The Availability Cascade

What it is

The loop needs an accelerant, and it has a name for it. Availability entrepreneurs are individuals and organisations that work to keep the flow of worrying news going, because the alarm serves them. Some are sincere, some are not, and the mechanism does not care which. Anyone who dissents at any point in the cycle is suspected of being in the pay of whoever is being blamed, which is why cascades rarely stop from the inside.

This is the availability heuristic run as a social process rather than an individual one. What comes to mind easily is not decided inside a head; it is decided by what has been in circulation. Worth recording that Thinking, Fast and Slow gives Sunstein the lead billing in its prose, discussing him at length and mentioning Kuran more briefly, while the idea belongs to the pair. Credit them together and put Kuran first.

The relationship to moral panic is close and not identical. A moral panic needs a folk devil, a group cast as morally rotten, and its criteria are sociological. An availability cascade needs only a risk and a feedback loop, and its mechanism is cognitive. Most techno-panics are both. An availability cascade about a chemical is not a moral panic, because a chemical cannot be a folk devil.

In effect

Kahneman gives two cases and flags both as still contested himself, which is unusual enough to be the most important thing about them.

Love Canal in 1979: buried toxic waste in a New York neighbourhood, heavy rain, an evacuation, national coverage, and a public alarm that produced federal law, giving the United States the Superfund programme. Kahneman's own line is careful, that claims of actual damage to health appear not to have been substantiated, and he records that opinions on the episode remain sharply divided. He is not saying nothing happened. He is saying the health case for the size of the response was not made and that the question is open.

Alar in 1989: a chemical sprayed on apples to regulate growth, reported as a carcinogen. Kahneman's account is that the cancer findings involved gigantic doses far beyond any realistic exposure, that congressional testimony followed, that the regulator banned the product, and that the net public-health effect was probably detrimental because the panic put people off apples.

The Alar case carries a documented disagreement. Paul Slovic put to Kahneman directly, in a personal communication dated May 2011 and reported in the book, that the public reaction was rational given what was broadcast and given the regulator's failure. People were told by credible sources that a chemical on children's food caused cancer, and the agency responsible had not settled the question before it reached them. Reacting strongly to that is not a cognitive failure, it is the correct response to the information available, and the failure sits upstream with the institutions that let it arrive in that form.

Both readings can be true at once and largely are. Kahneman is right about the toxicology: dose matters, and a finding at gigantic exposures does not licence the alarm that was raised. Slovic is right about the epistemics: you cannot call a public irrational for believing what its regulator and its broadcasters have jointly told it. The useful position is that a cascade is an institutional failure that presents as a public failure, which is also why blaming the audience is both the standard move and the wrong one.

What happens at the end of the cascade has its own name, probability neglect, which is Sunstein's. Once a risk is emotionally vivid the probability attached to it stops entering the calculation at all. It is not that people overweight the small number, it is that the number stops being consulted, which mirrors the instability in prospect theory's decision weights at the very low end of the probability range.

What it does not say

It does not report an effect. There is no effect size, nothing here replicates or fails to replicate, and it should be cited as an idea with an attribution rather than as a finding.

It does not settle either case study. Neither Love Canal nor Alar is settled history, both are marked contested on the author's own say-so, and any use of them has to carry that. No independent source has been consulted here, and no dose figure, cancer incidence figure or consumption figure has been invented.

It does not identify who the availability entrepreneurs are in any given case. The mechanism describes a role, not a conspiracy, and reading it as a claim that alarm is always manufactured by identifiable interests goes well past what the concept asserts.

It does not carry its own citation. Kuran and Sunstein are credited jointly and no paper, journal or year is recorded in this vault, and none has been invented. The same is true of probability neglect.

---