It is not difficult today to find an impressive claim. A city announces that it has achieved exceptionally high recycling rates. A company declares that it has become carbon neutral. An organisation presents a new circular economy model. A public authority announces that it has digitised its processes. An artificial intelligence company claims that its latest system is safer, more accurate and more reliable than anything that came before it.
These stories are not inherently problematic. On the contrary, society needs examples of progress. It needs to know that difficult problems can be addressed, that organisations can genuinely improve the way they operate, and that certain ideas can move from theory into practice. Without positive examples, public debate can easily become trapped in the identification of problems and in the belief that nothing can change.
Some time ago, I read another such story. It concerned an initiative in waste management and the circular economy. The results presented were impressive: high material recovery rates, significant diversion of waste from landfill, and the image of a model that could potentially serve as an example for other regions.
I had no reason to assume that what was being presented was inaccurate. Initiatives of this kind may involve real work, persistence, changes in everyday behaviour, investment in infrastructure and cooperation between people trying to achieve something meaningful. Every genuine improvement in waste management deserves recognition, especially when landfill is still treated in many cases as the easiest final destination rather than the last resort of a mature system.
Yet as I continued reading, I realised that the central question was not whether I believed the numbers. I was not trying to decide whether the story was true or exaggerated. Something else was occupying my mind: how could someone who had not participated in the initiative independently confirm that its claims were correct?
The difference between these two questions is fundamental. The first forces us to choose between trust and distrust. The second moves the discussion away from individuals, intentions and communication, and towards methodology, data and the design of the system itself.
In waste management, this becomes easy to understand. A high recycling rate may sound perfectly clear without actually being so. What exactly has been measured? The total amount of waste generated, the quantities collected separately, the material entering sorting facilities, or the material ultimately recovered and returned to the economy? Which waste streams are included, and which are excluded? How are processing residues, contamination and materials that were collected separately but could not ultimately be recovered treated?
The same ambiguity can exist in landfill diversion. Waste that has not been sent directly to landfill has not necessarily been recycled. It may have been temporarily stored, sent for another form of treatment, exported or eventually landfilled through a different route. Without clear traceability from the point of generation to the final destination, an impressive percentage may be numerically correct while still presenting an incomplete picture.
Even the term “Zero Waste” does not always carry the same meaning. For some, it describes a long-term direction built around prevention, reuse, recycling and the drastic reduction of landfill. For others, it refers only to a specific waste stream or to a limited period of time. In areas with major seasonal population changes, the timing of the measurement may also materially affect the result. An annual figure, an average or a percentage calculated over selected months does not necessarily describe the same reality.
These observations do not diminish the value of success. They protect it. The more important a claim is, the more valuable it becomes when it is supported by clear definitions, a consistent methodology, accessible evidence and a path that allows an independent observer to understand how the final conclusion was reached.
Without these elements, two organisations may announce exactly the same percentage while describing two entirely different situations. Neither needs to be lying. They may simply be using different definitions, different system boundaries and different calculation methods. Numbers do not eliminate ambiguity by themselves. Sometimes they conceal it more effectively.
This leads to a much broader issue. Almost every important decision in modern society rests on claims. A business claims that it has reduced its environmental footprint. An organisation announces that it has achieved its ESG targets. A government presents a public policy as successful. A hospital publishes improved quality indicators. A university presents the findings of a study. A technology company claims that its product is safe, effective or unbiased.
In all these cases, the discussion usually returns to the same question: whom do we trust?
Do we trust the organisation that published the data? The expert who interpreted it? The institution that approved it? The media outlet that reproduced it? The company that funded the study? The auditor who signed the report?
Institutions, experts and audit processes remain essential. No complex society can function without some degree of trust in people and organisations that possess knowledge, authority and responsibility. The problem begins when the credibility of a claim depends entirely on the status of the person or institution making it.
For much of history, trust was based on authority. We believed something because it came from a recognised expert, an institution, a government, a university or a major media organisation. That model has not disappeared, nor can it disappear entirely. Yet it is becoming increasingly inadequate in a world where claims multiply faster than our ability to examine them.
The answer to this crisis cannot be universal distrust. A society that questions everything without criteria does not become more rational. It becomes more vulnerable to misinformation, conspiracy theories and any narrative that confirms existing beliefs. Distrust is not, by itself, a method of verification, just as trust is not proof.
What is needed, therefore, is a different transition: from trust based primarily on authority to trust grounded as far as possible in evidence that can be examined. Not because people or institutions are necessarily unreliable, but because even the most honest organisations can make mistakes, use incomplete data, select inappropriate indicators or interpret the same evidence in different ways.
This is where the distinction between transparency and verifiability becomes important. The two are often used as though they were synonymous, but they are not.
An organisation may publish a report hundreds of pages long, including dozens of tables, charts and detailed descriptions of its methodology. It may be considered fully transparent because it has released a large volume of information. Yet if the reader cannot connect the final conclusions to the underlying data, reproduce the calculations or understand which assumptions affected the outcome, transparency alone does not create trust.
