The Internet does not begin with websites.

Websites are palace decorations. The real state is below them, stamping papers at the border.

That state is IP, the Internet Protocol.

IP is the regime that answers three questions:

• where is this packet going?
• where did it come from?
• how long may it wander before being executed?

Everything else rides on top. BGP announces where prefixes live. DNS turns names into addresses. NTP keeps the clocks from embarrassing the certificates. TCP, UDP, QUIC, HTTP, SSH, and every dashboard pretending to be infrastructure all depend on the same primitive ritual:

put a destination address on the packet and hope the republic of routers agrees.

I. What IP Actually Does

IP is a network-layer datagram protocol.

This means it does not promise delivery. It does not promise order. It does not promise uniqueness of suffering.

It offers a packet format, addressing, fragmentation rules, hop limits, and enough metadata for routers to forward traffic one step closer to a destination.

The political model is simple:

Application:  "send this request"
Transport:    "I will handle ports and reliability, maybe"
IP:           "I know the destination address"
Router:       "I know the next border crossing"
Ethernet/WiFi: "I know the next local citizen"

IP is not a delivery company. IP is the passport system.

The routers do not know your app. They do not know your feelings. They look at the destination address, consult forwarding tables, decrement a lifetime counter, and push the packet toward the next checkpoint.

If the packet dies, the state may or may not send a death certificate through ICMP.

II. The Four Bits That Caused The Succession Crisis

The first field in an IP packet header is the version.

It is only 4 bits.

That gives sixteen possible protocol version numbers: 0 through 15.

The Ministry of Numbered Things, known in the West as IANA’s IP Version Numbers registry, keeps the official table.

This table is much funnier than civilians expect.

Version	Official status	Political interpretation
0-1	Reserved	the throne room is sealed
2-3	Unassigned	early ghosts did not receive uniforms
4	Internet Protocol, RFC 791	the empire that actually won
5	Reserved historic, Stream Protocol lineage	the name was burned before the masses arrived
6	IPv6, RFC 8200	the official successor
7	Reserved historic, TP/IX	a claimant from the standards bunker
8	Reserved historic, PIP	already haunted before modern IPv8 appeared
9	Reserved historic, TUBA	bigger addresses by committee fossil
10-14	Unassigned	empty chairs in the Politburo
15	Reserved	emergency glass, do not break

The important lesson:

there is no clean dynasty.

There is IPv4, IPv6, several historic bodies in the wall, and now a modern IPv8 draft waving paperwork at the guard.

III. IPv4: The Empire

IPv4 is the protocol most people mean when they say “IP.”

It was specified in RFC 791 in September 1981, after earlier definitions such as RFC 760. This makes IPv4 older than both Kim Jong Rails and Kim Jong Un, a fact that explains why it still believes every machine should know its place.

It gave the world 32-bit addresses, written in dotted decimal:

192.0.2.10
198.51.100.23
203.0.113.7

Thirty-two bits gives about 4.3 billion possible addresses.

In 1981, this sounded enormous.

This was more addresses than there were humans alive at the time. The planners were not fools. They looked at the planet, counted the mammals with passports, and concluded that four billion network addresses should hold the republic for a while.

This was reasonable only because in 1981 nobody had yet assigned IP addresses to refrigerators, doorbells, cloud containers, televisions, watches, cars, thermostats, printers, game consoles, Kubernetes pods, and one suspicious fish tank thermometer in a casino.

IPv4 looked like this:

0                   1                   2                   3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+---+---+-----------+-------------------+---------------------------+
|Ver|IHL| DSCP/ECN  |    Total Length   | Identification            |
+---+---+-----------+-------------------+---------------------------+
| Flags | Fragment Offset | TTL | Protocol | Header Checksum          |
+-------------------------+-----+----------+--------------------------+
| Source Address                                             |
+------------------------------------------------------------+
| Destination Address                                        |
+------------------------------------------------------------+

The design was practical:

• 32-bit source and destination addresses
• header checksum
• Time To Live, now functionally a hop limit
• Protocol field identifying payloads like TCP, UDP, or ICMP
• fragmentation support inside the network
• options, because committees need a drawer for dangerous objects

IPv4 won because it was useful, deployable, and part of the TCP/IP suite that displaced rival networking stacks.

Then it ran out of addresses.

The empire did not collapse. It discovered bureaucracy.

IV. NAT: The Apartment Block With One Doorbell

IPv4 exhaustion should have forced a clean migration.

Instead, the Internet invented increasingly elaborate ways to pretend 32 bits were enough.

