Intent Integrity

Last night, I did what millions of people do when something feels off:

I searched: “How do I sleep better?”

That sounds trivial, but it isn’t.

Sleep is not some niche wellness hobby. In the U.S., about one-third of adults are not getting enough sleep, and insufficient sleep is linked by public-health authorities to elevated risks including depression, obesity, heart disease, injury, and other serious conditions.^[1]

So when someone types a basic question like “How do I sleep better?”, they are not just browsing. They are often entering a health-information journey.

And that is where the problem starts.

Because the open web does not present all information under the same conditions.

Academic research is slower, method-driven, peer-reviewed, and subject to procedural scrutiny by other experts. That process is imperfect, sometimes biased, and far from foolproof. But it is still designed to evaluate methods, evidence, assumptions, and interpretation through independent review.^[2]

Open-web content often operates under a very different logic: speed, discoverability, distribution, conversion.

That difference is not theoretical. Research evaluating online information about insomnia found that the overall quality of websites was only moderate, and that all the content assessed exceeded recommended readability levels for the general public. The same paper argued that trustworthiness needed improvement through clearer authorship, update dates, conflicts-of-interest disclosures, and evidence sourcing.^[3]

Research on sleep apnea information online found another important pattern: commercial websites performed worse in providing information about diagnosis and management, while scientific and foundation sites performed better.^[4]

That does not mean every commercial page is wrong.

And it does not mean every academic paper is right.

It means a peer-reviewed study and a commercially adjacent sleep article are not just two different sources.

They are outputs of two different incentive systems. And yet, on the results page, they often arrive dressed as if they belong to the same category: “information.” They do not.

That gap matters because online health-information seeking is already mainstream. Pew has reported that 72% of adult internet users have searched online for health information, and a 2021 systematic review found that 83% of health information consumers searched using a general search engine.^[5]

So this is not a marginal UX complaint. It is a structural question about how modern knowledge is delivered.

And the structure we have today is shaped by incentives at enormous scale.

In Australia alone, internet advertising reached A$18.4 billion in 2025, with search advertising at A$8.0 billion, according to IAB Australia and PwC. That is not a side market sitting next to information. It is one of the economic engines underneath it.^[6]

This is why Herbert Simon’s old insight feels less like theory and more like infrastructure: in an information-rich world, what becomes scarce is attention. And once attention is scarce, systems organise themselves around capturing it.

That is also why something like Goodhart’s Law shows up here. Once “helpfulness,” relevance, or informational visibility becomes a target for monetisation, the signals that once pointed to quality can become distorted. The moment informational real estate becomes commercially valuable, it stops being a neutral environment.

So here is the thought experiment:

What if we separated truth-seeking from commerce at the interface level?

What if an informational mode had a hard rule: no brands, no products, no vendors, no commercial entities in generated answers — just concepts, mechanisms, evidence, competing explanations, levels of confidence.

Then, separately, a commercial mode: products, brands, comparisons, reviews, prices, shopping pathways.

And if a company wanted to appear inside the informational mode? Fine. But only as a clearly labelled ad.

No blending. No disguised neutrality. No soft-sell hidden inside explanation.

That would not eliminate bias. Framing bias, omission bias, and institutional bias would still exist. But it would do something important: it would stop pretending that explanation and persuasion are the same genre.

Right now, we collapse too many different intents into one interface.

Search research has long recognised distinctions between informational, navigational, and transactional queries. In industry practice, people often add a fourth bucket, commercial investigation, because many searches sit between learning and buying.^[7]

The problem is not that these intents exist.

The problem is that we serve them through a single blended surface and then act surprised when epistemic trust erodes.

Maybe the next generation of search should not just optimise for relevance.

Maybe it should optimise for something else entirely: intent integrity.

That is my term, not an established framework. But I think it names the design challenge well:

Did the system respect why the user asked the question?

If I ask “How do I sleep better?”, I may eventually want products. But not at the start. At the start, I want understanding.

And maybe the biggest design failure of the modern web is that we no longer force systems to distinguish between those two moments.

References

1. Centers for Disease Control and Prevention. (n.d.). Sleep. CDC. https://www.cdc.gov/cdi/indicator-definitions/sleep.html
2. Pulverer, B. (2021). Reproducibility: Replication failure is a feature, not a bug. Communications Earth & Environment, 2, 55. https://www.nature.com/articles/s43247-021-00100-2
3. Koffel, J. B., & Rohde, S. D. (2017). Quality evaluation of online health information on insomnia. Frontiers of Medicine, 11(3), 381–390. https://link.springer.com/article/10.1007/s11684-017-0524-9
4. Mold, F., & Forbes, A. (2018). Quality assessment of sleep apnoea patient information available on the Internet. npj Digital Medicine. https://pmc.ncbi.nlm.nih.gov/articles/PMC6175814/
5. Pew Research Center. (n.d.). Health information online. Pew Research Center. https://www.pewresearch.org/topic/internet-technology/healthcare-online/
6. IAB Australia & PwC. (2025). Internet advertising revenue report: December quarter CY25. IAB Australia. https://iabaustralia.com.au/resource/internet-advertising-revenue-report-dec-quarter-cy25/
7. Jansen, B. J., Booth, D. L., & Spink, A. (2008). Determining the informational, navigational, and transactional intent of Web queries. Information Processing & Management, 44(3), 1251–1266. https://www.sciencedirect.com/science/article/abs/pii/S030645730700163X