Does our “need for speed” make our wi-fi suck?

Original Article ↗ Hacker News Discussion ↗

Wired vs Wi‑Fi for Home Networks

  • Strong consensus: anything that doesn’t move (TVs, consoles, desktops, set‑top boxes, stationary laptops, Sonos, etc.) should be on Ethernet. This frees airtime and dramatically improves Wi‑Fi reliability for phones/tablets.
  • Several argue that “the best way to speed up Wi‑Fi is to not use it” for heavy, continuous loads like video streaming or large model downloads.
  • Some push back that many people can’t or won’t run cable (rentals, apartments, concrete/brick walls, aesthetics), so Wi‑Fi remains the default.

Workarounds for “No Ethernet in the Walls”

  • Suggested options:
    • Flat or color‑matched cables routed along baseboards, crown molding, or via raceways.
    • MoCA over existing coax: widely praised as fast and stable (often ~1 Gbps+).
    • Powerline: highly mixed reports—works well for some, unusable and RF‑noisy for others; strong criticism that it radiates HF interference.
  • Some reuse old phone wiring (Cat3/4‑pair) for 100 Mbps–1 Gbps runs where it happens to work.

Wi‑Fi Configuration, Channel Width, and Roaming

  • Many recommend:
    • 20 MHz on 2.4 GHz, 40 MHz on 5 GHz; wide 80/160 MHz channels often hurt SNR and reliability, especially in dense environments.
    • Avoiding 40 MHz on 2.4 GHz because it occupies half the usable band.
    • More wired APs at lower TX power for better coverage and less co‑channel interference.
  • People report big real‑world gains from narrowing channel width (e.g., from 80 to 20 MHz) in marginal rooms.
  • Roaming remains painful: client‑driven behavior makes devices cling to weak APs; even with 802.11r and multiple APs, seamless roaming is rare in stone/brick houses.

IoT, Segmentation, and Airtime

  • Many households now have 20–60+ Wi‑Fi devices; IoT (plugs, bulbs, cameras, thermostats, etc.) drives the count.
  • Common pattern: put IoT on a separate 2.4 GHz SSID/VLAN, firewall it, and keep laptops/phones on 5 GHz. Rationale: cheap IoT radios use old standards, consume lots of airtime, and may be insecure.
  • Some prefer non‑Wi‑Fi IoT (Zigbee/Z‑Wave) to reduce contention.

Speed vs Responsiveness and Automated Speed Tests

  • Thread agrees with the article that responsiveness/latency often matter more than peak throughput, but users are conditioned to care only about “speed test Mbps.”
  • Skepticism about “automated high‑intensity speed tests” exists, but others confirm: ISP gateways (Xfinity, Verizon), Eero, Nest, DOCSIS modems, and SamKnows boxes do periodic tests—often in‑network, which may limit wider impact.

Ethernet and Standards Evolution

  • Debate on whether consumer Ethernet is “stagnant”:
    • One side: 1 Gbps has been “standard” for ~20 years and many TVs still ship with only 100 Mbps.
    • Other side: 2.5 GbE is now common on new motherboards/routers; 10 GbE and above are affordable used, and far more consistent than any Wi‑Fi.
  • Wi‑Fi 6E’s 6 GHz band and upcoming Wi‑Fi 8 are noted; future standards are said to be shifting from pure peak speed toward reliability, latency (95th percentile), and robustness under interference.