Influence
Part 3  The Weaponized Word
Chapter 69 of 360

Firehose of Falsehood: Volume and Speed as Weapons

In July 2016 the RAND Corporation published a short paper by Christopher Paul and Miriam Matthews describing a model of propaganda they had observed in Russian state media and associated networks. They called it the firehose of falsehood, and its distinguishing feature is that it does not care whether its output is consistent.

The model has four characteristics. It is high-volume and multichannel — the same message arriving through television, wire services, websites, social accounts, commentators and forums, with a large number of apparently independent sources. It is rapid, continuous and repetitive, first to the scene and endlessly repeated afterwards. It is not committed to objective reality, freely fabricating footage, witnesses and documents. And it is not committed to consistency: different outlets, and sometimes the same outlet, will offer mutually exclusive accounts of the same event without any attempt at reconciliation.

The first two characteristics are conventional propaganda, better resourced. The last two are the interesting part, because they violate everything traditional propaganda doctrine holds. Goebbels insisted on a limited number of consistent themes. This model abandons that deliberately.

Paul and Matthews explain why it works with findings from experimental psychology that they lay out explicitly. People judge messages more credible when they come from multiple apparently independent sources, and are poor at detecting that sources are coordinated. Repetition produces acceptance regardless of source credibility, via the illusory truth effect. High volume and speed exploit the primacy advantage — the first account of an event anchors interpretation, and later corrections are weighted less. And people are poor at identifying inconsistency across sources they encounter at different times.

The strategic goal is where this differs from ordinary lying. The firehose is not usually attempting to install a specific false belief. Multiple contradictory accounts cannot all be believed, and are not meant to be. What the volume produces is the sense that the truth about this event is unknowable, that all accounts are equally motivated, and that the effort of working it out is not worth making. Peter Pomerantsev's phrase for the surrounding doctrine is that nothing is true and everything is possible.

The result is not a rival belief. It is disengagement — an epistemic learned helplessness in which the audience stops trying to determine what happened, which serves the propagandist perfectly well, because an audience that believes nothing will also not act.

RAND's defensive recommendations are notable for what they concede. Do not expect to counter the firehose by refuting it: refutation is slower than production and you will lose on throughput. Instead, forewarn audiences before exposure, which is prebunking. Compete for the audience's attention with a positive alternative rather than only a negative correction. Undermine the objectives rather than the individual claims — ask what this flood is intended to achieve, and address that. And apply pressure to the distribution channels themselves.

Every one of those is an admission that the argument cannot be won item by item. Which is the correct assessment, and the same conclusion the next chapter reaches from a much smaller case.

The case

Christopher Paul and Miriam Matthews‘ RAND Corporation report ’The Russian Firehose of Falsehood Propaganda Model' (RAND PE-198, 11 July 2016).

The mechanism

The model abandons consistency and plausibility in favour of high-volume, multichannel, rapid, continuous messaging, because experimental psychology shows that quantity of sources and repetition drive acceptance more than internal coherence. Contradiction is not a bug: it exhausts the target’s limited executive resources and produces epistemic learned helplessness rather than a rival belief. RAND’s key defensive finding is that refutation is slower than production, so the model wins on throughput, not on argument.

What this chapter covers

  1. RAND names the propaganda model
  2. Volume overloads limited executive resources
  3. Paul and Matthews, RAND PE-198 2016
  4. Russian state media and troll networks
  5. Contradictions that are never corrected

Cross-ref: Part 11 information hygiene