Daniel sent us this one — he's been thinking about electoral interference, but not the usual playbook. The standard story is that a foreign power picks a candidate they like and tries to get them elected. Russia, twenty sixteen, Trump. But Daniel points out that for a regime like the IRGC in Iran, Israel isn't a place where you have a preferred side. It's a mortal enemy, full stop. So the objective shifts. Instead of backing one faction, you amplify all the fringe voices — far left, far right — and wait for the center to tear itself apart. He calls it a kindling effect. And he wants us to walk through how this has actually been used, including the funding of protests and the amplification of voices at the margins.
This is exactly the conceptual shift that most coverage misses. Everyone's still looking for the smoking gun that proves Iran tried to elect candidate X or defeat candidate Y. But that's not the game. The game is to make the entire system ungovernable. I've been digging into this and the operational logic is genuinely chilling once you see it laid out.
It's not picking a winner. It's burning the house down and hoping everyone inside takes each other out before the fire trucks arrive.
That's the image, yeah. And the IRGC has been remarkably explicit about this internally. There were leaked documents in twenty twenty-three — internal strategic assessments — that prioritized what they called "internal erosion" over direct military confrontation. The logic is straightforward: why risk a shooting war with a superior military when you can get Israelis to tear each other apart?
Which is also cheaper and much harder to trace back to Tehran.
And that's the core paradox Daniel's getting at. How do you prove a negative? If Iran is amplifying existing voices rather than creating new ones, the people being amplified are real Israelis with real grievances. They're not Iranian agents. They may have no idea they're being boosted. So when someone says "this protest is being amplified by Iranian bots," the immediate response is "so you're saying these legitimate concerns are fake?" And that's precisely the trap.
The deniability is baked into the tactic itself. You're not inventing the anger, you're just making sure everyone hears it at maximum volume.
Let me lay out what I think of as the three mechanisms. The first and most visible is social media amplification. Iranian state-linked accounts — and we're talking about networks tied to the Islamic Republic of Iran Broadcasting, or IRGC-affiliated hacktivist groups — they don't post original content. They identify fringe Israeli voices on both extremes and pour gasoline on them.
They're not writing the tweets, they're just making sure the most inflammatory ones go viral.
And the sophistication is in the symmetry. During the twenty twenty-three judicial overhaul protests, researchers found coordinated amplification of both the most extreme anti-government protestors and the most extreme pro-government counter-protestors simultaneously. Same networks, same infrastructure, pushing content that made both sides look more unhinged and more threatening to the other side.
Which makes each camp more convinced the other is about to destroy the country.
And that's the kindling Daniel's talking about. You don't light the fire. You find the dry tinder — the existing grievances, the real anger — and you fan it from every direction at once.
That's mechanism one. What's the second?
And this is where it gets sophisticated. In twenty twenty-four, Shin Bet released a report that I found remarkable — they'd uncovered evidence that Iranian intelligence had attempted to fund both sides of the Israeli protest movement through separate, unaware intermediaries. Different cutouts, different front organizations, none of whom knew the other existed or that their money traced back to the same source.
So the same Iranian operation was funding the people trying to defend the judicial overhaul and the people trying to stop it?
That's exactly what Shin Bet found. And think about the operational elegance of that. If you fund one side and it's discovered, you've embarrassed your preferred faction. If you fund both sides and it's discovered, you've tainted the entire protest movement. Every genuine activist now has to spend energy proving they're not on Tehran's payroll.
That's the both-sides trap you mentioned. The discovery of the operation becomes itself an operational success.
Exposure doesn't defeat the tactic, it amplifies it. Because now the conversation shifts from "what are we protesting about?" to "who's a foreign agent?" And that conversation is itself polarizing and paralyzing.
The worst case scenario for Iran isn't that they get caught. It's that nobody notices. And the second worst case is that everyone notices and starts accusing each other.
That's the strategic elegance of it. You win either way. Either the amplification works undetected and the center erodes, or it's detected and the ensuing witch hunt erodes the center anyway.
Alright, mechanism three. You mentioned deepfakes.
