The Clapping Stopped - Decipon Insights

The Number

On June 17, 2026, Laura Loomer posted a video to X. In it, she said she had fallen for Russian propaganda. She described how RT had reached out to her during her years of deplatforming, offered her an audience when no American platform would have her, and how she had, over time, started advancing the narratives that served their interests. She said she never took money. She said she had been to Ukraine and seen what Russia had done.

In the five days following the post, her X account received 751,000 likes, comments, and reposts.

In the three five-day periods before the post, she had averaged 1.05 million.

That is a 28 percent drop. The drop was sudden. It did not taper over weeks as a disillusioned audience slowly peeled away. It fell off a cliff within days of a single post.

Researcher Matthew Kupfer noted the obvious implication: “Such a sudden implosion does not suggest these were authentic users.”

He is right. And that number — 28 percent, five days — is not just a data point about Laura Loomer. It is a measurement of an influence operation’s infrastructure, accidentally exposed.

How the Loop Works

Loomer was first permanently banned from Twitter in 2018 for anti-Muslim tweets. Over the following years, she was removed from Facebook, Instagram, payment processors, ride-share apps. The deplatforming was broad and, from her perspective, total.

During this period, RT reached out. Russian state media had a standing practice of recruiting Western political figures who had been marginalized by mainstream platforms — people with established audiences, legitimate grievances about censorship, and no institutional relationship that could prohibit appearances. Loomer fit the profile. She appeared. She says she was not paid.

What happened next did not require a contract.

When you are cut off from organic distribution and a channel appears that amplifies your content to millions, you respond to that channel. Not because you are instructed to. Because the engagement tells you your content is resonating. This is how all content feedback loops work — you post, you measure the response, you adjust. If posts advancing one narrative consistently outperform posts advancing another, you produce more of the first kind. The feedback is the instruction.

The mechanism has a name in behavioral science: operant conditioning. Behavior is shaped by its consequences. When an organism receives a reward contingent on a specific action, it produces more of that action — not because it understands the reward structure, but because the reward signal is indistinguishable from genuine success.

Loomer’s engagement metrics, during her years of amplified distribution, could not be distinguished from authentic audience approval. Some of it was. Some was not. She had no method for telling the two apart. The synthetic component shaped her output without her knowledge, without her consent, without anyone writing a memo.

This is the architecture that does not need an operator.

Contrasting Two Models

Yesterday’s Decipon breakdown examined Tulsi Gabbard and the Science of Identity Foundation — an operation with a directive channel, explicit memos, verbatim execution. Someone in an identified position was receiving instructions and following them. The tell was the script: 24 of 32 television interviews, almost verbatim from memos.

The Loomer operation worked differently. There was no directive channel. No memos. No one at RT or in Russian state media infrastructure needed to tell her what to say. The amplification network selected for content that served their interests, rewarded it with distribution, and depressed content that did not. She produced more of the former.

The directed-proxy model requires an operator and a channel. It is detectable through the documentary evidence — find the memos, find the source of instruction, map the execution fidelity.

The amplification-reward model requires no operator and no channel. Its fingerprint is statistical: the correlation between content type and distribution support. But you only see that fingerprint from the outside, if you have access to sufficiently granular engagement data, and only after the fact.

Or — as in this case — when the target exits the loop and the amplification suddenly withdraws.

Why She Couldn’t See It

The most common response to cases like Loomer’s is: why didn’t she notice? The framing implies a failure of critical thinking — if she had been more skeptical, she would have caught it.

This misunderstands how feedback loops work.

Loomer was not being deceived about facts. She was not being shown false evidence about specific events and told to believe it. She was being rewarded, with engagement, for a category of output. The reward felt like authentic audience response because it was partially authentic. Real people also engaged with her content. The synthetic amplification was mixed in with the real. The ratio was not visible.

When all your distribution feedback points in the same direction, you update in that direction. This is not credulity — it is how anyone navigates content decisions. Every successful creator has optimized for engagement signals. If those signals are partly manufactured, the optimization runs on corrupted data. You do not notice the corruption because the signal looks like every other signal you have ever responded to.

