A public, longitudinal, multi-platform recon system that watches how hate narratives form, moves them from first source through the mainstream, and names the payer wherever the trail permits. Free weekly bulletin. Data cited and dated.
Trace the narratives. Name the payers. Print the figures.
Why we're building this
Online hate is a business. Campaigns pay to spread it. Big accounts amplify it. Platforms get paid to host it. In 2026 we got a real number on the fix: cutting 80 percent of the exposure costs a platform about 2.9 percent of revenue. The fine for leaving it up runs to 10. They know the price. They aren't paying it.
This is part of Hobocode's ongoing reporting on how hate moves online. We laid the argument out in full, every figure sourced, in Online Hate Is a Product Problem Now. It ends with a plain ask: reporters and regulators should stop publishing trend lines and start publishing the payer and the dollar figure. That's the data set we're trying to build.
What it does
A longitudinal, multi-modal, multi-platform recon platform that watches the biggest surfaces in public and does the work most researchers can't do alone.
Ingest public feeds from X, YouTube, TikTok, Reddit, Rumble, select Telegram channels, podcast transcripts, and mainstream news. Classify narratives by claim structure, not by keyword, so coded hate registers alongside the obvious kind.
Temporal clustering across the ingest graph identifies the earliest appearance we can see. Where the true origin sits behind a private channel, we get as close as public data allows, and we say so plainly.
Watch a narrative move from a fringe post, to a creator's show, to a cable segment, to a campaign ad, to a Sunday op-ed. Each hop is timestamped, sourced, and linked, so anyone can walk the same trail.
Impressions where they're published, unique-account touch where they're not, adjusted for the well-documented finding that verified and high-status accounts move hate content roughly 3.5 times farther than the baseline.
Framing shifts, language codes, and cross-narrative fusion (child protection borrowing, replacement language, hidden-hand explanation). Public taxonomies from ADL, GLAAD, and CCDH provide the seed vocabulary.
Which group the narrative is aimed at, when the target changes, which platforms carry it hardest, and what the reachable audience looks like by geography and demographic proxy.
Detect brands appearing against hateful content in the wild. Feeds directly into brand-safety programs that lost their industry standards body when GARM shut down in 2024.
Cross-reference PAC filings, campaign finance data, and corporate disclosures to pin dollar figures to narratives. Where the trail goes dark, we say so and show the last known step.
The public product
Every Monday: three or four narratives rising this week, one being retired, one crossing into mainstream media, one tied to a dollar figure. Short, dense, sourced. Written to be forwarded to a brand-safety lead, a reporter, or a regulator, and understood in a minute.
The bulletin is free. The underlying dataset, API, and enterprise dashboards are what supporting organizations pay for. That mix keeps the mission public and the operation sustainable.
Why now
Who it helps
Support the launch
The Weather Report is being organized as a mission-driven non-profit, so the bulletin stays public and editorial independence is written into the operating documents. Same rule as the essay: every figure verified against a primary or original reporting source.
We're funding a 12-month pilot: build the ingest, ship the first six monthly bulletins, and open the data feed. Support at any level helps. Larger institutional backers are what get this started sooner rather than later.
Governance
Non-profit. The Weather Report is being structured as a mission-driven non-profit. The public Monday bulletin stays free. Paid data feeds and enterprise licenses fund the operation, no single customer sets the agenda.
Editorial independence. Written into the operating documents. Sponsorship and funding never buy narrative selection, framing, or exclusion. Where a funder is themselves in the data, we say so and let another qualified voice write it up.
Sourcing standard. Every figure verified against a primary or original reporting source. Every claim in the bulletin links to the underlying data. Same rule that carried the essay.
Third-party classifications. We measure and report on movement. We do not adjudicate what is hate. Seed taxonomies come from ADL, GLAAD, CCDH, ISD, and the published academic literature.
Related reading
This page draws on our reporting on the blog, including the sourced figures above. New pieces go up regularly.