Using the internet like it's 1999

· web ai coding · Source ↗

TLDR

  • Joshua Blais argues that RSS (Miniflux), self-hosted XMPP with OMEMO, protocol-layer tools, and POSSE publishing reclaim attention from algorithmic silos.

Key Takeaways

  • Miniflux + RSS is the single recommended replacement for algorithmic feeds: you choose sources, no platform manipulates the queue.
  • Self-hosted XMPP with OMEMO encryption gives E2E group chat and DMs without trusting a host; Matrix is explicitly rejected as solving nothing over XMPP and requiring Electron.
  • POSSE (Publish on Own Site, Syndicate Elsewhere): author content at your own domain first, push to socials via API, never consume them as a reader.
  • Search precision matters: time-bounded queries like before:2025 net/http go language outperform generic keywords on any engine.
  • DNS-level blocklists at the router plus a JS-disabled browser or uBlock Origin are the baseline mitigations recommended for browsing the modern web.

Hacker News Comment Review

  • Commenters immediately flagged the irony: the post advocating 1999-era minimalism loads roughly 1 MB of assets, which would have taken over a minute on a 56k modem.
  • The forward-looking question went mostly unanswered: how do you preserve protocol-layer openness without reimposing 1999-era constraints on bandwidth, discoverability, and UI friction?
  • One commenter pointed to fark.com as a still-running counterexample: human-curated aggregation with a Totalfark subscription model funding the operation without algorithmic amplification.

Notable Comments

  • @Terr_: frames the core failure as the internet delivering corporate-controlled distribution rather than the widely-available capital that lets individuals control machines for their own needs.

Original | Discuss on HN