Built by No One, Run by Everyone: A Deep Dive into Internet Governance and Routing

How the Internet Actually Works: From Vint Cerf’s TCP/IP to the IP Address on Your Phone

We use “the internet” like it’s one single thing, but it’s really a layered system: a handful of global organizations setting policy, a tiered system of regional and national registries handing out address space, undersea cables physically carrying the data, and routing protocols deciding which path your data takes across thousands of independently-run networks. This post walks through the entire stack — history, governance, and the actual physical/logical journey of a packet — in one place.

Where It All Began: ARPANET and TCP/IP

The story starts in 1969 with ARPANET, a U.S. Department of Defense project (via ARPA, the Advanced Research Projects Agency) that became the first network to use packet switching — breaking data into small packets, sending them independently, and reassembling them at the destination. This design choice is why the internet is decentralized and fault-tolerant to this day.

In the 1970s, Vint Cerf and Robert Kahn co-invented TCP/IP (Transmission Control Protocol/Internet Protocol), the protocol suite that let separate, incompatible networks interconnect and exchange data reliably. This breakthrough is what turned a collection of isolated networks into an actual “network of networks” — the internet as we know it. Cerf is widely credited as one of the “fathers of the internet” for this work, and later went on to help found the Internet Society.

The Standards Bodies: Who Decides How Things Work

Three organizations are worth knowing here, and they’re often confused with each other:

• IETF (Internet Engineering Task Force) — the open, volunteer-driven body that develops the actual technical protocols and standards (HTTP, TCP/IP refinements, etc.), published as RFCs (Request for Comments).
• IEEE (Institute of Electrical and Electronics Engineers) — a much broader global engineering body, not internet-specific, but responsible for standardizing key hardware/link-layer technologies like Ethernet (802.3) and Wi-Fi (802.11).
• ISOC (Internet Society) — a nonprofit founded in 1992, with Cerf as a founding figure, that promotes an open internet and serves as the organizational home for IETF.

These bodies define how things should technically work. They don’t allocate addresses — that’s a separate chain of organizations.

The Address Allocation Hierarchy: Who Gives You an IP

No single company owns IP address space. It flows down through a strict hierarchy:

ICANN → IANA → RIR → NIR → ISP → End User

• ICANN (Internet Corporation for Assigned Names and Numbers) — a California-based nonprofit (founded 1998) that sets global policy for domain names, IP address allocation, and protocol parameters.
• IANA (Internet Assigned Numbers Authority) — a function operated by ICANN. This is the actual operator: it maintains the DNS root zone, assigns protocol/port numbers, and allocates large IP blocks to five Regional Internet Registries.
• RIR (Regional Internet Registry) — five regional bodies split the world: APNIC (Asia-Pacific), ARIN (North America), RIPE NCC (Europe/Middle East/Central Asia), LACNIC (Latin America/Caribbean), and AFRINIC (Africa). They manage allocation within their region.
• NIR (National Internet Registry) — a country-level layer under an RIR. India’s NIR is IRINN (Indian Registry for Internet Names and Numbers), operated by NIXI, which allocates IPs and AS numbers to Indian ISPs and organizations.
• ISP (Internet Service Provider) — gets blocks from the NIR/RIR and distributes individual addresses or smaller blocks to end users.
• End User — your device, typically getting a dynamic IP via DHCP from your ISP.

How IP Blocks Actually Get Split Up (Subnetting & CIDR)

Addresses aren’t handed out one at a time — they’re distributed in blocks using CIDR notation (e.g., 203.0.113.0/24), where the number after the slash indicates how many bits are fixed for the network portion, leaving the rest for individual devices.

A typical allocation chain looks like this:

1. IANA gives APNIC a huge block — say a /8 (~16.7 million addresses).
2. APNIC gives IRINN, or a large ISP directly, a /16 (~65,000 addresses).
3. IRINN gives an ISP like Jio or Airtel a /20 or /22, sized to their customer base.
4. The ISP gives a business a smaller block — maybe a /28 (16 addresses) — or a single address to a home user.

