Thinking outside the algorithm

On creating a better climate of ideas

Thinking back to when the internet was new is to remember a time when the world was promised an era of radical openness. The World Wide Web, we were told, would connect us and, in doing so, help us to better think, understand, and imagine. In some ways, the internet has delivered on this promise. It has linked anyone with web access to a flow of information that—in its speed, dynamism, and sheer complexity—resembles thought itself, a kind of global hive mind augmenting our ability to engage with everything, everywhere, all at once.

This Substack is, perhaps, a case in point, a small part of the great river of words and data that is the internet in 2025. The conversations, ideas, and connections that are the heart of The Healthiest Goldfish are enabled and deepened by the online world we have created, a world that has indeed opened new opportunities for how we communicate.

But as digital media has opened our world it has also narrowed it. What we read and see online, the information we consume, now comes to us by way of finely tuned systems that shape our online exposures. We are all, at core, consuming what comes to us through algorithms. While the workings of algorithms can be complex, their function is fundamentally simple. They track what we do in order to determine what we like so they can give us more of it. They keep us engaging, keep us using the social media platform, watching the video site, buying from the online store. The algorithms have grown so sophisticated that we can now visit a web site that provides a service confident not necessarily that we will find what we are looking for—we may not be looking for anything—but that the algorithm will provide what we would have wanted had we been wanting. It anticipates desire, and, where none exists, stokes it, then offers more content, endlessly, to keep it keen.

For example, say you, the reader, were to leave this essay for a moment to watch a video on YouTube. Even if you do not have a video in mind, the site definitely does, thanks to algorithms that allow it to recommend an unlimited supply of videos tailored to users’ preferences, anticipating what you might like to view based on what you watched before. Over 70% of YouTube views are now recommended by the site’s algorithms. More and more, we do not go to watch videos on YouTube, we simply go to watch YouTube. It is a difference which captures the power of algorithms to narrow our vision. Where once the internet might have been a place for true exploration, it is now a place where attention is increasingly channeled and consolidated by algorithms and the companies that control them.

This is perhaps most true on social media sites, where algorithms are prime movers. These sites have become centers of online gravity, not just for sharing and connecting, but also as hubs of news and information. With 53% of U.S. adults saying they “often” or “sometimes” get news from social media, these sites have become as core to how we learn about and interpret the world as the nightly news was 50 years ago.

This challenges our ability to understand the moment and the issues and ideas that shape it. When our media consumption is driven by algorithms, we risk overlooking much that is important simply because it does not align with what these systems think we want, what they think will keep us engaging with a site’s content. This can have the effect of homogenizing our media diet, feeding us more of what we already click on. Variety, when it is found, is no longer in the diversity of content, but in its intensity, its extremity. Where once, for example, we may have read news from a liberal perspective and a conservative perspective, the algorithm now gives us more of what it thinks we like, which tends to be what we already agree with, and the more it gives, the deeper it is likelier to take us down rabbit holes of ideological uniformity, performative outrage, and demonization of the other side. In past writing, I have emphasized the importance of embracing unifying ideas rather than polarizing ones. Algorithms, in many ways, are a strong headwind against ideas that could bring us together.

This has had downstream effects on all kinds of forces and trends in the present moment, starting with how we think. There is a narrowing of the mind that comes when we outsource so much of our intellectual and cultural life to algorithms. Once, the business of reading, talking, taking in entertainment meant coming into contact with what is outside of ourselves. Words and ideas, books and films—these reflect the perspectives of others, perspectives we might not share. To venture among them was to encounter much that is pleasing but also much that is uncomfortable, much that takes time to reckon with and understand. Algorithms help us to avoid this messy, necessary process. They keep us focused inward, on what we already know, or think we know. By influencing how we think, they influence what we do. And what have we done in the age of the algorithm? We have become more polarized, more generating of heat rather than light, less able to see the value in different points of view, and more likely, as a consequence, to develop intellectual blind spots that place at risk everything from the future of academia to the integrity of movements that aspire to create a better, healthier world.

Creating such a world means, then, finding ways to “think outside the algorithm.” How can we do this? Algorithms are not going away. They are, and will likely remain, central to how we consume information on the internet, how the flow of knowledge travels. However, I have long argued that while structural forces are core to shaping outcomes in health and behavior, it is also true that we have, all of us, individual agency and can take steps to mitigate some of the ways that the forces in this moment can lead us astray. We could hope for regulatory oversight, which may or may not come, or we could make it a habit of mind to push back against the algorithm. As I have reflected on this in the past while, it seems to me that three such habits can help guide us towards a healthier engagement with our algorithm-driven media ecosystem.

First, we can be deliberate in our media consumption. This means actively following sources outside our normal online networks. This can involve seeking sources that represent points of view that differ from our own, or using an app designed to show us news from a variety of places and perspectives. It can also mean looking beyond the digital world for what we read. It can be easy to forget that print media is still in existence, but it is, and well worth perusing.

Second, we should be skeptical but not cynical about the content we encounter. We can, and should, question what we read, even—particularly—when it comes from a source with which we are inclined to agree. It is important to remember that we expose ourselves to different perspectives to improve our thinking, not to let others do our thinking for us. Oscar Wilde wrote, “Most people are other people. Their thoughts are someone else’s opinions, their lives a mimicry, their passions a quotation.” This has become an occupational hazard of life in the digital age, when whole intellectual identities can be constructed out of what algorithms feed us. A healthy skepticism can provide a sense of needed perspective, letting us know when we risk being carried away by the online current.

Third, when we sense the dynamics of the discourse becoming counterproductive, we should be willing to change the conversation. For example, when we sense our spaces, either online or in “the real world,” are becoming too defined by ideological bubbles and groupthink, we should work towards new norms of curiosity in our discourse, with less emphasis on consensus for its own sake. Such norms deeply inform our work at the School of Public Health at Washington University in St. Louis, where they are core to what we teach our students and how we aspire to conduct ourselves as thinkers and teachers in the field of health. A great school, and an effective movement for health, depends on accessing new ideas, new ways of thinking, and the insights that come from unfamiliar—sometimes unexpected—sources. It depends on building new channels where information can flow, new pipelines from thought to action, towards the creation of a healthier world.

I suspect that algorithms will are going to continue to shape what we see for a long time. I would like to think that they need not define what we choose to see. Perhaps exercising discernment in our consumption of media, choosing to be deliberate, skeptical, and being willing to change the conversation when we need to, can nudge us towards an intellectual climate worthy of this moment and a future worthy of our best aspirations.

__ __ __

Also this week

A second piece with Austin Kilaru and Julie Donohue on “Study design and methods for health policy and services research: A look back and a look ahead” in JAMA Health Forum.

Also in the journal, “Health implications of recent policy changes affecting LGBQ+ populations”.