:

  • January 1, 1983 matters because ARPANET switched from NCP to TCP/IP in one coordinated Flag Day.
  • DNS changed numeric addressing into names such as usc.edu, while IoT deployments still need to plan cache TTLs and failover names.
  • The Internet became practical because BSD Unix, NSFNET, and open protocols let more hosts join without redesigning every application.

4.5 The World Wide Web (1991-1999)

4.5.1 Tim Berners-Lee’s Gift to Humanity

Tim Berners-Lee, a British scientist at CERN (the European particle physics laboratory), was frustrated that researchers couldn’t easily share documents across different computer systems. In 1989, he proposed a system combining three ideas:

  1. HTML (HyperText Markup Language) — a way to format documents
  2. HTTP (HyperText Transfer Protocol) — a way to request and send documents
  3. URLs (Uniform Resource Locators) — a way to address any document on any server

On August 6, 1991, he published the first website at CERN: http://info.cern.ch. Crucially, Berners-Lee and CERN made the technology free for everyone. They could have patented it and charged licensing fees, but they chose to give it away. This single decision — perhaps more than any technical achievement — is why the Web became universal.

4.5.2 The Browser Wars and Explosive Growth

  • 1993: Marc Andreessen releases Mosaic — the first graphical web browser. Downloads increase 341,000% in one year
  • 1994: Andreessen co-founds Netscape. Amazon and Yahoo are founded
  • 1995: JavaScript created by Brendan Eich in 10 days. Internet Explorer launched. eBay founded
  • 1996: Larry Page and Sergey Brin begin the BackRub project at Stanford (later renamed Google)
  • 1998: Google incorporated. RFC 2616 defines HTTP/1.1 (still widely used)
  • 1999: Kevin Ashton coins “Internet of Things” in a presentation at Procter & Gamble

The growth of the Web was unlike anything before:

YearWebsitesInternet UsersGrowth Rate
19911~4 million
1993130~14 million130x sites
199523,500~40 million181x sites
19971 million~120 million43x sites
200017 million~413 million17x sites

From 1 website to 17 million in just 9 years — compound annual growth rate of approximately CAGR=(17,000,0001)1/91536%\text{CAGR} = \left(\frac{17{,}000{,}000}{1}\right)^{1/9} - 1 \approx 536\% per year.

Interactive CAGR Calculator: Try calculating compound annual growth rate for different Internet metrics:

Quick Check: Test your understanding of the World Wide Web:

4.6 The Mobile and Cloud Era (2000-2015)

4.6.1 From Desktop to Pocket

  • 2001: Wikipedia launches — collaborative knowledge on the Internet
  • 2003: Skype launches — VoIP goes mainstream
  • 2004: Facebook launches. Web 2.0 emerges (user-generated content)
  • 2007: Apple iPhone released — the smartphone revolution begins
  • 2008: Android launched. GitHub founded. First Bitcoin paper published
  • 2010: Instagram launches. iPad introduced. “Cloud computing” enters mainstream vocabulary
  • 2013: Cisco estimates 10 billion devices connected to Internet

4.6.2 The Seeds of IoT

Several parallel developments in this era made IoT possible:

  • Cheap, ultra-low-power compute: ARM’s Cortex-M series (first released 2004) put a 32-bit processor with as little as 8 KB of RAM and sub-microamp sleep currents within reach of mass-market embedded devices
  • Cheap sensors: MEMS accelerometers dropped from $5 to $0.50
  • Low-power wireless: Bluetooth Low Energy (BLE) standardized in 2010, able to send a small packet once a second for a year on a single coin-cell battery
  • Cloud platforms: AWS IoT (2015), Azure IoT Hub (2015), Google Cloud IoT (2017)
  • IPv6: With 21282^{128} addresses (3.4×10383.4 \times 10^{38}), every grain of sand on Earth could have its own IP address

4.7 The IoT Era (2015-Present)

4.7.1 Kevin Ashton’s Vision Realized

When Kevin Ashton coined “Internet of Things” in 1999, he was working on RFID supply chain tracking at Procter & Gamble. His insight was simple but profound: computers depend on humans to input data, but most of the real world isn’t in any database. If machines could sense the physical world directly, they could manage things far more efficiently.