Argument does not break this loop. Counter-evidence, fact-checking, debunking — none of it interrupts the feedback mechanism, because the mechanism does not operate through belief. You can believe Russia is propaganda while continuing to post content the amplification network rewards, because the reward arrives whether you endorse it consciously or not.

What broke Loomer’s loop was not argument. It was physical removal from the feedback environment. She went to Ukraine with an embedded journalist, saw the destruction firsthand, experienced a reality that the engagement metrics could not reframe. External direct experience, impossible to discount, broke the signal’s credibility.

That is a remarkably fragile defense mechanism to depend on at scale.

What the 28 Percent Reveals

The engagement drop has two possible explanations that observers have debated.

The first is organic: real followers who held pro-Russia sympathies, who had followed Loomer because of those sympathies, who departed when she reversed. Genuine audience churn following a political betrayal.

The second is inauthentic: Russian amplification infrastructure that had been artificially inflating her engagement, which withdrew when her content no longer aligned with operational interests.

Both could be true simultaneously. They are not mutually exclusive. A real audience with sympathetic views, amplified by an inauthentic infrastructure, will produce compound effects in both directions — natural and synthetic engagement on the way up, natural and synthetic disengagement on the way down.

But the suddenness is diagnostic. Authentic audiences drift. They grumble, some leave, others stay, the process unfolds over weeks. A 28 percent drop in five days, triggered by a single post, is not a gradual drift. It is the signature of coordinated, near-simultaneous withdrawal.

What the number measures is the delta between Loomer’s organic reach and her amplified reach. The gap — whatever fraction of 28 percent was inauthentic — is the operation’s footprint, made visible by its own withdrawal.

She exposed the operation by leaving it.

The Defense Implications

Media literacy frameworks typically target the belief layer: teach people to evaluate sources, identify propaganda, think critically before sharing. These interventions are designed to interrupt the transmission of false content.

They do not address feedback-loop capture, because that mechanism does not operate through belief.

An influencer being shaped by synthetic amplification is not primarily holding false beliefs — though she may come to hold some. She is operating in a corrupted incentive environment. Her output is shaped by rewards she cannot distinguish from authentic ones. Telling her to think more critically does not change the reward structure.

What would address this:

Transparency into the amplification layer. If platforms provided creators with data on the authenticity composition of their engagement — not perfect, but probabilistic — the feedback loop would be partially visible. Creators could see when their engagement on specific content types diverges from their baseline in ways that suggest external amplification.

Slowing the feedback cycle. The shorter the feedback loop, the more powerful the conditioning. Real-time engagement metrics train behavior faster than weekly aggregates. Platforms that prioritize immediate engagement signals are building more effective conditioning environments, including for adversarial amplification.

External anchoring. Loomer’s loop was broken by an experience that the feedback environment could not absorb. This suggests a category of intervention that has nothing to do with the platform: connections to reality outside the content ecosystem, in forms that engagement metrics cannot represent.

None of these are complete answers. The amplification-reward model is harder to defend against than the directed-proxy model, precisely because it requires no directive and leaves no memos. The operation’s only trace is the withdrawal — visible only in retrospect, only when the target exits.

What She Did

The admission Loomer made on June 17 is not trivially easy to make. She has spent years building an identity on a particular set of positions. Reversing in public, with millions of followers watching, is a reputational cost that most people in her position would avoid calculating.

She paid it. The engagement drop followed.

What she did not do — could not know to do — is diagnose the mechanism that had been shaping her output for years. She framed it as falling for propaganda: believing false things that she has now corrected. That framing is accurate as far as it goes.

The more complete account is that she was inside a reward structure she could not see, producing outputs the structure incentivized, with no available signal telling her anything was wrong. The operation did not need her to believe anything specific. It needed her to keep getting rewarded for content that served its interests. For years, that condition held.

The 28 percent is not primarily a story about Laura Loomer. It is a measurement of how much of any amplified influencer’s reach may be synthetic — invisible while it runs, visible only in the diagnostic moment when it stops.

This article is part of Decipon’s Manipulation Breakdowns series, examining specific influence operations through the Influence Tactics Protocol.

Sources: