From the Vault: Algorithmic Amplification

Tech Policy Podcast1h 0mApril 28, 2026

Get the full intelligence

Search transcripts, export clips, track mentions, and explore all topics from “From the Vault: Algorithmic Amplification” inside PodZeus.

AI-Generated Summary

This episode from the Tech Policy Podcast's vault revisits a 2022 conversation with Stanford scholar Daphne Keller and Ari Cohn of the Time Free Speech Council, examining Keller's influential 2021 paper on algorithmic amplification. The discussion centers on the constitutional challenges of regulating how social media platforms promote content, particularly under the First Amendment. Host Corbin Barthold and guests argue that attempts to regulate 'amplification'—such as demoting or de-amplifying content—are legally indistinguishable from outright censorship, as they restrict the distribution and prominence of speech, which courts treat with strict scrutiny. The episode critiques the 'sound truck' analogy for online amplification, favoring the more nuanced bookstore staff pick analogy, and highlights the epistemological confusion in imagining a 'de-amplified state' as a return to some pure, unmediated truth. The conversation also explores the limits of legal liability, the failure of tort claims against platforms for algorithmic amplification, and the dangers of overregulation that could chill free expression. Ultimately, the guests advocate for indirect solutions—like privacy regulation and middleware competition—to empower users and decentralize content control, rather than top-down legislative mandates. The episode underscores a broader skepticism about the political and policy fixation on content moderation as a panacea for societal ills. The hosts question whether the intense focus on algorithmic amplification reflects a moral panic, noting how the 'current thing' has shifted from 'noxious content' to 'child online safety' over time. They warn against overestimating the power of platforms to solve systemic problems, arguing that social media's influence is often overstated. Instead, they push for structural reforms—such as privacy laws and competitive middleware models—that respect constitutional boundaries while addressing real harms. The discussion ends on a note of cautious optimism, acknowledging that technologists are actively exploring solutions like Twitter’s Project Blue Sky, even as the authors remain wary of overregulation and the 'overproduction of elites' in policy discourse.

Key Takeaways
1

Regulating algorithmic amplification is legally equivalent to regulating speech distribution and thus subject to strict First Amendment scrutiny.

2

The 'sound truck' analogy for online amplification is misleading; the bookstore staff pick analogy better captures personalized recommendations.

3

There is no 'de-amplified state' that serves as a neutral baseline—platform algorithms are necessary to combat inauthentic behavior and spam.

4

Tort liability for algorithmic amplification is unlikely to succeed due to lack of a legal cause of action and the absence of a duty to protect people from harmful speech.

5

Indirect regulation via privacy laws and competition (e.g., middleware) offers a constitutionally sound path to empower users and decentralize content control.

…and 3 more takeaways available in PodZeus

Chapters
0:00
15 min

Introduction and Context: A Time Capsule on Algorithmic Amplification

You should always be wary of giving the government control over speech. So we have the receipts. Here they are. Daphne is principled. Good work.

Highlight
15:10
22 min

The First Amendment Problem: Amplification as Speech Regulation

If lawmakers restrict the distribution... of speech or restrict the prominent placement of speech. The Supreme Court has said those burdens get the same strict scrutiny as outright bans.

Highlight
36:58
25 min

Debunking Analogies and the Epistemological Mess of 'De-Amplification'

The whole idea of a de-amplified state is an epistemological mess.

Highlight
1:01:46
17 min

The Limits of Liability and the Case for Indirect Regulation

The discussion turns to the legal feasibility of suing platforms for algorithmic amplification. Ari Cohn explains that most harms (e.g., emotional distress from a TV show) lack a legal cause of action, and courts consistently reject liability for promoting harmful speech. The conversation shifts to alternative solutions: privacy regulation and middleware. These approaches aim to give users control over their feeds without direct government mandates, thereby sidestepping First Amendment barriers.

1:19:14
19 min

Rethinking the Focus: Is Content Moderation the Right Battleground?

The hosts question whether society is over-investing energy in content moderation debates, framing it as a symptom of moral panic. Corbin Barthold and Ari Cohn suggest that the intense focus on platforms may distract from larger societal issues like climate change, voter rights, and economic inequality. They argue that the rhetoric around content moderation is unproductive and that the real work should focus on process—transparency, due process, and competition—rather than substance.

High-Impact Quotes
The whole idea of a de-amplified state is an epistemological mess.
Daphne Keller36:58
Viral: 90.0
The idea of a de-amplified state is an epistemological mess.
Daphne Keller22:08
Viral: 90.0
We should be very grateful for those First Amendment barriers. You know, we might be sympathetic to wanting a government thumb on the scale that is going to suppress lawful but terrible racist speech, for example. But having that power in the hands of every future administration... I think is something that should scare everybody.
Daphne Keller18:34
Viral: 88.0
Speakers

Host

Corbin Barthold

Guests

Daphne KellerAri Cohn
Topics Discussed
algorithmic amplification95%first amendment and free speech90%middleware and platform competition88%section 230 and platform liability85%content moderation and platform governance80%privacy regulation and data control75%moral panic and policy focus72%epistemological challenges of online discourse70%
People & Brands

Daphne Keller

person

12xPositive

Ari Cohn

person

10xPositive

Corbin Barthold

person

8xPositive

Facebook

organization

6xMixed

Twitter

organization

5xPositive

Digital Services Act

other

4xMixed

Netflix

organization

3xNeutral

Tech Freedom

organization

3xPositive

Stanford University

organization

3xPositive

13 Reasons Why

media

2xNegative

Get the full intelligence

Search transcripts, export clips, track mentions, and explore all topics from “From the Vault: Algorithmic Amplification” inside PodZeus.

Start discovering podcast insights today

Start with a 7-day trial and explore a growing catalog of popular podcasts. No credit card required.

No credit card required • 7-day trial • Cancel anytime