A quick fun fact about Nevada: it is the least-densely populated state in the union. There’s Las Vegas and Reno/Carson City, the Tahoe towns, then some smaller municipalities like Elko, Pahrump, and Fernley. But other than that, you don’t drive past many other “Welcome To…” highway signs with a population number greater than 5,000. Driving through Eastern Nevada, there’s a real chance you can make it through stretches of Highway 50 or Highway 6 and not see another sign of human life for a couple of hours at a time.

Nevada is also home to the most named mountain ranges in the U.S. It boasts Pilot Peak, Wheeler Peak, Boundary Peak, and Charleston Peak just to name a few, all erupting from the high desert expanse. There’s a certain charm about this combination of natural prominence and desolation. A humbling effect mixed with a small relic of ancestral exploratory drive, the sentiment of being 250 miles from the nearest gas station with no cell reception somewhere between Ely and Tonopah and realizing that a flat tire could become a real problem.

I took one such road trip last year, but noticed some slight alterations to the landscape’s barren allure. Driving 30 minutes east of Reno used to involve a pit stop in Fernley followed by endless desert. Now, it spits you out into industry. Sprawling distribution centers, the Nevada Tesla Gigafactory, and graded hillsides with track home developments to boot. And, in the last few years, the appearance of the all-powerful AI Data Centers.

Nevada’s deserts have long offered the assumption of free rein grounds for industry. What else are miles upon miles of emptiness good for? From mineral mining operations to nuclear detonation test sites and now onto (quite literally) moving mountains for Gigafactories and data centers, it wouldn’t possibly bother anyone to see this landscape change, right?

These data centers have been getting a pretty bad rap of late, and it’s not terribly difficult to see why. For all the compute they put out, data centers demand utilities at a staggering rate. The question remains whether existing grids can accommodate the growing power demand, let alone surrounding communities.

Just within the last couple of weeks, NV Energy (my own power supplier) announced that they will be stopping their partnership with Liberty Utilities, which provides power to some ~50,000 Lake Tahoe residents. NV Energy decided that they need to re-route that supply to new data center projects in northern Nevada, which are creating “unprecedented times” and requests to triple the provider’s peak supply. Business seems to be booming. The residents, however, are going to have to figure out how to weather rate hikes in a yet-unfilled replacement power supply. A power source that was once dedicated to serving the human beings living in an area has effectively been redirected to serving the machine.

This kind of thing isn’t just happening in Nevada. The more threads I pull, the more geographies it seems to impact. Farm communities in the midwestern U.S.1 and central California are watching their agricultural land and resources get gobbled up by developers. Highly revered natural landscapes are not spared either; I’ve seen plans floated for data center builds a mile from Horseshoe Bend in Arizona and next to Montana’s Blackfoot River.

This is yet another unfortunate example of asymmetric risk profiles. When it comes to data centers in agricultural towns and nature’s landmarks, the people who make the decision to build and the people who live with its consequences are never the same people. The operators and investors who set up contracts from Palo Alto offices absorb the upside of successful data center builds, while locals in Wisconsin, Montana, Lake Tahoe, and elsewhere absorb the downside of a tainted landscape and diverted resources.

I have seen footage from town hall meetings in rural communities pertaining to data center builds. They tend to draw crowds that I have to believe are comparable to the local county fairs, brimming with people excited to give their $0.02 and stick it to the legislators underwriting the permits. In almost all cases, the citizens in the communities where these data centers are going up do not want to see the data centers go up. So a farmer drives to their community center, signs in on the check-in sheet, and gets three minutes on the microphone.

But this charade is usually nothing more than lip service, astroturfing, and consent manufacturing. The permits were already filed, the risks to the developers and AI investors already externalized, and Farmer Joe was given the illusion of a voice. The town hall meetings are info sessions on Mustafa Suleyman’s coming wave, not civic participation.

But wait, don’t I appear to be leaving out a key detail here? These data centers will create jobs, Danny. They’ll stimulate the local economy, bring in more investors, and make the local population boom. The Colorado River might get strained and the neighbors a couple of towns over might see their power bills double in the winter, but there will be more jobs.

For all the liberal economics the U.S. champions, technological innovation done poorly can take a rather Soviet view of resources: a natural landscape, a natural reserve, or a utility service has no value beyond what industry can extract from it.

I hope for America’s organic and regenerative farmers to succeed and for their communities to retain their livelihoods. Yes, I want local communities to have rich career markets, but I also don’t think that those jobs should come at the expense of preexisting citizens.

