November 30, 2025
Numbers vs Beige
A Second Look at Geolocation and Starlink
Did Starlink conquer Yemen? Commenters yell math fail and mock a beige stripe
TLDR: Huston reexamines Yemen’s Starlink numbers, arguing ad‑based measurements likely overcount shared or roaming connections. Commenters clash over a 2.3M vs 8M user gap and crack jokes about a beige CSS bug, highlighting how messy counting the internet can be—and why it matters.
Geoff Huston took a second look at a wild claim: Yemen might have up to 7 million Starlink users, which would be most of the country’s internet. He says that’s implausible, and walks through possible reasons the data went off the rails—like ships and planes using Starlink while passing by, cross‑border roaming, or one satellite dish feeding an entire hotspot so ad systems count every phone behind it. He also flags how civil unrest can make measurements messy, since the method here aligned ad impressions with market share. Translation for non‑nerds: counting ads to guess users can be a shaky move. The comments? Absolute popcorn material. User wmf points out Huston’s global estimate of 2.3 million Starlink users vs SpaceX’s claimed 8 million customers—cue math war over whose numbers count: browsers, dishes, or paid accounts. Some joke that the Yemen count is just “ship Wi‑Fi doing laps,” while others argue that hotspot‑style sharing can make one dish look like an army of users. And then daemonologist strolls in with the ultimate plot twist: the site itself has a “beige stripe” breaking the page, offering a DIY CSS fix and turning a data debate into a front‑end roast. The vibe is half statistical showdown, half “pls fix your website,” and all Internet drama. Read Huston’s piece here and SpaceX’s claims here.
Key Points
- •Ad-based measurement suggested 6–7 million Starlink users in Yemen, which the author deems implausible given national user base and GDP.
- •Starlink maps IP addresses to countries, including those used on ships and aircraft in international areas, potentially skewing geolocation.
- •Starlink permits roaming, but service in unauthorized locations may be limited or terminated; enforcement details are unclear.
- •Community redistribution hotspots can cause many end users behind a single Starlink service to be counted as individual users in ad-based measurements.
- •Civil unrest and methodological assumptions (stability, uniform ad presentation across ISPs) may bias market share estimates, prompting a re-examination of Yemen data.