Why some ideas go viral in Brussels, but most don’t

Sometimes, I try to write a paper/strategy without mentioning “Competitiveness”, only to fail miserably. It could be because I lack willpower. Or because I am intellectually bankrupt. Or because of a concept in physics known as Self-Reinforcing Cascades (SRC).

SRC explains how “Competitiveness” became an unavoidable phenomenon overnight, but “Strategic Autonomy” didn’t. It explains why some ideas go viral, but most don’t.

A recent paper in the Physical Review Letters has modelled how things like ideas spread in complex environments. To oversimplify, it finds that when ideas don’t differ strongly (like in our homogenous Bubble), what decides how well they propagate is not how complete or compelling they are, but how often they get repeated in their early days.

This tracks with my experience in Brussels (which behaves like any other complex system). Below is my intuitive application of the model to the rise of “Competitiveness” and the petering out of “Strategic Autonomy”.

What the paper says

The article’s premise is that when the difference between two choices (in this case policy ideas) is too negligible to determine outcomes, ideas spread not because they are superior, but because they get repeated early on.

Once an idea is repeated enough times, it crosses a threshold, after which it becomes self-perpetuating. The probability that someone repeats it increases simply because others have done so before. The system then locks the idea in, even if the idea remains vague or incoherent.(1)

That’s fairly shocking: that repetition matters more than meaning, analogous to “clicks matter more than content”. But, hopefully, I’ll be able to conclude that we shouldn’t regress to slop-posting, and there’s still space for ideas in agenda-setting.

Step 1: Seeding

The starting point for the model is similar to received Brussels wisdom. Ideas are sensitive to initial conditions. This is campaigns tend to start with an early advantage if they are thrown into the mix by the right actor, in the right way, at the right moment.

“Competitiveness” was teed-up and inserted directly into the debate by Mario Draghi – one of the EU’s most respected figures – at a moment when the bloc was on the ropes and knew it.

“Strategic Autonomy” didn’t have that singular founding moment. It emerged erratically, initially in French strategic discourse. No one “owns” it across the Bubble like Draghi does “Competitiveness”. The closest it came to having a defining moment was a speech at the Sorbonne by Macron in 2019, which the Bubble did not follow as closely as it did Draghi.

Step 2: Drifting

The model argues that after “seeding’”, ideas sort of “drift” in a semi-random manner.(2) As they drift, the probability that they go viral or die out (irrespective of their strength) fluctuates. In this case, the “drift” would mean testing and prodding a concept in speeches, op-eds etc. before it dies or formally becomes a “thing”.

“Competitiveness” had favourable drift conditions. It raised few antibodies. (It could not be accused of being too French, too protectionist, too dirigiste, too federalist, too market-fundamentalist.) It was inoffensive and means nothing and everything to everyone: an empty signifier, like ’08 Obama’s “Hope”.(3)

“Strategic Autonomy” came out of the blocks on the defensive. It started out tilted by the French toward sovereignty and faced multiple critics on multiple fronts.(4) This resistance hampered its “drift”.

Step 3: Reinforcing

The crux of the paper’s argument (and where things start to diverge from the Brussels norm) is the threshold. Once an idea has been repeated often enough (and what constitutes “enough” is surprisingly small), the probability of further repetition increases sharply. The more it’s discussed, the more it’s taken up. The more it’s taken up, the more it’s discussed. The process becomes self-accelerating. All this means ideas can go viral even in imperfect conditions, so long as they reach the tipping point faster than competing ideas.(5).

“Competitiveness” crossed this threshold almost immediately because the system was primed in steps 1 and 2. Then Draghi repeated it. Politicians repeated Draghi. Civil servants repeated their leaders. Briefings repeated civil servants. Everyone repeated the briefings. It became sticky and – eventually – self-reinforcing.

“Strategic Autonomy” took years to reach reinforcement because it kept changing shape. Still today, my intuition is no one talks about “Strategic Autonomy” (or “Open Strategic Autonomy”) anymore, but everyone talks about Europe’s economic / technological / supply chain / research / internal / economic “security”. And “security” already seems to be morphing into “resilience”. Each redefinition effectively resets the accumulated reinforcement, slows momentum, and delays crossing the threshold.

Step 4: Diffusing

With reinforcement, the model predicts a shift from randomness to diffusion through clusters. Ideas spread through repeated interactions of tightly knit groups before jumping between clusters like fires merging. Brussels is highly conducive to this. It is a system of overlapping, semi-isolated, but still interacting, clusters: DGs, cabinets, Parliament political groups, think-tank networks, expert committees etc.