I consider myself a Teddy-Roosevelt-esque conservationist, so I also see the need for humans to be responsible stewards of our ecosphere and not its exploiters. I used to make fun of people causing projects to come to a screeching halt for considering the impact on local trout populations. Now, I find myself understanding the sentiment. I see (unnamed) colleagues on my LinkedIn feed who run VC funds from New York City championing a groundbreaking million-square-foot data center east of Reno. One that will make a lot of money for an AI company so they can turn around and sponsor a new building for the local university. But these VCs have probably never fished the Truckee River, have never hiked the Sierras, and haven’t driven through the powerful desolation of Eastern Nevada’s deserts.

I have done these things, and it’s part of why I find NV Energy’s “unprecedented times” concerning. I know I’m not alone, because there are plenty of other communities seeing these same trends. I just happen to be watching them in my Nevadan backyard, and I’m not so sure I want to see the landscape dotted with data centers.

My goal is not to drift into doomerism. I want to root for technological progress, but I hold that it needs to progress responsibly and thoughtfully. Dealing with AI and its infrastructure responsibly will be a test of incentive structures. The question might need to shift from, “Can our power grids keep this up?” to “Who are we even building all this stuff for anyways?” What is the point of building something that “provides jobs and boosts the local economy” if the people who make up that local economy are having their resources siphoned into a black box?

AI and its growth today are dependent on these data centers. It’s not a debate about whether AI is useful or not. It is an extremely valuable tool that has some very real use cases: cancer screening, democratization of health and biometrics, and education are just a few that come to mind. But just as nail guns are tools and not toys, the infrastructure is evidently not up to speed with AI as a consumer toy.

Who’s paying for it when it’s not used responsibly but demand persists? Right now, the answer is those who had the most to lose, but least amount of say in the matter. Until we can fix the infrastructure layer for AI, I would rather be able to ask vendors at my local farmer’s market about how to grow seasonal vegetables than watch their farmland get reconstituted for data center use so I can ask Claude instead.

“Zero,” or at least “as little as possible.”

An unusual goal for the world’s most successful venture capitalist, the pursuit of zero. But that something is introspection, which Marc Andreessen believes is no good professionally or personally and has been a liability for the last hundred years. It allegedly emerged from Vienna in the early 1900s, a founding pillar of the Freudian Therapy Industrial Complex.

For all my personal interest in startups and frontier tech, I’ve admired Marc Andreessen for what he’s built at a16z and for the lasting impact Mosaic had during its tenure. He’s a startup guy, an innovator, one of the best in the space. But he’s collected some bad press and critics over this particular take, and I’ll add my name to that list. But before I do so, I’d like to steelman the merit of his perspective.

What he’s getting at isn’t totally outlandish:

“People who dwell in the past get stuck in the past.”

On its face, I agree. Guilt is not a useful feeling. No mover, no shaker, no person in the arena who has had a lasting positive impact on humanity has been universally liked at the time of their work. It’s been said that it makes sense to strive to be polarizing, because it’s an indicator of having an impact. Analysis paralysis can be real, so move fast, break things, don’t look back, don’t dwell on the past, figure out the parachute after you’ve already jumped, and certainly don’t introspect. That’s how you get things done.

What Andreessen describes about getting stuck in the past is true, and others have already pointed out that harping on the past is a case of rumination more than it is introspection. But it was his subsequent double-down on X that made me think this wasn’t just a semantics problem, and that he might genuinely scorn the practice of self-examination.

“If you go back 400 years ago, it never would have occurred to people to be introspective.”

In more ways than one, I think this is objectively untrue. For instance, some Great Introspectors of History that you might have heard of include: John Locke, Anne Conway, Buddha, and Jesus Christ. The Renaissance and the Enlightenment were rooted in introspection. And even if introspection is a gimmick of the Viennese salons of the early 20th century, Albert Einstein and Kurt Gödel emerged from these circles as frontier scientists that set the stage for many of the technological innovations to which Marc Andreessen owes his livelihood today.

It’s not hard to poke holes here, and plenty of others on LinkedIn, X, and The Atlantic already have. Such a perspective is to take the “move fast and break things” mantra to its logical extreme. It reads as if someone saw Dostoyevsky’s Nikolai Stavrogin as an ideal to emulate instead of a warning parable.

The obvious, gut-check reaction is to see someone say they practice zero introspection and label them a sociopath. I don’t think that’s quite right either. I wonder if Marc Andreessen isn’t just a little bit confused, making provocative statements from an ivory tower and steering into the skid without fully reckoning with the possibility that his own success might owe something to his contemplations.

Introspection does not mean dwelling in the past; it’s studying the past to understand how it shaped the present, studying the present to understand how it might shape the future, and probing what one’s role as an individual might be in that dance. Before too long, the present will become the past, and the future will become the present that was shaped by decisions made today. So, as a rule of thumb, it makes sense to think carefully about how those things interact and about one’s own role in the interaction.