Today, that vision is becoming reality:

Metric201520202025 (projected)
Connected IoT devices3.8 billion11.7 billion27 billion
Global IoT spending$698 billion$749 billion$1.1 trillion
Data generated daily2.5 exabytes40 exabytes463 exabytes

4.7.2 The People Behind IoT Protocols

The IoT revolution built on protocols designed by dedicated engineers:

ProtocolCreator(s)YearKey Innovation
MQTTAndy Stanford-Clark (IBM), Arlen Nipper1999Publish-subscribe for unreliable networks
CoAPZach Shelby, Carsten Bormann (IETF)2014REST for constrained devices
6LoWPANGeoff Mulligan, Nandakishore Kushalnagar2007IPv6 over low-power wireless
LoRaWANSemtech, IBM Research2015Long-range, low-power wide-area
MatterGoogle, Apple, Amazon, Samsung2022Universal smart home standard
ThreadARM, Google/Nest, Samsung2014IPv6 mesh networking for IoT

Physics PhoebeCheckpoint: Web to IoT

You now know:

  • The Web is an application on top of the Internet, not the Internet itself.
  • Web growth ran from 1 website in 1991 to 17 million in 2000, while IoT device counts grew from 3.8 billion in 2015 to 11.7 billion in 2020.
  • MQTT, CoAP, 6LoWPAN, LoRaWAN, Matter, and Thread extend the older packet, layer, and open-standard pattern into constrained devices.

Quick Check: Test your understanding of Internet pioneers and their contributions:

4.8 Timeline: 55 Years of Networking

1957 ━━ Sputnik launched → ARPA created (1958)
1961 ━━ Kleinrock publishes packet network theory
1962 ━━ Baran designs distributed networks at RAND
1965 ━━ Davies coins "packet switching" at NPL
1969 ━━ ARPANET: first message "LO" (UCLA → SRI)
1971 ━━ Email invented by Tomlinson (@)
1973 ━━ First international ARPANET connections
1974 ━━ Cerf & Kahn publish TCP paper
1978 ━━ TCP split into TCP + IP
1983 ━━ "Flag Day" — TCP/IP becomes Internet standard
      ━━ DNS invented by Mockapetris
1985 ━━ First .com domain (symbolics.com)
1986 ━━ NSFNET backbone (56 Kbps)
1989 ━━ Berners-Lee proposes the World Wide Web
1991 ━━ First website goes live at CERN
1993 ━━ Mosaic browser — the Web goes graphical
1995 ━━ JavaScript, Internet Explorer, eBay
1998 ━━ Google incorporated
1999 ━━ Kevin Ashton coins "Internet of Things"
      ━━ MQTT protocol created at IBM
2004 ━━ Facebook launched
2007 ━━ iPhone released, 6LoWPAN standardized
2010 ━━ Bluetooth Low Energy standardized
2014 ━━ CoAP standardized (RFC 7252)
2015 ━━ LoRaWAN 1.0 released
2020 ━━ 11.7 billion IoT devices connected
2022 ━━ Matter 1.0 — universal smart home standard
2025 ━━ ~27 billion IoT devices projected

Quick Check: Place these Internet milestones in the correct chronological order:

4.9 The Unsung Heroes

Many crucial Internet pioneers rarely appear in popular histories:

  • Radia Perlman — Invented the Spanning Tree Protocol (STP) that prevents network loops. Called “Mother of the Internet”
  • Elizabeth Feinler — Ran the ARPANET Network Information Center, managing the naming system before DNS
  • Bob Metcalfe — Invented Ethernet at Xerox PARC in 1973, connecting devices over local networks
  • Jon Postel — Managed Internet Assigned Numbers Authority (IANA) for decades, edited most early Internet standards (RFCs)
  • Steve Crocker — Wrote the first RFC (Request for Comments) in 1969, establishing the Internet’s open standards culture
  • Geoff Huston — Pioneered Internet governance and routing analysis

The big picture: When your IoT temperature sensor sends a reading today, it follows the SAME fundamental process invented in 1969 — just billions of times faster.

