The New Internet

Original Article ↗ Hacker News Discussion ↗

A blog post from Tailscale pitching its mesh-VPN platform as the basis for a “new internet” sparked debate over whether this vision reduces or merely recentralizes power on the network. Commenters compare Tailscale’s overlay model to long-promised but unevenly adopted standards like IPv6, IPsec and DNSSEC, arguing over whether the real blockers to end‑to‑end connectivity are technical, economic, or social. Many praise Tailscale’s usability and self‑hosting options (e.g. via Headscale) but express unease about dependence on a proprietary coordination layer and the broader incentive for such companies to keep underlying internet problems unsolved.

Reactions to the “New Internet” framing

  • Many find the essay well-written but see the “New Internet” branding as grandiose or marketing for an IPO/acquisition ramp.
  • Some view the conclusion as “old landlord bad, new landlord us,” noting the irony of denouncing rent‑seeking while proposing Tailscale as the new gatekeeper.
  • Others see it as a normal “platform + killer app” strategy pitch and not uniquely sinister.

Tailscale’s value and real‑world friction

  • Strong appreciation for “it just works” UX compared to raw WireGuard/OpenVPN; especially for small teams and homelabs.
  • Concrete complaints: confusing onboarding with family accounts and ACLs, non‑obvious behavior when inviting users, Windows firewall/ICMP issues, and performance/battery problems on some clients.
  • Some note that for serious remote support, tools like TeamViewer/AnyDesk can be more straightforward.

Centralization, control planes, and rent

  • Core worry: solving NAT/connectivity via a proprietary coordination layer recentralizes power; Tailscale becomes another chokepoint.
  • Self‑hosting Headscale is seen as a partial answer, but people question how long official clients will support third‑party controllers.
  • Several point out that Tailscale requires external identity providers (OIDC), which itself reinforces dependence on big platforms.

IPv6, NAT, and protocol design debates

  • Large sub‑thread debates IPv6: some call it a design mistake and blame poor adoption; others argue “IPv6 has already happened” in mobile/large networks and NAT is only tolerable because IPv6 offloads pressure.
  • Disagreement over NAT as “security”: one side treats it as essential isolation; the other calls that security theater, saying real protection comes from firewalls and zero‑trust, and IPv6 can be filtered similarly.
  • Broader point: Tailscale is a pragmatic overlay exploiting the messy, NAT‑heavy IPv4 world rather than fixing the underlay.

Self‑hosting and home networks

  • Some are energized by a trend toward self‑hosted services (photos, media, home labs) with Tailscale‑like mesh as glue; envision “family network citadels” run by one technical person.
  • Others have moved everything to big clouds to avoid homelab maintenance toil; see local hosting as work they don’t want at home.

Alternatives and open ecosystem

  • Multiple overlay contenders mentioned: ZeroTier, Nebula, NetBird, OpenZiti, tinc, yggdrasil, plain WireGuard plus tooling, etc.
  • View that the real issue is proprietary vs non‑proprietary overlays; Tailscale is praised as a polished option but not technically unique.