Lalo Martins

I’ve watched enough Outer Wilds playthroughs to develop a theory that is either insightful or a sign I should touch grass. Possibly both.

One of the things Outer Wilds does—quietly, insistently—is ask what different kinds of minds look like when they’re allowed to shape a civilization. Not as labels. Not as diagnoses. As ways of being intelligent.

Seen through that lens, the game starts to look less like a puzzle box and more like a thought experiment in neurodiversity.

(By the way, Outer Wilds is a game that has to be experienced for yourself without previous knowledge — and this article is packed with spoilers. If you haven't played yet, do us both a favor: bookmark this, stop reading right now, and come back after you finished it. Thanks.)

The Nomai: “Identify and Explain”

Let’s start with the Nomai.

If you dig deep enough — especially around the Quantum Moon — you eventually learn that Nomai philosophy can be reduced to two verbs:

Identify. Explain.

These are the two tenets of Nomai philosophy; to seek out and to understand is our way of living.

These aren’t just scientific habits. They’re moral ones.

To identify is to resist premature conclusions: to ask what something actually is, where its boundaries lie, how it behaves under observation. To explain is to build a model that survives contact with reality and can be shared with others.

This is an entire civilization built around what, in human terms, looks a lot like autistic cognition at its healthiest: intense curiosity, comfort with abstraction, systematic thinking, externalized memory, and zero cultural stigma around saying “we were wrong, let’s revise.”

(Also: they awkwardly put their feelings into written words. “I don't know how to be me without you.” “In romantic matters, her density rivals a neutron star.” They cautiously probe around each other. They engage gingerly but productively with their rivals (Poke/Cassava, Pye/Idaea). Very Spectrum communication style.)

The Nomai don’t conquer. They don’t mythologize themselves. They document failures with pride. Their cities feel like collaborative notebooks stretched across a solar system.

They are unapologetically curious—and tragically vulnerable to a universe that does not care how good your epistemology is.

The Hearthians: ADHD at the End of the Universe

Now contrast that with the Hearthians.

By the time they evolve, the universe is old, resource-scarce, and actively hostile. Long planning horizons are a luxury. Survival favors minds that are:

• adaptable rather than optimized,
• exploratory rather than methodical,
• good at improvisation under pressure.

In other words: ADHD as an evolutionary feature, not a bug.

The Hearthians are brilliant generalists (they have to: their population is minuscule). They build rockets with wood, vibes, and courage. Every single tech device they use has duct tape somewhere. They outsource executive function to cameras and autopilot (thanks, Slate). They are distractible, fearless, and absurdly resilient.

They don’t understand the universe deeply (okay, except for Gabbro) — but they engage with it constantly. And in a chaotic environment, that matters more.

Quartz, the Hatchling: The Convergence

And then there’s the protagonist. Or, as I like to call them, Quartz.

Quartz is what happens when you give an ADHD brain infinite retries and a strict 22-minute timebox.

“What if,” Quartz seems to ask, “I could do everything… in the same day?”

This is the Superman-of-ADHD fantasy realized. Hyperfocus without permanent consequences. Novelty without punishment. Failure as information instead of shame.

Quartz succeeds not by being the smartest individual, but by borrowing ways of thinking as needed: Nomai patience here, Hearthian boldness there, infinite sass where there's someone to talk to, relentless curiosity everywhere.

Intelligence, in motion.

Quartz also stops us from asking “but which approach is better?” because the question then seems stupid. There's no such a thing. They're optimized for different problems — and they beautifully complement each other. That's why we call neurodiversity, well, neurodiversity.

The Final Campfire: A Mind Assembled

By the time you reach the final campfire, it stops feeling like a reunion and starts feeling like a metaphor.

Each character represents not a personality, but a faculty:

Esker Patience and reflection. Esker is the one who stepped away from the noise and kept listening. Solitude not as withdrawal, but as signal amplification. The part of intelligence that says: slow down, the universe will still be there. Without Esker, you rush past meaning. Telling: you don't have to go “gather” Esker. He's already there. He's always been there.