Transparency concerns what an organisation chooses to reveal. Verifiability concerns whether an independent third party can follow the same path and reach the same, or a comparable, conclusion.
This distinction becomes even more important in the age of artificial intelligence. Until recently, producing a well-designed report, a persuasive text, a professional presentation or sophisticated visual material required time, skill and financial resources. The quality of presentation was never proof of truth, but it often functioned as an indirect signal of seriousness and investment.
That signal is rapidly losing value. Artificial intelligence is dramatically reducing the cost of producing persuasive content. A polished article, an impressive chart, a comprehensive report or a professional video can now be created in very little time. This development does not merely make falsehood easier. It makes truth, exaggeration, error and half-truth equally easy to package convincingly. Most importantly, it makes their outward appearance increasingly difficult to distinguish.
When the cost of producing a persuasive claim approaches zero, persuasiveness itself ceases to be a meaningful indicator of credibility. The central challenge will no longer be who can create the best narrative, but who can substantiate what they claim.
This also changes the way systems need to be designed.
After years of working with information systems architecture, I have come to believe that it is not enough for a system to execute a process correctly. It must also preserve the history of that process. It should be able to show which data entered the system, who created or modified it, which transformations were applied, which rules were triggered and how the final result was produced.
If this capability is not designed from the beginning, it is extremely difficult to add reliably afterwards. A report created at the end cannot recover data that was never recorded. A dashboard cannot reconstruct a chain of traceability that never existed. An external audit cannot confirm with certainty events for which the system retained insufficient evidence.
This applies equally to a waste management system, a digital public service, a sustainability report or an artificial intelligence model. Proof should not be treated as an additional reporting requirement that appears after the operation has been completed. It should be a natural product of the operation itself.
A mature waste management system should not merely announce how many tonnes it recycled. It should be able to connect incoming quantities to their source, the stages of collection and treatment, processing residues, deliveries to final recipients and the methods used to calculate performance indicators. The final result should emerge from the functioning of the system rather than from a separate communications exercise at the end of the year.
A digital public service should not merely claim that it has accelerated a process. It should be able to demonstrate, through comparable evidence, how long the process took before, how long it takes now, which steps were removed, how many cases were completed and where delays still remain. A company reporting on ESG targets should not be limited to producing an attractive report; it should maintain consistent data provenance, clear definitions and the ability to reproduce its core calculations.
Likewise, an artificial intelligence system cannot be considered reliable solely because it performs well in a set of tests selected by its creator. It requires clear documentation of its data, limitations, evaluation methods and the conditions under which its performance declines. The more a system influences human decisions, the more important it becomes to examine not only the result, but also the process that produced it.
This principle could be described as “Verifiability by Design”: the capacity for verification should not be added after a system has been built, but embedded in its architecture from the beginning.
This is not merely a technical principle. It is a different understanding of trust. In such a system, credibility does not depend exclusively on the good intentions of the operator, the reputation of the organisation or the quality of the presentation. It is supported by the structure itself: by data provenance, traceability, controls, documentation and the possibility of independent examination.
This does not mean that every piece of information must be made public without limits. There are legitimate requirements relating to personal data, commercial confidentiality, security and sensitive processes. Verifiability is not the same as the unrestricted disclosure of every detail. It means that sufficient mechanisms exist for appropriate and independent parties to examine a claim according to clear rules, without relying solely on the narrative of the organisation being assessed.
Nor does it mean that data are neutral or infallible. Data can be incomplete, inaccurate, biased or irrelevant to the question being asked. That is why verification cannot be reduced to the existence of numbers. It also includes definitions, methodology, provenance, quality, limitations and interpretation.
The goal is not to replace blind trust with an equally blind faith in data. It is to build systems in which claims can be examined meaningfully.
I believe this need will become one of the central challenges of the next decade. As the volume of generated content increases, as systems become more complex and as more decisions are based on automated processes, it will become increasingly difficult to assess credibility through appearance, status or the persuasiveness of a narrative.
So I will make a prediction, even at the risk of being proven wrong: verifiability will become one of the most important concepts of the next decade.
We will encounter it in waste management and environmental targets, in sustainability reporting and financial disclosure, in public administration, science, journalism and, above all, artificial intelligence. Not as yet another compliance requirement, but as a fundamental characteristic of systems that expect to be trusted.
In the previous century, much of our effort was directed towards building systems that could operate quickly, efficiently and at scale. These qualities will remain important. But they will no longer be enough.
The systems of the future will need to prove that they operated correctly.
Perhaps this will be the most important shift in the way we understand trust: moving away from the idea that trust is produced through communication, reputation or authority, and beginning to see it as a property of design.
Because an important claim does not gain value merely by being true. It gains real power when it can be proven.
