The specially commissioned research used by Sadiq Khan to call for more internet censorship is full of holes. Who’d have thought…
Sadiq Khan used a speech in Cambridge last month to call for “a new central body” to combat online disinformation backed by GLA research claiming there has been a 350% rise in migration-related narratives and a 150–200% increase in “London in decline” content on X. The Guardian wrote it all up with Khan’s spin…
Guido has dug out the underlying research, produced by one man, which has serious problems:
- Secret methodology: The annex credits unnamed “proprietary communications and mapping tools” operated by an unidentified research firm. The Boolean keyword searches used to generate the headline statistics are not published. The classification criteria for labelling accounts as “coordinated” or “inauthentic” are not disclosed.
- No numbers: The annex claims “London in decline” narratives rose 150–200% while overall London-related X posting rose “only 7%.” It does not provide absolute numbers. A 350% increase from a low baseline in any narrowly-defined keyword search could represent a tiny fraction of London discussion…
- Migration concern rising 350%: Ipsos data from the same period shows public concern about immigration doubled between 2024 and 2026, reaching its highest level in a decade.
- Anecdotal evidence: The annex’s most concrete evidence of disinformation: Bolton footage misattributed to London, a staged TikTok video stripped of context, three monetised X accounts posting the same false New Year’s Eve claim. These are the only genuine examples of misleading content…
- Caveats: The GLA’s own rapid evidence review says the evidence base on interventions is “underdeveloped,” causal effects are “difficult to isolate,” prevalence surveys measure self-reported exposure rather than verified falsehoods, and that promoting media scepticism can backfire by increasing distrust of legitimate sources.
- Bad reporting: The structured sample of platforms beyond X is never quantified with no sample sizes, no platform breakdown, and no selection criteria disclosed.
- Bad actors: Attribution of actors (ERW, Russian-aligned, Beijing-aligned, MAGA-aligned) relies on “consistent analytical criteria” that are never specified, validated only by unnamed “OSINT experts.”
Khan said off the back of this: “We need a new central body with the agility and authority to protect our democracy from disinformation, and deal with the scale and speed of this crisis.” The research paper also recommends using “trusted local messengers” to spread government lines including “health and care workers, faith leaders, community organisations, schools, youth workers and other local networks.” Another recommendation involves pre-emptively exposing people to government-selected examples of manipulation techniques. Might as well shut down the internet…