This nested splitting (a /16 breaks into 256 /24s, each /24 breaks into smaller /28s, etc.) means organizations of any size get an appropriately sized chunk without wasting address space. Companies also use subnetting internally to segment networks — separate IP ranges for, say, finance, engineering, and guest Wi-Fi — for security and traffic management.

Companies and servers generally get static IPs (fixed, since they need to be reliably reachable), while home users typically get dynamic IPs, reassigned periodically via DHCP since they don’t need a fixed address.

How Data Actually Crosses the World

Allocation is the logical layer. Here’s the physical and routing layer that makes global communication work.

Submarine cables. Despite the popular image of satellites, more than 95% of international internet traffic travels through undersea fiber-optic cables, often laid by consortiums of ISPs and tech giants like Google and Meta. Data moves as light pulses through fiber at near light-speed.

Internet Exchange Points (IXPs). These are physical data centers where multiple ISPs connect directly to exchange traffic locally instead of routing it internationally. In India, NIXI operates IXPs in cities like Mumbai, Delhi, and Chennai so domestic traffic stays domestic — cheaper and faster than sending it abroad and back.

Tiered ISP structure.

• Tier 1 ISPs (e.g., Lumen, NTT, Tata Communications) own the global backbone and peer directly with each other, paying no one for transit.
• Tier 2 ISPs (e.g., Jio, Airtel) peer with some networks but pay Tier 1s for transit to reach distant parts of the world.
• Tier 3 ISPs are local, last-mile providers buying connectivity from Tier 2s to deliver to homes and businesses.

BGP (Border Gateway Protocol). Once your data is on the network, BGP decides the actual path packets take across thousands of independently-operated networks, called Autonomous Systems (each with its own ASN, also allocated by RIRs/NIRs). BGP routers continuously exchange information about the best-known paths to every IP block, automatically rerouting around failures — which is why the internet keeps working even when a cable gets cut.

Packet switching. True to its ARPANET roots, data is broken into packets, each tagged with destination info, sent independently (sometimes via different routes), and reassembled at the destination. This is the core design that makes the internet decentralized and resilient.

Putting It All Together: The Full Journey of a Request

1. You type a website address. Your device needs an IP, resolved via DNS (also overseen by ICANN/IANA).
2. Packets leave your device through your home router, using your ISP-assigned IP.
3. They travel up through your ISP’s network (Tier 3 → Tier 2).
4. If the destination is far away, packets pass through an IXP and onto a Tier 1 backbone, potentially via a submarine cable.
5. BGP routing determines the actual path across multiple Autonomous Systems.
6. Packets arrive at the destination server’s ISP, then the server itself — which has its own static IP from its own allocation chain.
7. The server responds, and packets retrace a path (not necessarily the same one) back to you.

A Few Extra Things Worth Knowing

• IPv4 exhaustion. IPv4 (the original 32-bit system, ~4.3 billion addresses) ran out of fresh allocations at the RIR level years ago — a major reason IPv6 (128-bit, effectively unlimited) exists, though global adoption is still gradual.
• ASNs and BGP go together. Every ISP or large network needs an ASN to participate in BGP routing, and these are allocated through the same RIR/NIR chain as IP addresses.
• ICANN vs. IETF, one more time. ICANN handles naming/addressing policy and allocation. IETF handles the technical protocol standards everything runs on. They’re complementary layers of the same overall system, not competitors.
• NIXI does double duty. Beyond running IRINN (India’s NIR), NIXI also operates the country’s IXPs, keeping a meaningful share of India’s internet traffic routed domestically.

None of this is centrally “run” by any one company or government — it’s a deliberately distributed system of policy bodies, regional registries, ISPs, and physical infrastructure that, layer by layer, turns a request typed into a browser into a packet that can find its way anywhere on Earth and back.