Yeah, and this one's particularly nasty because it exploits the speed of information versus the speed of verification. In twenty twenty-five, there was an incident where a fake IDF spokesperson appeared in a video announcing a blanket draft exemption for the ultra-Orthodox community. The video was a deepfake. It wasn't particularly good by technical standards — if you paused and looked closely, you could spot the artifacts. But it didn't need to be good. It needed to circulate for forty-five minutes before it was debunked.
In those forty-five minutes, you've got secular Israelis furious about draft dodging and ultra-Orthodox Israelis furious when they find out it was fake and assume it was a secular provocation.
Both sides inflamed simultaneously, from a single piece of content that took maybe two hundred dollars worth of compute to generate. And here's the thing — even after it's debunked, the emotional residue remains. People remember the anger more than they remember the correction. The deepfake gets retracted, but the fight it started keeps going.
Because the fight was real, even if the spark was fake.
That's the whole game. Iran doesn't need to convince anyone of anything. It just needs to make sure Israelis are too busy fighting each other to notice or respond to external threats.
Let's put some operational names on this. You mentioned Operation Coral Reef.
This was a twenty twenty-five Mossad disruption of an IRGC front network — more than forty fake news sites, all publishing content in Hebrew, all designed to look like legitimate Israeli news outlets. And here's what was clever about it: the sites didn't all push the same line. Some attacked Netanyahu from the right, some attacked him from the left, some attacked the opposition from both directions. The network wasn't trying to build a coherent narrative. It was trying to make sure that whatever your existing grievance was, you'd find content that validated and escalated it.
It's a custom radicalization pipeline. Whatever direction you're already leaning, there's a site that looks credible pushing you further in that direction.
And because the sites looked like real news — they had similar layouts, similar branding, they'd sometimes even scrape and republish legitimate articles alongside the fabricated ones — most readers would never think to check who was actually behind them.
This is where I think the kindling metaphor Daniel used is actually more precise than he might have intended. A fire needs three things: fuel, oxygen, and an ignition source. Iran is providing the oxygen — the amplification — and sometimes the ignition — the deepfakes, the provocations. But the fuel is already there. The grievances, the historical tensions, the genuine political disagreements. Those are real.
That's what makes this so hard to talk about without sounding like you're dismissing legitimate dissent. If I say "Iran is amplifying far-left anti-Zionist voices," someone will hear "all anti-Zionist voices are Iranian assets." That's not true, and it's not what anyone is claiming. But the tactic is specifically designed to make those two statements blur together in the public mind.
It's a contamination strategy. You don't need to defeat your enemy's ideas if you can make them all smell like foreign manipulation.
Let me give you a concrete case that illustrates this perfectly. The twenty twenty-four Flag March in Jerusalem. For listeners who don't follow the details, this is an annual march through the Old City that's already a flashpoint. In twenty twenty-four, researchers tracking Iranian-linked amplification networks found something striking. The same cluster of accounts was simultaneously amplifying the most extreme voices calling for settler violence during the march and the most extreme voices calling for violent resistance to the march.
Both sides of the confrontation were being fed the same escalation.
And the accounts didn't coordinate with each other in any visible way. They weren't replying to each other or retweeting each other. They were operating in parallel, each targeting their own audience with content designed to make the other side seem maximally threatening.
Which makes the confrontation on the ground more intense, which generates more dramatic footage, which gets amplified again. It's a feedback loop.
Here's the knock-on effect that I think is underappreciated. After a few cycles of this, genuine activists start to wonder if the people standing next to them at a protest are real or if the whole thing is being manipulated. That doubt is corrosive. It undermines the basic trust that civil society requires to function.
You start seeing Iranian bots behind every opinion you don't like.
Which is itself a form of polarization. And Iran doesn't have to lift a finger to make that happen. Once the idea is planted that foreign manipulation is everywhere, people do the work themselves, dismissing anyone they disagree with as a bot or a useful idiot.
The paranoia becomes a weapon even after the operation is exposed.
That's the third knock-on effect I want to highlight. Intelligence community paralysis. When Shin Bet or Mossad looks at a surge of extremist activity, they now have to ask: is this genuine domestic radicalization, or is this being amplified from outside? And the answer is usually both, which makes the operational response incredibly complicated. If you crack down hard and it turns out you're mostly seeing amplified fringe voices, you've alienated citizens who feel persecuted for their opinions. If you go soft and it turns out there's a real domestic terror cell in the mix, you've failed to prevent an attack.