Gabbro Calm, openness, acceptance of uncertainty. Chill™. Gabbro doesn’t solve quantum mechanics; they coexist with it and turn it into art. This is epistemic non-attachment. “I might be wrong, and that’s okay.” The antidote to panic. The hammock at the edge of the abyss.

Chert Mathematics, strict observation, uncomfortable truth. Chert keeps watching even when the data becomes existentially rude. No avoidance. No narrative sugar. Just: this is what the universe is doing. Intelligence that doesn’t look away. And appropriately, it collapses into panic, since that's the failure mode of this faculty — but that doesn't make it less valuable.

Riebeck Memory, history, continuity. The past as a dataset, not nostalgia. Riebeck is the reminder that understanding is cumulative, that knowledge persists through careful preservation. Fear doesn’t disqualify you from insight. It just means you proceed gently.

Feldspar Curiosity and adventure, unfiltered. The willingness to go first, to test the impossible, to be gloriously wrong at high velocity. Feldspar is raw exploratory drive—the spark without which nothing ever starts. Or at least until he gets tired… because taking a break is important too.

None of these traits is sufficient alone. Together, they form something like a complete intelligence: reflective, courageous, grounded, and adaptable.

And then there’s Solanum stepping in to integrate them with the one thing that makes it all work.

Is Solanum the Real Main Character? (Semi-Serious)

This is where I allow myself a little trolling.

Solanum’s arc can be read as the actual philosophical spine of the game.

Child: What if the Eye is evil? Teen: Maybe it doesn’t mean anything. Young adult: The universe is, and we are. Ancient quantum hermit: I think of us as friends.

That’s not a plotline. That’s intellectual maturation.

Solanum embodies the hardest cognitive skill of all: updating one’s worldview without losing oneself. She is the 看板娘 of changing your mind in response to evidence—without panic, without nihilism, without clinging to certainty.

If the Nomai give us the method, Solanum gives us the emotional maturity to use it.

So What Is Outer Wilds Actually About?

It’s not a puzzle game. Calling it that does real harm.

It’s a game about how intelligence works when it’s allowed to be curious, fallible, and brave. About how different kinds of minds survive — or don’t — under different cosmic conditions. About learning deeply enough that, when explanation runs out, you can still choose to leap.

You do have to YOLO into the Eye in the end. You do probably crash into the sun at least once on purpose. And that’s the point.

Understanding gets you to the edge. Meaning requires the jump.

Intelligence, in Outer Wilds, isn’t a stat you possess. It’s a process you practice—until you know enough to let go.

And maybe that’s why the game hits so hard: it quietly insists that becoming smarter isn’t about becoming different.

It’s about becoming willing.

We tend to talk a lot about “smart people.” Smart engineers. Smart founders. Smart hires.

Teams, however, do not fail or succeed because of individual intelligence. They fail or succeed because of how information flows, how decisions get updated, and how disagreement is handled once reality starts pushing back.

In other words: intelligence at the team level is not a trait. It’s a process. (Not that it’s a trait at the individual level either—but that’s a separate topic.)

This is not a story about ideal teams. I haven’t worked in many of those. What I have seen—repeatedly—is what makes teams fail, often long before anyone is willing to name it. The patterns are surprisingly consistent.

---

Intelligent teams optimize for learning, not for being right

Every team says it values learning. Far fewer design for it.

Intelligent teams don’t try to eliminate mistakes. They try to make mistakes cheap, early, and informative. They shorten feedback loops. They test assumptions before turning them into commitments. They treat early wrongness as progress, not embarrassment.

This is not about being cautious or slow. It’s about reducing the cost of course correction. Teams that require certainty before acting tend to move confidently in the wrong direction for a very long time.

Blameless postmortems are not a “nice culture practice.” They are epistemic infrastructure.[¹][²]

---

They make updating beliefs socially safe

Many teams are good at gathering data and terrible at acting on it.

The reason is rarely technical. It’s social.

In unintelligent teams, changing your mind is expensive. It costs status. It invites scrutiny. It feels like backtracking. So people quietly defend decisions long after the evidence has shifted.

Intelligent teams normalize updating. Decisions are treated as hypotheses, not verdicts. The quality of a decision is separated from the outcome it happened to produce.[³] “We learned something” is allowed to be a win.