Metaphysics and therapy sessions aside, introspection makes one a better business operator. Understanding and contemplating multi-order effects of decisions made in the past and present is Systems Thinking 101. Introspection, especially in the startup investment arena, is paramount to success, whether Marc Andreessen wants to acknowledge that or not. How many times has he had to work through speculation: If I make X bet on Y founder, I expect Z to happen in the future. This bet may be unconventional, but I trust my intuition and understanding of the landscape to confidently place this bet.

That is introspection, and it’s required to be a successful systems thinker, builder, operator, and investor.

Worth noting too is that introspection and deep original thought, rather than high-speed pattern matching, remain key advantages of human cognition over machines. As AI systems accelerate, the ability to examine oneself and one’s decisions is becoming more valuable, not less. Introspection is a uniquely human skill that should be revered, not chastised.

I’m personally proud of my inclination to self-examine. Introspection could be the thread that holds my entire professional and personal thesis together, one of my own biggest value propositions. It’s how I approach my writing. It’s how I approach my own startup investments and professional projects. I can’t speak for Marc Andreessen and anyone else in the space, but I want to lean into exploring how my decision making relates to a greater system instead of running from it and denying any responsibility for its consequences.

Which brings me to my real problem with “zero introspection.” It’s a liability waiver more than it is a productivity hack. Such an operating thesis would not have held water for much of human history. As a small thought experiment, let’s place Marc Andreessen in a jungle 25,000 years ago in a tribe of 150 fellow Amazonian hunter gatherers. In such a tribe, a cheater detection mechanism would almost certainly snuff out his aversion to self-reflection and considering the ripple effects of his actions. His fellow tribe members might cut him off from shares of hunted game or forget to inform him about the swimming hole full of piranhas. But with a16z operating in a country of ~330,000,000 people and many billions of dollars in AUM, the consequences of anti-introspection get externalized onto everyone else with the upside borne by a few.

To point the finger and offer no replacement framework is of no use to anyone seeking wisdom. So I’ll posit that instead of championing zero introspection, perhaps advocating for ownership is a better alternative. The frontier technology space seems to celebrate the voice that says,

“I’m an innovator. I’m moving fast. Get out of my way. If I break something, screw you, somebody else can clean it up.”

But another voice is out there, the one that says,

“I’m an innovator. I’m moving fast. I might break a thing or two for better or for worse, and if things go awry, I’ll own and reconcile that.”

I, and many others, would probably like to see the latter championed more frequently.

Leave a comment

Share Danny Nikolai

Most days this week have demanded that I get up around 04:30-05:00 AM. I wake up a little bit confused, well before sunrise, parched from the dryness of the high desert. The timing, by itself, isn’t the worst thing in the world, and is pretty consistent with my normal circadian rhythm. But instead of waking up to a calm house with a dedicated space to stretch and do some breathwork first thing, I’ve risen alongside heavy panting, the clanging of paws on metal, and whining.

We rescued a dog last week from a local shelter. She’s a morning lark, needs to go outside, and I was voluntold by my fiancee to handle taking her out in the morning.

We named her Miso Hamachi Ragnar Danneskjöld (she goes by Miso) because of her miso-golden eyes. Getting a dog is a big step, and we want to make sure we do it right. This means setting boundaries, giving her proper attention, and putting our own mark on her whims after a year of her living in a shelter. She is, in some ways, a new project that I’ve added to my plate. But she’s different from a writing assignment or an engineering problem, because I can’t bargain with her for a couple extra days before she needs to go for a walk or get fed.

I tend to be pretty Type A. I’m an engineering Ph.D., am a systems thinker, I love to talk about baseball statistics like a savant, and I try to build daily routines for myself with intention because I know exactly how they work for me. I like to capitalize on my early-morning chronotype to get my high-leverage work done before noon, am typically in the gym or out for a run/bike/swim by 06:30ish in the morning, and I try not to take meetings on Fridays. I like to time block for deep writing sessions or engineering problem solving, and have a plan going into my days, weeks, months ahead to minimize decision fatigue and internalize my locus of control. It’s been said that too much flexibility without any structure devolves into chaos, and my personal procedures definitely bias toward that structure.

Miso, however, does not give a damn about my daily routine, goals for the day, or professional deliverables. Her daily goals are walk, ball, sniff, play. On top of taking care of her at home, we’ve spent a lot of time in the last week running errands to get food, supplies, leashes, meet with trainers, meet with local boarding houses for when we travel, on and on. She is a pitbull/German pointer mix, a unique sort of freak athlete. The strength and broad frame of a pit mixed with the hyperfocus, olfactory senses, and speed of a pointer means most errands and walks take twice as long as they should when we’re trying to teach her to heel and walk without tension in the leash.