The main survival tools were:

• CIDR: stop wasting address blocks through classful allocation
• RFC 1918 private space: allow internal networks like 10.0.0.0/8
• NAT: hide many private hosts behind one public address
• CGNAT: make the ISP hide many customers behind shared public addresses

NAT is the great apartment-block deception.

Inside the building, everyone has a room number. Outside the building, the world sees one street address. The doorman keeps a translation table and hopes nobody asks philosophical questions.

inside host          NAT table                    public internet
10.0.0.23:51544  ->  203.0.113.9:40001  ----->    93.184.216.34:443
10.0.0.24:51545  ->  203.0.113.9:40002  ----->    93.184.216.34:443
10.0.0.25:51546  ->  203.0.113.9:40003  ----->    93.184.216.34:443

This worked so well that it damaged history.

NAT solved the immediate crisis while making the proper solution feel optional.

The Republic recognizes this tactic. It is how every temporary emergency power becomes a permanent ministry.

V. IPv5: The Name That Was Already Occupied

People ask:

“If IPv4 was followed by IPv6, what happened to IPv5?”

The answer is not that engineers cannot count.

The answer is that protocol version 5 was used by the experimental Stream Protocol lineage, including ST and ST2 material later marked historic. It was not the next public Internet Protocol for ordinary packets.

IPv5 is therefore a perfect bureaucratic corpse:

• important enough to reserve the number
• not important enough to become the Internet
• famous mainly because everyone asks where it went

The Supreme Leader respects this outcome.

Many ministers serve the state best by becoming footnotes.

VI. IPv6: The Official Heir

IPv6 is the legitimate successor.

RFC 8200 specifies IPv6 as the modern Internet Standard, replacing the earlier RFC 2460 specification. Its headline change is address expansion from 32 bits to 128 bits.

That is not “more addresses.”

That is “we are no longer pretending the village registry can run an empire.”

IPv6 addresses look like this:

2001:db8:582:ae33::29
2001:db8::1
fe80::1

IPv6 changed more than address size:

Area	IPv4	IPv6
Address size	32 bits	128 bits
Broadcast	yes	no broadcast; multicast/anycast instead
Header checksum	in IP header	removed from IP header
Fragmentation	routers and hosts may fragment	source handles fragmentation; routers do not fragment transit packets
Local discovery	ARP	Neighbor Discovery over ICMPv6
Configuration	manual, DHCP, later extras	SLAAC, DHCPv6, Router Advertisements
Options	IPv4 header options	extension headers

This was technically coherent.

It was also comically large.

IPv6 has 2^128 possible addresses, roughly 340 undecillion. This is not a numbering plan. This is a cosmic land reform program.

If IPv4 gave the planet one cramped apartment block, IPv6 looked at the observable universe and said:

“Every dust particle may apply for residency. Bring two forms of identification.”

It was also operationally inconvenient.

IPv6 required operators, vendors, applications, firewalls, monitoring systems, training material, embedded firmware, load balancers, VPNs, and corporate procurement departments to behave like adults at the same time.

This did not occur.

Instead, the world deployed dual-stack, tunnels, translation, NAT64, DNS64, 464XLAT, carrier-grade NAT, and many diagrams that look like a committee trying to confess.

IPv6 is not dead. It is widely deployed and necessary.

But IPv4 remains alive because old protocols do not die when the standard says so. They die when the last billing system, printer firmware, VPN appliance, and industrial control box stops needing them.

The year of that event is classified because it has not been invented.

VII. IPv7, IPv8, IPv9: The Historic Claimants

The version-number registry preserves three other historic claimants:

• IPv7: TP/IX, referenced through historic next-generation work
• IPv8: PIP, the P Internet Protocol
• IPv9: TUBA, TCP and UDP with Bigger Addresses

These were not mass-adopted Internet successors.

They were artifacts from the succession crisis around “what comes after IPv4?”

The 1990s had many proposals because the problem was obvious:

IPv4 addresses would not last forever.

The solution was not obvious because every possible successor had to fight physics, politics, hardware, routing-table growth, host stacks, application assumptions, and the terrifying sentence:

“We need everyone to upgrade.”

This is why the registry looks like a royal cemetery.

VIII. Modern IPv8: The New Claimant

Now we arrive at the comedy item.

There is a modern Internet-Draft called Internet Protocol Version 8, draft-thain-ipv8-02, last updated on April 17, 2026.