When changing your mind doesn’t damage your standing, people do it earlier—and that’s when it still matters.

To be clear, this isn’t universal.

My current team is, let’s say, still improving at gathering data—but they are exceptionally good at acting on what they have. Decisions move. Feedback gets incorporated. When something stops making sense, we don’t pretend otherwise.

That’s one of the things that makes me proud to work with them.

---

They externalize thinking relentlessly

Smart teams don’t rely on memory, heroics, or oral tradition.

They write things down. They draw diagrams. They keep decision logs. They document not just what was decided, but why.

This isn’t bureaucracy. It’s shared cognition.

Externalizing reasoning reduces cognitive load, exposes hidden assumptions, and makes disagreement inspectable rather than personal.[⁴] Writing is a forcing function: if a decision can’t be clearly explained, it probably isn’t fully understood yet.

Most experienced ICs already know this intuitively. The challenge is making it visible, repeatable, and supported at the team level.

---

They design for cognitive diversity instead of fighting it

A common failure mode—especially in agile organizations—is misinterpreting “cross-functional team” to mean interchangeable people.

The justification is usually well-intentioned: reduce bus factor, avoid silos, keep the team functioning if someone is unavailable.

The result, however, is often a homogeneous group optimized for parallel execution.

Parallelization makes known work faster. It does not make unknown work solvable.

Research on collective intelligence consistently shows that groups perform better on complex tasks when they include diverse cognitive styles and perspectives, even when individual ability is held constant.[⁵]

A four-core machine with a GPU and an NPU will outperform a 32-core CPU on real workloads for exactly this reason. Heterogeneous systems handle heterogeneous problems better.

Friction isn’t a personality problem. It’s a signal that multiple models are in play—and that something important may be hiding there.

---

They treat disagreement as signal, not threat

In failing teams, disagreement feels personal. In intelligent teams, it feels informative.

This doesn’t mean endless debate. It means taking the time to understand why someone disagrees before moving to resolution. “Help me understand your reasoning” is not a rhetorical move; it’s a diagnostic one.

Teams that can separate task conflict from relationship conflict consistently outperform those that suppress disagreement in the name of harmony.[⁶]

The goal isn’t consensus. It’s calibrated alignment: moving forward with a shared understanding of tradeoffs, risks, and unknowns.

---

Strengths can mask weaknesses—and that’s okay (for a while)

A team that acts decisively on weak signals can afford to miss some early warnings. When issues that are caught get addressed quickly and effectively, improving detection doesn’t always feel urgent. Concretely and empirically, it may not be the current bottleneck.

I mentioned earlier that my current team is good at acting on information and less good at collecting it. That’s not a contradiction—it whiffs strongly of causality.

When response is strong enough, gaps in sensing simply don’t compound fast enough to demand attention.

The risk, of course, is that scale changes the math. What was once absorbable noise can turn into systemic blind spots. Intelligent teams revisit these tradeoffs deliberately, rather than assuming yesterday’s strengths will automatically scale.[⁷]

---

What intelligent teams actively avoid

They avoid hero culture. They avoid speed without feedback. They avoid confusing confidence with competence.

None of these failures are dramatic on their own. They compound quietly.

---

Intelligence is a team habit

Intelligent teams are not magically assembled. They’re shaped by small, repeated choices: what gets rewarded, what gets documented, what gets challenged, and what gets ignored.

This isn’t about hiring smarter people. It’s about building environments where people are allowed—and expected—to think well together.

This is the kind of environment I do my best work in. Not because it’s comfortable, but because it’s honest.

And honesty, in complex systems, turns out to be a competitive advantage.

---

References

[1] Edmondson, A. (1999). Psychological Safety and Learning Behavior in Work Teams. Administrative Science Quarterly. [2] Dekker, S. (2014). The Field Guide to Understanding ‘Human Error’. Ashgate. [3] Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux. [4] Clark, A. (2008). Supersizing the Mind: Embodiment, Action, and Cognitive Extension. Oxford University Press. [5] Woolley, A. W. et al. (2010). Evidence for a Collective Intelligence Factor in the Performance of Human Groups. Science. [6] De Dreu, C. K. W., & Weingart, L. R. (2003). Task versus Relationship Conflict. Journal of Applied Psychology. [7] Senge, P. (2006). The Fifth Discipline. Doubleday.