They're damned if they overreact and damned if they underreact.
Iran knows this. The whole point is to force exactly that dilemma. Every shekel and hour Israeli intelligence spends trying to untangle domestic extremism from foreign amplification is a shekel and hour not spent on countering Iran's nuclear program or its conventional military buildup.
That's the resource diversion angle. Even if the kindling never actually starts a fire, just forcing your enemy to spend resources watching for sparks is a win.
The cost asymmetry is brutal. Running a network of forty fake news sites and a few hundred amplification bots costs what? A few million dollars a year, maybe? The counterintelligence resources required to detect, attribute, and disrupt that network cost orders of magnitude more.
Let's pull back a bit. Daniel's prompt is really about whether this is a distinct playbook, and I think the answer is clearly yes. How does this compare to the Russian model everyone's familiar with?
The Russian model, especially what we saw in twenty sixteen, is what researchers sometimes call a "firehose of falsehood." You flood every channel with pro-Kremlin content, you create multiple conflicting narratives, you make it impossible to know what's true. But the content is all oriented in one direction — toward Russian interests. Iran's approach is different. It's more surgical. You don't create new voices, you amplify existing local ones. You don't push a single narrative, you push every narrative that's destructive to the target society.
Russia wants you confused and supporting Russia. Iran wants you angry and fighting each other.
That's the distinction. And it reflects different strategic positions. Russia sees itself as a great power competing for influence. It wants allies, or at least compliant neutrals. The IRGC sees Israel as an entity to be eliminated. There's no scenario where Israel becomes a friendly state, so there's no point in trying to influence its elections toward a preferred outcome. The only rational objective is to weaken it.
Which makes the kindling approach not just one option among many, but the logically necessary strategy given the premises.
And this isn't unique to Israel. We've seen the same playbook in Iraq, where Iranian-linked networks have amplified both Shia militias and Sunni grievances simultaneously. In Lebanon, where they've boosted both Hezbollah's hardest-line voices and their most extreme critics. The goal in each case isn't to win, it's to make sure nobody else can govern.
There was also the US campus protest dimension you mentioned. Twenty twenty-four, twenty twenty-five.
Yeah, this was fascinating and undercovered. During the peak of the Gaza campus protests, researchers found Iranian-linked accounts amplifying both the most extreme pro-Palestinian voices — including calls that went well beyond what most protesters were demanding — and simultaneously amplifying the most extreme pro-Israel voices, including content that framed all protestors as terrorist sympathizers.
Again, the goal isn't to help one side win. It's to make the conflict as bitter and intractable as possible.
To make sure that Americans are fighting each other about Israel rather than having any kind of productive debate. Every hour the US political system spends in polarized fury about campus protests is an hour it's not spending on, say, scrutinizing Iran's nuclear progress.
The Overton window shift is real too. You mentioned this in the context of Israeli politics specifically.
Yeah, this is one of the most measurable effects. By twenty twenty-six, positions that were considered fringe even five years ago have become mainstream talking points in Israeli political discourse. On the left, open calls for a binational state — not two states, one state — which was basically unsayable in mainstream Israeli politics a decade ago. On the right, the language of "transfer" has crept from the absolute margins into Knesset speeches.
The amplification mechanism accelerates this because politicians see these ideas getting traction online and assume there's a constituency for them.
The amplification creates an illusion of grassroots support that then becomes real as politicians chase what they think voters want. It's a self-fulfilling prophecy, and Iran just has to keep the amplifiers running.
What do we do about this? Daniel's prompt is really an invitation to understand the tactic, but I think listeners are going to want something actionable.
I've been thinking about this in terms of what I'd call "kindling literacy." The first step is just understanding that the tactic exists and recognizing its signature. When you see a story that seems perfectly designed to inflame both sides of a conflict simultaneously — when you can feel in your gut that this is going to make everyone furious at everyone else — that's a signal to pause and check the amplification chain.
What does someone actually do?
