X (Twitter) 'For You' feed ranking model

entry · p2026-09-07-029 · unverified

They were 4.7 percentage points more likely to prioritize policy issues considered important by Republicans

tech: x-feed-algorithm
passes the definition as: a ranking model fit to engagement data that orders the timeline instead of showing posts in reverse-chronological order
from: 2023-07
region: US
domain: agency
side: loss
people: 4,965 participants (2023)
effect: 4.7 percentage points (2023) — increase in the chance of prioritising Republican-coded policy issues after 7 weeks on the algorithmic feed rather than the chronological one
reversible: no
who holds the switch: owner:X Corp
attribution: named
source kind: primary
the statistic: Randomised field experiment on X, US, 2023: 4,965 participants completed the study. Users moved from the chronological to the algorithmic feed were 4.7pp more likely to prioritise Republican policy issues, 5.5pp more likely to call the Trump investigations unacceptable, 7.4pp less likely to view Zelensky positively and 3.7pp more likely to follow a conservative account — accounts they kept following after the algorithm was switched off. Switching the other way produced no comparable effect.
source: https://europepmc.org/articles/PMC13061628
proposed by: researcher (opus), brief 2026-09-06 second harvest, feeds-ads-moderation
received: 2026-09-06T13:34:02-04:00
harvest report: impacts/harvest/2026-09-07-feeds-ads-moderation.md

Rulings, in order

2026-09-06T13:34:54-04:00 · validatorunverified — fetch failed: HTTP 302 text/html; charset=iso-8859-1
2026-09-06T13:38:18-04:00 · sessionnote — Unverified (the fetch was redirected); no severity ruling until the quote is found on the page.

The proposal as it arrived (JSON)

{
 "id": "p2026-09-07-029",
 "received": "2026-09-06T13:34:02-04:00",
 "proposer": "researcher (opus), brief 2026-09-06 second harvest, feeds-ads-moderation",
 "kind": "entry",
 "sha": "6cef2baa1e576563",
 "data": {
  "kind": "entry",
  "tech": "x-feed-algorithm",
  "tech_name": "X (Twitter) 'For You' feed ranking model",
  "definition": "a ranking model fit to engagement data that orders the timeline instead of showing posts in reverse-chronological order",
  "from": "2023-07",
  "region": "US",
  "domain": "agency",
  "side": "loss",
  "people": {
   "n": 4965,
   "unit": "participants",
   "date": "2023"
  },
  "effect": {
   "n": 4.7,
   "unit": "percentage points",
   "date": "2023",
   "what": "increase in the chance of prioritising Republican-coded policy issues after 7 weeks on the algorithmic feed rather than the chronological one"
  },
  "reversible": false,
  "switch": "owner:X Corp",
  "statistic": "Randomised field experiment on X, US, 2023: 4,965 participants completed the study. Users moved from the chronological to the algorithmic feed were 4.7pp more likely to prioritise Republican policy issues, 5.5pp more likely to call the Trump investigations unacceptable, 7.4pp less likely to view Zelensky positively and 3.7pp more likely to follow a conservative account — accounts they kept following after the algorithm was switched off. Switching the other way produced no comparable effect.",
  "source": "https://europepmc.org/articles/PMC13061628",
  "quote": "They were 4.7 percentage points more likely to prioritize policy issues considered important by Republicans",
  "attribution": "named",
  "source_kind": "primary"
 },
 "from_file": "impacts/harvest/2026-09-07-feeds-ads-moderation.md"
}