I want to argue for an unfashionable idea—one that tends to make people uncomfortable for reasons that have very little to do with evidence:

Intelligence is not primarily an innate trait. It is a process.

Not entirely a process. Biology matters. Injury matters. Development matters. Pretending otherwise would be irresponsible. But for the vast majority of people, the differences we casually label as “smart” or “stupid” are far better explained by how a mind operates than by what that mind is made of.

This isn’t a motivational slogan. It’s a claim about mechanisms.

What do we actually mean by intelligence?

Modern definitions—across cognitive psychology, neuroscience, and even artificial intelligence—converge on a familiar cluster:

• solving problems,
• making decisions under uncertainty,
• adapting behavior when new information arrives.

Not raw memory. Not speed alone. Not vocabulary size. Those are ingredients, not the meal.

What’s striking is that all three depend less on storage than on retrieval, control, and revision. Intelligence doesn’t live in what you know; it shows up in how you move through not knowing.

A mind can hold vast amounts of information and still behave unintelligently if it cannot:

• access the right knowledge at the right moment,
• tolerate uncertainty long enough to explore,
• or revise its internal models when they stop matching reality.

That already points away from intelligence as a fixed quantity.

The failure modes that get mislabeled as “low intelligence”

In everyday life, when people come off as “stupid,” what’s usually being observed isn’t lack of capacity. It’s one of a small number of predictable process failures.

Rigidity. The refusal to update beliefs in the face of evidence. This is the most damaging failure mode, because it hijacks intelligence itself: reasoning power gets redirected from learning to defending.

Avoidance. “I’m not a math person.” “This is too hard.” “I don’t like thinking about this.” These aren’t diagnoses; they’re self-protective narratives. They reduce discomfort now while quietly training the brain not to engage later.

Premature certainty. Finding an answer that works well enough and then clinging to it long past the point where it explains what’s happening. Shortcuts can be intelligent—until they aren’t. The skill lies in knowing when to let go.

Notice what’s missing here: raw intellect. None of these failures require a small brain. They require habits.

Intelligence as a loop, not a trait

A more useful way to think about intelligence is as a feedback loop:

1. Observe
2. Hypothesize
3. Test
4. Update

If this feels familiar, that’s not an accident. It’s the classical formulation of the scientific method. There’s a reason our species became collectively smarter once this loop was formalized, taught, and socially rewarded.

Break the loop anywhere and intelligence collapses—no matter how capable the underlying brain might be.

Crucially, every part of this loop is trainable.

Neuroscience supports this view. Executive function (the ability to plan and inhibit), attentional control, error monitoring, and cognitive flexibility are not fixed traits. They are plastic systems shaped by use, stress, feedback, and reward. Brains specialize in what they practice.

Practice:

• defending beliefs instead of revising them,
• avoiding confusion instead of sitting with it,
• speed instead of accuracy,

and the mind will reliably behave as if it were unintelligent—even if the hardware is perfectly fine.

Why this idea makes people uneasy

If intelligence were fixed, failure would be tragic but blameless. If intelligence is a process, improvement becomes possible—and that implies responsibility.

Not blame. Agency.

The claim here isn’t “anyone can become a genius.” It’s more modest, and more hopeful:

Many people operate far below their cognitive potential because they were never taught—nor rewarded—for using their minds well.

We reward confidence more than calibration, certainty more than accuracy, speed more than understanding. Then we act surprised when people struggle with complex, ambiguous systems.

The uncomfortable but optimistic conclusion

If intelligence is a process, then it can grow.

Not infinitely. Not uniformly. Not without limits.

But meaningfully.

People become more intelligent when they learn to:

• treat being wrong as information rather than failure,
• tolerate confusion without rushing to anesthetize it,
• revise their models of the world without experiencing it as a threat to identity.

None of this requires a different brain. It requires a different relationship with thinking.

That’s not self-help. It’s cognitive mechanics.

Which leads to a reframing I find hard to unsee:

Intelligence is not something you are. It’s something you do.