There are tools. Bot Sentinel, Hoaxy — these let you check whether the same accounts are boosting content from opposing viewpoints. If you see an account that's retweeting both the most extreme settler voices and the most extreme anti-Zionist voices, that's not a confused person. That's an amplifier.
For journalists covering protest movements or political polarization, the question should always be: who benefits from this specific conflict? If the answer seems to be "both sides benefit from escalation," look for a third party.
That's the analytical lens. And for tech platforms — and I know this is a bigger ask — the detection algorithms need to evolve. Right now, most coordinated inauthentic behavior detection looks for accounts that are all pushing the same message. That catches the Russian model. It misses the Iranian model, where the coordination is in the symmetry. Flagging coordinated amplification of both sides of a conflict should be a standard red flag.
Which is harder to detect because it looks like organic disagreement.
It looks exactly like organic disagreement. That's the point. You need to look at the infrastructure, not the content. Are the accounts amplifying the far left and the accounts amplifying the far right sharing the same creation patterns, the same timing, the same network characteristics? That's the tell.
As AI-generated content gets cheaper — and we're already seeing this with the deepfakes — the cost of running one of these operations drops toward zero. What does that mean for the future?
I think we're going to see this become the default influence tactic of the late twenty twenties. Not just for Iran, but for any actor whose goal is destabilization rather than persuasion. And the next frontier is what I'd call real-time kindling. Imagine generative AI that monitors a live political crisis and instantly produces content designed to inflame every faction simultaneously. It posts the deepfake, it writes the inflammatory tweet, it generates the fake news article — all within seconds of a real event, before anyone has time to verify anything.
You don't even need to plan the operation in advance. You just point the system at a society and wait for something controversial to happen, which in the current information environment is roughly every fifteen minutes.
Then the system amplifies every fringe reaction to that event, making sure that whatever the real story is gets buried under an avalanche of manufactured outrage from all directions.
That's the thing that keeps me up. Not that this tactic exists, but that it's getting cheaper and faster and harder to detect, and our defenses are still built for the last war.
The one thing I'll say that's slightly hopeful — and I want to be careful not to overstate this — is that awareness itself is a defense. The kindling tactic works best when nobody knows it's happening. Once people understand the playbook, they start asking the right questions. They get suspicious of content that seems too perfectly designed to enrage everyone. They check sources. They look for amplification patterns.
It's not a cure, but it's a vaccine. Exposure to the tactic in a controlled way builds some immunity.
That's the hope. And that's really what Daniel's prompt is doing — it's not just describing a problem, it's building the conceptual framework people need to recognize the tactic when they encounter it in the wild.
Let's land this. The kindling effect is a distinct influence playbook. It doesn't try to pick winners. It amplifies every fringe voice until the center can't hold. It's deniable by design, it's cheap to run and expensive to counter, and it's becoming the default strategy for any actor whose goal is destabilization rather than persuasion. And the defense starts with knowing it exists.
Then asking the right question every time you feel yourself getting outraged by something online: who wants me to feel this way, and why?
One last thought before we wrap. The open question I keep coming back to is this: how do democracies defend against a strategy that weaponizes their own freedoms? The whole point of an open society is that people get to have extreme opinions and say them loudly. If we start treating every fringe voice as a foreign operation, we've lost what we're trying to protect. But if we do nothing, we're letting adversaries exploit that openness to tear us apart.
That's the tension. And I don't think there's a clean resolution. The best we can do is be aware of the dynamic, teach people to recognize it, and build detection systems that look for the infrastructure of amplification rather than policing the content of speech.
Which is messy and imperfect and probably the best we're going to get.
Welcome to defending democracy in the twenty-first century.
Now: Hilbert's daily fun fact.
Hilbert: During the Cold War, a linguist studying click consonants in Khoisan languages discovered that a field manual for tracking operations in the Simpson Desert contained a transcription error so persistent that it accidentally encoded a complete set of!Xóõ tone patterns into what was supposed to be a list of water sources.
That's either the most impressive clerical error in history or someone at the CIA was very, very bored.
I have so many questions and I know I'll never get answers to any of them.
This has been My Weird Prompts. Thanks to our producer Hilbert Flumingtop. If you found this useful, share it with someone who needs to understand how influence operations are evolving. We'll be back with a new episode soon.
