Feeds Rank by Engagement and Hide Why, So Keep a Log of Design Choices
Ranking shapes what reaches debates about truth and law, so effects should be traceable to specific choices.

emerald.com · Algorithmic cognitive programmes: observing ambient architectures of attention with social systems theory | Kybernetes | Emerald PublishingIn an academic theory paper that presents no new data, Massimo Terenzi, who writes about platforms and public communication, argues that feeds rank content mainly by predicted engagement. Politics, science, business, and culture then rise or fall on clicks, likes, and watch time, not on their own standards for what matters.
To describe the stakes, he turns to what theorists call societal constitutionalism, the idea that powerful private systems can crowd out other areas of life and need internal limits. When he calls this a constitutional problem, he means one system's logic spreads into fields that should run on different rules, with no built-in way to check it.
That, he argues, is what engagement ranking does. A health post rises or falls not on accuracy or lawfulness, but on whether it draws a measurable reaction. The same test then shapes what people can find and debate across very different subjects.
Two forces drive this expansion into other fields, in his account. The first is economic, since platforms need sustained attention to earn revenue. The second is technical, since machine learning revises the ranking formulas through feedback, producing results that even designers may neither intend nor fully understand.
Terenzi does not call for publishing every line of code or second-guessing every post. He says that demand might be both impractical and unwise, and says fixing only bad outputs misses the cause. What he wants instead is what he calls constitutional observability, the ability to trace a visible effect back to a specific design choice.
In practice, that would mean a dated, auditable log of changes to ranking goals and weights, plus more varied ranking rules and an independent oversight body with access to operative parameters. That body, he argues, would not choose specific posts for specific users. It would sit inside the platform design and set structural limits.
Those limits could include minimum diversity in feeds and reserved space for material chosen for reasons other than engagement. Then, he argues, concerns from law, politics, and science could actually change the dials, rather than simply complain about the results.
Questions people ask
What does societal constitutionalism mean here?
It is the idea that powerful private systems, not only governments, can expand one system's logic into fields that run on different rules, threatening their integrity. The remedy, in this account, is self-limiting mechanisms within those systems that keep one logic from crowding out other ways of deciding what matters.
How is constitutional observability different from transparency?
It means outsiders can see what a ranking change did, link it to a particular design decision, and push back in a way that changes the design. Terenzi contrasts this with full disclosure of code, which he says might be both impractical and unwise.
Would this oversight body decide what posts people see?
No. Terenzi describes an independent oversight body with access to operative parameters that would adjust the structures of ranking. It would set program-level constraints, not determine which content is shown to which user.