Sometimes clumsily. Sometimes elegantly. Almost always better with practice.

The most important question, then, isn’t “how smart am I?” It’s “how willing am I to update when reality disagrees with me?”

That question, unlike an IQ score, has an answer you can actually work on.

So you want to play Outer Wilds on stream or video, and you heard it should be played completely blind, but you're still worried about providing the best content for your viewers? Fear not! Huge Outer Wilds enthusiast here, I've watched countless people play it (including two couples as of this writing — couple streamers are my new obsession!), and I know a bit about content creation. I prepared a spoiler-free list of tips to help you out: practical content things, tips to maximize your enjoyment (and your audience's), and a couple of borderline-spoilers to prevent missing content. Keep those in mind, and enjoy the game!

It is not a puzzle game.

That's maybe my main takeaway. Outer Wilds is an archeology game, if you're unfamiliar with the subgenre, think of it as an exploration/story game. It can be played as a puzzle game, but it's a pretty mediocre one (except for one explicit puzzle section, you'll know when you're there), while it's one of the best archeology games out there. I've seen a few creators approach it as a puzzle game and walk out disappointed.

Content creator things.

The best place for your camera or avatar (or, really, the only good place) is the middle of the right-hand side. Only the DLC ever puts anything important there, and even that is easy for the audience to deduce by context, so nothing will be missed.

Try to end every video or stream in the campfire (where the game starts). It's the best way to make sure nothing important happens when you're not recording/streaming.

If you like recaps, a good way to start a stream/episode is reviewing the log.

There's a discord called The Interloper, created specifically for viewers to chat with each other while watching Outer Wilds streamers, without fear of spoilers. If you're okay with that sort of recommendation, I'd suggest pointing your viewers there.

The Outer Wilds viewer community has a fun convention. The emoji ::) basically means, “oh boy I wish I could tell you something but I'm holding back spoilers”.

Tutorials.

If you're the type to rush the start, two non-spoiler tips on the village:

• Every child in the village is a tutorial. There's one tutorial that isn't a child, but it's also the longest and most skippable. Absolutely stop to talk to all children you meet, their tutorials are hugely helpful.

• Do read everything in the observatory: both in the beginning and the end of the game.

Progress.

Don't rush. If you feel stuck, stop and have a careful look around. If you feel you're done with an area, stop and have a careful look around.

On the other hand, don't frustrate yourself either. If you have looked around carefully and tried everything, maybe you're just not ready for this area yet. Make a note of it and come back later.

On that note: when viewing things with multiple pages (especially in the DLC), make sure you paged all the way to the end.

Come back to all travellers again and again as you learn more things. They're not one-shot.

There is no “correct” order. The game is designed so you can do it your way and everything will still come together.

The log is your best friend. Lavish it with attention.

There are optional, hidden easter eggs, so feel free to play creatively, your way.

The DLC is best played after finishing the main content.

There is no section 5.

There's so much more I'd like to say, but it's better to discover it yourself. Enjoy! And make sure to smell the pine trees along the way.

Any project that uses JavaScript or one of its variations needs a package and script manager. For historical reasons, we have two options in common use: npm, the Node Package Manager, developed by npm, Inc (a subsidiary of GitHub, which is a subsidiary of Microsoft), is the standard one, installed along with Node; and Yarn, created by Facebook but currently developed by a regular open source community, is an alternative client for the npm system with a slightly different interface.

Yarn was created at a time when development of npm was in a rough patch. At the time, it had better performance, additional features, and provided more deterministic behavior. But npm eventually caught up, and now the differences are essentially only in interface.

Since npm comes with the Node installer, essentially everybody who works with JS (or one of its variations) has npm. Yarn has to be installed separately. And being the default, npm is used in most examples around the web.

Both are open source and hosted on GitHub (npm, Yarn). A quick look at the repositories shows npm is substantially more active in every metric.

In the end, as with most things in tech, you should use the best tool for the job, according to your project and your team. My recommendation is:

• If you work for Facebook or a subsidiary thereof, you should use Yarn.
• Everybody else should use npm and try to forget Yarn exists.

Hope that helped, thanks for reading.