Step-by-step breakdown:

  1. Your sensor generates data: A temperature reading of 23.5°C (just like Charley Kline typing “L-O-G-I-N”)
  2. TCP/IP wraps it in layers: Application data → TCP segment → IP packet → Ethernet frame (Cerf & Kahn’s 1974 design)
  3. DNS resolves the server name: “iot-server.example.com” → 93.184.216.34 (Mockapetris’s 1983 invention)
  4. Packets route through the Internet: Each packet finds its own path, just like Baran envisioned in 1962
  5. The server receives and processes: Using HTTP/MQTT/CoAP — protocols standing on 50+ years of evolution

Why this matters: The Internet wasn’t designed by one person or one company. It was built collaboratively by thousands of researchers who shared their work freely. That culture of openness — embodied in the RFC process started by Steve Crocker in 1969 — is why IoT protocols like MQTT and CoAP are also open standards that anyone can implement.

ConceptRelates ToRelationship
Packet switchingCircuit switchingPacket switching replaced circuit switching as the fundamental network paradigm
TCP/IPOSI modelTCP/IP is the practical implementation that won over the theoretical OSI model
ARPANETInternetARPANET was the prototype; the Internet emerged when TCP/IP connected multiple networks
World Wide WebInternetThe Web runs ON the Internet — it’s an application, not the network itself
IoT protocolsTCP/IPMQTT, CoAP, and 6LoWPAN are modern adaptations of the TCP/IP architecture for constrained devices

Cross-module connection: The layered protocol architecture discussed here is formalized in OSI and TCP/IP Models. The constrained IPv6 adaptation used by many low-power networks is covered in IPv6 and 6LoWPAN.

4.10 See Also

Time: ~10 minutes | Difficulty: Beginner | No hardware needed

Challenge: Create a personal “Internet history” timeline showing how networking technology has affected YOUR life.

Steps:

  1. List 5 Internet-enabled services you use daily (social media, streaming, smart home, etc.)
  2. For each one, trace it back to the historical innovation that made it possible (e.g., Netflix → Web → HTTP → TCP/IP → ARPANET)
  3. Calculate: How many of the 1969-1999 innovations from this chapter does each service depend on?
  4. Identify which service would STOP WORKING first if you removed innovations in reverse chronological order

Start with the most recent innovation (e.g., Matter/Thread for smart home) and trace backwards. You’ll find that ALL modern services ultimately depend on TCP/IP (1974) and packet switching (1962).

Every Internet-connected service depends on at least these 5 foundational innovations:

  1. Packet switching (1962) — without it, no shared network
  2. TCP/IP (1974/1978) — without it, no universal protocol
  3. DNS (1983) — without it, you’d need to memorize IP addresses
  4. HTTP/HTML (1991) — without it, no web interfaces
  5. Wi-Fi/cellular (various) — without it, no mobile access

Smart home devices additionally depend on MQTT (1999), BLE (2010), and IPv6 (1998).

4.11 Common Pitfalls

Pitfall 1: Treating the Web and the Internet as the Same Thing

The Web is an application that runs on the Internet. IoT systems often use Internet connectivity without using web pages at all; they may send MQTT, CoAP, DNS, NTP, or device-management traffic over the same packet network.

Pitfall 2: Assuming TCP/IP Was a Single Finished Invention

TCP/IP evolved through research, papers, implementation pressure, and operational cutover. The split between IP and TCP, the later use of UDP, and the growth of DNS all show why Internet architecture is best read as a set of design decisions, not a single launch event.

Pitfall 3: Copying Old Dates Without the Architecture Lesson

The important point is not only that ARPANET began in 1969 or that Flag Day happened in 1983. The engineering lesson is that open protocols, layered boundaries, packet forwarding, and interoperable implementations let independent networks become one Internet.

4.12 Knowledge Check

4.13 Deep Dive: Packet Switching, Layers, and the IP Narrow Waist

The Internet that IoT rides on rests on two ARPANET-era ideas. The first is packet switching: instead of reserving a dedicated circuit, data is chopped into independently addressed packets that the network routes hop by hop, sharing links efficiently and rerouting around failures. The second is layering: the protocol stack is split into layers, each with one job, so they can evolve independently.

The TCP/IP model has four layers: link, internet, transport, and application. A sensor sending a reading and a browser loading a page use the same four layers; only the application protocol differs. Understanding that stack is the key to understanding every IoT protocol above it.

