IPv6 just turned 30 and still hasn't taken over the world

Adoption: “Failed” vs “Everywhere”

• Thread splits between “IPv6 is a failure” and “IPv6 is already half the internet.”
• Several participants cite stats (Google, APNIC) and operational experience: ~50% of Google traffic, most mobile networks, many home ISPs and big clouds use IPv6 heavily, sometimes IPv6‑only with IPv4 via NAT64/464XLAT.
• Others note their own devices or ISPs have no IPv6, so they perceive it as dead; people conflate “I don’t see it” with “it isn’t used.”

Enterprise, VPNs, and Address Planning

• Many enterprises and data centers still run IPv4‑only inside, using RFC1918 (especially 10/8) plus NAT.
• Engineers describe real scaling problems: overlapping 10/8 ranges across business units, mergers, and inter‑company VPNs causing painful renumbering, async routing, and nested NAT.
• Some large orgs have exhausted public /8s and private space and are kludging with additional “reserved” ranges; IPv6 is seen as “distant aspiration.”

NAT, CGNAT, and Peer‑to‑Peer

• NAT is widely (mis)treated as a security feature and privacy layer; many home admins explicitly prefer “NAT as default-deny.”
• Others stress that security comes from firewalls, not translation, and that CGNAT breaks or complicates P2P, VoIP, gaming, and self‑hosting.
• Some see CGNAT as accidental privacy (many users per IPv4), others call it “the devil” and expect ISPs to monetize logs and IPv4 scarcity.

ISPs, Prefixes, and Practical Pain

• Experiences vary wildly: some ISPs give stable, generous prefixes (/56 or /48) and working PD; others offer no IPv6, only CGNAT, or dynamic /64s that change frequently.
• Dynamic IPv6 prefixes break self‑hosting and firewall rules; workarounds (ULA, dynamic DNS, NAT66) are considered clunky.
• A number of users report ISPs that once deployed IPv6 then quietly turned it off.

Complexity, Design Choices, and Alternatives

• Critics argue IPv6 suffers from “second‑system effect”: 128‑bit addresses, SLAAC, privacy extensions, multiple address types, and uneven DHCPv6 support (notably on Android) make it feel over‑engineered.
• Others say at layer 3 it’s simpler than IPv4 (fixed header, straightforward subnetting, plenty of space) and that most complexity is operator self‑inflicted.
• Long subthreads debate “IPv4+”: adding extra bits or options while keeping dotted‑quad semantics; defenders of IPv6 say any real extension would have required essentially the same dual‑stack migration.

Privacy, Security, and Centralization

• Some disable IPv6 entirely, fearing easier device‑level tracking and loss of “casual anonymity” from shared IPv4 and CGNAT.
• Counterpoint: both stacks are equally subpoena‑able, browser fingerprinting dominates tracking, and IPv6 privacy extensions randomize interface IDs.
• Several argue that lack of end‑to‑end reachability (NAT, CGNAT) helped push the ecosystem toward centralized cloud platforms and away from true peer‑to‑peer.