Important detail:

the IETF Datatracker says plainly that anyone may submit an Internet-Draft, and that this draft is not endorsed by the IETF and has no formal standing in the standards process.

That sentence is the difference between:

• “the Internet has a new protocol”
• “a document entered the building”

The draft proposes a managed IPv8 suite with 64-bit addresses, an ASN-based routing prefix plus host field, Zone Server concepts, DNS8, WHOIS8, BGP8, OSPF8, ARP8, ICMPv8, and claims of IPv4 backward compatibility.

Its basic addressing idea can be summarized like this:

IPv8 draft address shape:

r.r.r.r.n.n.n.n

r.r.r.r = 32-bit ASN routing prefix
n.n.n.n = 32-bit host address

IPv4 subset representation:
0.0.0.0.n.n.n.n

This is fascinating.

It is also not deployed Internet reality.

Routers do not obey draft energy. ASICs do not forward aspirations. Middleboxes do not wake up and say, “I have read your individual submission and accept version 8 packets.”

Also, version number 8 is already marked historic in IANA’s registry because of PIP. The draft asks IANA to assign version 8 to modern IPv8, but asking the palace for the throne is not the same as sitting on it.

The Supreme Leader respects the ambition. The Supreme Leader also checks the registry.

The unofficial summary is simpler:

IPv8 is what happens when someone looks at IPv6 and says:

“What if we rewrote this in Rust?”

Not literally Rust, of course. The draft is protocol text, not a Cargo workspace. But the energy is familiar: take a deployed old system, declare the correct successor insufficiently modern, introduce stricter ownership of the address space, add new management services, and imply that the old pain existed because the previous generation lacked discipline.

Every era has this instinct.

In operating systems, it says “rewrite the kernel.” In databases, it says “rewrite PostgreSQL.” In networking, it says “rewrite IPv6, but with a registry, zones, and better life choices.”

The packet remains unimpressed until silicon forwards it.

IX. Why IPv4 Still Rules The Palace

IPv4 survived because the Internet is not a clean software project.

It is a global installed base with money attached.

IPv4 has:

• decades of operational knowledge
• hardware fast paths
• firewall assumptions
• legacy embedded devices
• cloud primitives
• cheap tutorials
• NAT workarounds
• application compatibility
• the psychological advantage of looking familiar

IPv6 has:

• the correct address space
• cleaner neighbor and multicast model
• mandatory relevance for modern networks
• fewer excuses every year
• the curse of being correct before everyone is ready

Modern IPv8 has:

• a draft
• ambition
• a numbering problem
• the energy of a governor announcing a new capital city before the roads exist

This is the succession map:

Protocol	Status	Kim’s evaluation
IPv4	dominant legacy reality	old dictator still controlling the army
IPv5	historic stream experiment	name burned in a classified hallway
IPv6	official successor	rightful heir, slow coronation
Historic IPv7/8/9	failed or abandoned claimants	royal cemetery of next-generation IP
Modern IPv8 draft	individual Internet-Draft	new claimant waving papers at border control

X. The Real Story (Suppressed)

Officially, IP means Internet Protocol.

Unofficially, the Ministry of Packet Affairs records the original expansion as:

Imperial Passport.

Every packet required papers. Every router was a checkpoint. Every TTL decrement was a border guard asking, “how many countries have you already crossed?”

IPv4 was supposed to be temporary enough to evolve. Then it became too successful to replace cleanly.

IPv6 was designed as the heir. Then the old king discovered NAT and refused to leave the balcony.

IPv5 disappeared into streaming history. IPv7, IPv8, and IPv9 entered the standards catacombs. Modern IPv8 arrived later with a new uniform and declared:

“The number was only mostly dead.”

The registry coughed. The routers continued forwarding IPv4. The IPv6 operators sighed in hexadecimal.

Somewhere in Pyongyang, a border router accepted exactly one route-map and trusted no one.

XI. The Lesson

IP is not interesting because packets have addresses.

IP is interesting because an addressing scheme became civilization.

IPv4 proved that simple, deployable systems can conquer the world before anyone notices the long-term costs.

IPv6 proved that technically correct successors still need operational politics.

IPv5, IPv7, historic IPv8, and IPv9 proved that version numbers are not a product roadmap.

Modern IPv8 proves that in networking, a draft can be entertaining, ambitious, and technically provocative while still not being the Internet.

The decree is simple:

do not confuse numbering with succession.

The packet does not care who claims the throne. The packet cares which routers forward it.

— Kim Jong Rails, Supreme Leader of the Republic of Derails