Most business frameworks arrive fully formed and fully confident. Four boxes, a catchy acronym, a promise. The Roth Complexity Lab does something unusual with its own framework: it lists the ways it might be wrong, then asks other people to check.

What is the Roth Complexity Lab?

The Roth Complexity Lab is a research lab and think tank in Budapest, founded by Miklós Róth. It describes itself as a public scientific institute advancing research into complex adaptive systems, AI, and a generative model of how new mathematics comes into being, all built on one lens: the S·I·C·T framework (Structure, Information, Cohesion, Transformation). It is remote-first and collaborates internationally. Its home is rothcomplexity.org.

The lab sits next to Róth's commercial work. CRS AI Marketing & SEO Agency and Roth AI Consulting supply the real-world problems. The lab is where the underlying ideas get formalized, tested and, when necessary, thrown out.

Why would an agency founder start a research lab?

Róth has been unusually candid about the motivation. In a LinkedIn post he wrote that he had at times rushed toward answers too quickly, mistaken complexity for depth and grown attached to ideas before testing them hard enough, and that some of the most valuable things he knows came from being wrong. He presents S·I·C·T in the same spirit: not as a finished theory, but as a proposal whose worth depends on whether it survives contact with reality. Where does it work? Where does it fail? What should change, and what should be abandoned?

That stance is rare in consulting, where frameworks are usually products. In the lab, the framework is a hypothesis.

How does the lab label its claims?

Every assertion the lab publishes carries an epistemic tag that says how well it is supported:

• Rigorous: proved mathematically or tested against data.
• Interpretive: a grounded reading of the evidence, but a reading all the same.
• Analogical: a useful parallel between fields, not a proof.
• Speculative: an open bet, explicitly marked as one.

The lab calls S·I·C·T pre-paradigmatic systems science: a rigorous proposal with a concrete path to stronger validation, not yet a settled theory. Readers can see at a glance which parts of an argument are load-bearing and which are exploratory. In an era flooded with confident-sounding text, that labeling discipline may be worth as much as any single result.

What does the lab actually research?

Its published work spans a deliberately wide range, because a framework that claims to describe systems in general has to be tested on very different systems. Topics on rothcomplexity.org include:

• A mathematical formalization of the S·I·C·T protocol, including the stability condition S + C ≥ I + T.
• Multi-agent cooperation, through the SICT Meta-Model, which bears directly on how teams of AI agents coordinate.
• Quantum versus classical physics, read through the S·I·C·T lens.
• Early-warning signals of fragility in financial systems, including preregistered diagnostic approaches.
• How new mathematics is born: the idea that when a field's information and pressure overwhelm its structure and cohesion, coherence returns at a higher level of abstraction.
• Cognitive operationalization of the framework for agentic AI systems.

Some of these sit far from marketing. That's intentional. If the same four coordinates say something useful about a central bank, a physics problem and a company rolling out AI, the framework has earned some credibility. If they don't, the lab wants to know.

How can a framework publish how it should die?

By stating in advance which observations would count against it. In its open research program, the lab invites outside researchers to run specific tests on data it cannot reach itself, to check whether S·I·C·T-derived indicators add predictive skill beyond standard early-warning indicators (with a null result explicitly counted as publishable), and to try to break its predictions using pre-registered measures.

It is the classic logic of falsification, applied to a working consultant's own toolkit. It also explains who the lab is looking for: systems scientists, AI governance specialists, organizational leaders, computational social scientists and investigative journalists working where institutional fragility meets fast technology.

Which of the lab's habits can a company borrow?

You don't need a research lab to use its discipline. Three habits translate directly into business, and they matter more as AI makes polished-looking analysis cheap.

Tag your claims. Before a board deck or strategy memo goes out, mark each key statement: measured, inferred, assumed or hoped. It takes ten minutes and changes the conversation, because everyone can see which numbers the plan actually rests on.

Write the failure conditions before the pilot. Decide in advance what result would make you stop an AI pilot, and write it down. Researchers call this pre-registration. In companies it prevents the familiar drift where every pilot is declared a success after the fact.

Invite the attack. Ask someone outside the project to argue that it won't work. A red team, a skeptical colleague, a customer. The goal isn't to kill ideas but to find their weak points while they are still cheap to fix.

What does this mean for Róth's clients?

More than it might seem. A consultant who labels his own claims is easier to believe when he says something is certain. When Róth tells a client through Roth AI Consulting that their AI rollout is outrunning their structure, the diagnosis comes from a framework that has been exposed to criticism rather than shielded from it. And when he isn't sure, the habit is to say so. The framework is explained here.