“Competitiveness” crossed communities so rapidly because it was neutral. Anyone could adopt it without it threatening their identity. Everyone from the fossil fuel industry to environmental NGOs has tried to fit themselves into the “Competitiveness” box.

By contrast, “Strategic Autonomy” fractured along cluster lines. Rather than trying to fit themselves into the “Strategic Autonomy” trend, different communities tried to fit Strategic Autonomy into their desired end state. For example, security and trade policymakers used it differently. So, while the idea did cascade within different policy networks, it never scaled across them like “Competitiveness” did. On the contrary, these different meanings clashed, which slowed the spread.

Step 5: Saturating

Next, the idea becomes the baseline. It is no longer just an idea but the conceptual lens we all see everything through. Like in my opening example, people reach for it in papers/speeches/strategies because it is a path of least resistance. It’s easy to use without having to justify it.

“Competitiveness” became the baseline remarkably quickly. It is the default frame for talking about any measure intended to strengthen Europe’s long-term performance at a time when that is all anyone is talking about. Its conceptual vagueness remains its strategic advantage: it can mean productivity, innovation, geopolitics, industrial scale-up, regulatory reform, sectoral wish lists, anything.

“Strategic Autonomy” never became a default because of ideological scepticism and conflicting meanings. It has always been possible to talk about economic or foreign policy without mentioning it, while today one simply can’t talk about innovation or productivity without mentioning “Competitiveness”.

Step 6: Lock-in

The final stage in the model is lock-in. At a certain point, abandoning the idea is harder than continuing to use it. Even if the idea is vague or flawed, the system prefers continuity to the cost, coordination, and explanation that comes with switching to something else.

Today, “Competitiveness” is locked in. It would be a misstep for a policymaker to omit, contradict or ignore it. I still don’t quite know what “Competitiveness” means, but that’s not important. The point is that it is now a Bubble fixture. And, combined with the EU’s taste for path dependency (another hallmark of complex systems), it seems unlikely to be going anywhere.

“Strategic Autonomy” never reached lock-in. When Trump was elected in November 2024, I wrongly argued voting patterns in the European Parliament on strategic autonomy suggest MEPs still didn’t know what it was, that they were still making their minds up on whether they supported it, and that US elections would focus its definition and force an EP position on it. 12 months later, it semi-exists, but is now being tweaked and rebranded (as “Economic Security”, “Supply-chain Resilience” etc.). The fact it can be replaced shows it was never locked in.

Conclusion

At first glance, the conclusions might seem alarming: if the model is taken literally, ideas don’t win because they are better, but because they are repeated. It’s not about winning arguments, it’s about winning the space then filling it with your substance.

For all the talk in Brussels of “evidence-based policy”, if the model holds, then it really is Gramsci’s world and we’re just living in it. It would be a metapolitical victory: establishing a conceptual territory first, with the policy flowing downstream from that.

If we start seeing agenda-setting this way, Brussels just becomes a deterministic battlefield where victory goes to the ideas with the lowest conceptual friction and the highest reinforcement momentum. In this arena, the trick is not problem-solving, but cascade engineering. Then it is only a small step to rage-baiting and the kind of Ghiblified slop in the image accompanying this blog.

But the EU’s “Complexity” and unpredictability are also its strengths here. Complex systems are exposed to randomness, exogenous shocks, and dynamics. Unlike the model, actors and networks don’t function in idealised states. In the real world, ideas can be framed, but can also be re-framed and counter-framed. Networks can act randomly, but can also be coordinated. Probability is not fixed, but can be reinforced organically or through engineering.

All this creates edges in EU policymaking’s structured randomness, that you can exploit if you have the political agency to do it.

***

(1)     The model still assumes people react to small preference differences, but once an idea crosses the threshold, sheer repetition matters more than those preferences.

(2)     In the model the drift isn’t totally random (people still lean toward whatever seems slightly more useful) it’s just not strong enough to steer the outcome.

(3)     Perhaps some saw it as too neoliberal, but the point stands – if you go to a panel on competitiveness with 5 speakers, you’ll end up with six favourable definitions of competitiveness, which in this case, was its strength.

(4)     Germany feared it implied decoupling. Nordics feared protectionism. It was too economic for the security policy bubble and to geopolitical for economics bubble.

(5)     Positive feedback loops and the possibility of “fat tails”, further strengthen the case that Brussels is a complex system first and a political system second.

- Brendan Moran is Director at DELTA-V Public Affairs