4.13.1 The Four Layers and Encapsulation

Each TCP/IP layer wraps the layer above it in its own header, a process called encapsulation:

LayerJobExamples
ApplicationThe actual message semanticsHTTP, MQTT, CoAP
TransportEnd-to-end delivery and portsTCP, UDP
InternetAddressing and routing across networksIP (IPv4/IPv6)
LinkDelivery over one physical hopEthernet, Wi-Fi, IEEE 802.15.4

A sensor sending {"temp":21.4} illustrates the process. The application layer defines that message; the transport layer, perhaps UDP, adds source and destination ports; the internet layer adds IP addresses so routers can forward it; the link layer adds a frame header for the next physical hop. At the receiver, the headers are peeled off in reverse.

Each layer reads only its own header. A router looks at the IP header and ignores the payload, which is why the network can carry protocols it has never heard of.

4.13.2 Why the Core Stays Simple

The Internet’s most consequential design choice is the end-to-end principle: keep the network core simple so it just forwards packets, and put intelligence such as reliability, ordering, encryption, and application meaning in the endpoints. A router does not track your whole conversation; TCP in your device handles retransmission. This is why anyone can invent a new application protocol without asking the network for permission.

The stack also has a famous narrow waist. Many link technologies sit below and many applications sit above, but everything meets at IP in the middle. Ethernet, Wi-Fi, cellular, and IEEE 802.15.4 can all carry IP, and HTTP, MQTT, CoAP, DNS, and many other protocols can all ride above it.

That narrow waist is what let IoT grow on the existing Internet. A constrained radio and a data-center server interoperate because both meet at IP, with technologies such as 6LoWPAN carrying IPv6 even over tiny 802.15.4 frames.

A CoAP-based sensor can talk to a cloud service written years earlier. Neither the routers between them nor the link technologies they use were designed for CoAP, yet it works because CoAP rides UDP, UDP rides IP, and the simple IP-forwarding core carries it unchanged. The end-to-end principle and the IP narrow waist are the reason IoT could be built on the existing Internet rather than on a separate network.

4.14 Summary

The Internet’s journey from 4 ARPANET nodes in 1969 to 27+ billion IoT devices today was shaped by key individuals, pivotal decisions, and a culture of open collaboration:

  • Paul Baran, Donald Davies, and Leonard Kleinrock independently invented packet switching — the foundation of all Internet communication
  • Vint Cerf and Bob Kahn designed TCP/IP — the universal protocol that connects everything
  • Tim Berners-Lee created the World Wide Web and gave it away for free
  • Kevin Ashton envisioned the Internet of Things — extending the Internet to physical objects
  • The Internet’s open standards culture (RFCs, free protocols, shared specifications) is why IoT protocols like MQTT, CoAP, and LoRaWAN are freely available today

The most remarkable aspect of this history is that the fundamental architecture designed in the 1970s — packet switching, layered protocols, end-to-end design — still works perfectly for IoT devices that are millions of times more numerous and diverse than anything the original designers imagined.

4.15 What’s Next

Now that you can trace the Internet’s evolution from ARPANET to IoT, explore how these protocols work in modern systems:

TopicChapterDescription
Protocol ArchitectureOSI and TCP/IP ModelsExamine the formal layered architecture that grew from Cerf and Kahn’s TCP/IP design
IoT MessagingMQTT Publish-Subscribe BasicsExplore how publish-subscribe messaging evolved from IBM’s 1999 protocol for IoT
Systems EvolutionSystems Evolution to IoTTrace how computing economics and miniaturization enabled the IoT revolution
IoT HistoryIoT History and Paradigm ShiftsIdentify lessons from technology adoption patterns across Internet eras
IPv6 and 6LoWPANIPv6 and 6LoWPANDescribe how modern addressing and adaptation support constrained devices

4.16 Key Takeaway

The history of the Internet shows why layered design, open protocols, interoperability, and fault tolerance matter. IoT systems inherit those lessons when many vendors and networks must work together.

4.17 Continue Your Route

This final part closes the route from Checkpoint: TCP/IP and DNS through Key Takeaway. Return to Internet History: ARPANET and TCP/IP or continue from the fundamentals module index.