There are two truths here: Miso is a love, and it’s great to have her around. She has also severely disrupted my operating system in the last week. Trying to carve out time to answer a few emails between dog errands, breaking a deep work writing rhythm every five minutes to get the dog off the couch, limping into the gym for a truncated workout at noon instead of bright and early when I feel fresh. All the while trying to remember that just as flexibility without structure turns to chaos, structure without flexibility creates fragility. I hear echoes of Bruce Lee almost mocking me for my frailty:

“... adjust to the object, and you shall find a way around or through it. If nothing within you stays rigid, outward things will disclose themselves. Empty your mind, be formless. Shapeless, like water.”

For all its cliche - it’s a quote I’ve seen a gazillion times, worked into B-roll kung fu movies, plastered on corny motivational posters - it is not wanting for wisdom, and I value Lee’s sentiment. But through the first week of having a shelter dog, I’m experiencing a massive gap between knowing the principle and living it.

I wish I could say this essay is a redemption story, about hearing Bruce Lee in my head, fully owning my situation, making some excellent lemonade during the chaos, and figuring out how to adapt my systems for when life gets a bit hectic so I can share a “how-to” with my audience. I could have developed a minimalist routine, leaving protected space for my essentials (writing, training, deliverables) instead of letting them plummet to the cellar of my priority list so that I have a memory of how to get shit done when life gets less predictable.

But I didn’t do that. I punted my core responsibilities, letting myself wither into a stress response instead of working within my situation. More alarming still is that this slight disruptive blip wasn’t even that obscene. I welcomed a mostly chill shelter dog from the local rescue into my home and let it upend my routine.

I know life will, without a doubt, get more strained than this. If a loving dog, a few extra errands, and a few disrupted days left me questioning my entire ability to accomplish important tasks, what does that mean for me?

So what are my takeaways here, what can I share with a reader as a source of trusted wisdom? For one, it’s the unsettling admission that I knew exactly what I was supposed to do (Be Like Water), and I proceeded to do the opposite (Be Like Rock?). I just tried to force my routine on top of the added stressors. I watched myself approaching the situation incorrectly, and couldn’t bring myself to course correct. So my systems, it appears, are fragile. Going forward, my options are to either double down on my rigidity and force things, or I can learn to modify my routine to accommodate my essentials alongside any hiccups that might occur.

Second: these hiccups, I think, are worth it. Some of the endeavors I look forward to in life - having a dog, raising kids, business ventures with interesting people - will almost certainly not conform to my vision of structure. Lifestyle turbulence is a byproduct of meaningful pursuits. Next time, if I can keep my essentials intact instead of throwing the entire system to the wind at the first sign of disruption, I’ll consider that a victory.

Leave a comment

Share

A little over a year ago, I relocated my home base and many of my business ops to Reno/Tahoe, NV from Austin, TX. I didn’t do so because I soured on Austin. I love Austin, still consider it a home away from home, and I find excuses to go back regularly. I shipped off to Reno because I love the mountains and driving 25 minutes is easier than catching a flight to Colorado or Utah.

For a while, I second-guessed my decision to head for the hills. A large part of my professional life sits at the intersection of engineering, technical startups, and pairing investors with companies pursuing meaningful innovation. Austin is, by most measures, one of the most robust U.S. hubs for such an ecosystem. I felt like I had traded living in the land of opportunity for a city that, professionally speaking, felt like an entirely different planet. That’s no fault of Reno’s. It just didn’t feel like Austin.

And I complained about it. You can ask my fiancée. I was insufferable for a few months, thinking that I had pulled the e-brake on my own professional development for the sake of skiing after finishing my Ph.D. I walked around for the better part of a season of life complaining about skiing all winter instead of working with new people as if I was forced into it.

Eventually, I got over myself, noting the lunacy in my gripes.

Part of what pulled me out of my slump was stumbling upon the Sierra Angels. SA is Nevada’s primary angel network, founded in 1997, with members between Las Vegas, Reno/Tahoe, and very rarely the hundreds of miles of barren high desert in between. The organization is as old as I am, and older than most of the startups it supports by a long shot. I joined the group last summer, making me a relatively new member. At a pitch night last week, I got to talking with Jeff Saling, the current head of the organization.

I didn’t know it until talking to him, but Jeff had a similar trajectory to my own. He lived in Austin for a time, building and successfully exiting three companies while he was a Texan. He was intimately plugged into Capital Factory’s Austin location, building within and learning from one of the country’s pr

emier startup accelerators. Then he moved to Reno, and quickly noticed the same thing I did: the startup support infrastructure in Nevada just wasn’t the same as it was in Austin, where it existed at all.

The difference is what Jeff did with that information. I whined about it for a few months. Jeff took it upon himself to build StartupNV in 2017, a Nevada-based accelerator that has helped ~2,500 Nevadan early stage founders across four funds. This essay isn’t a plug for StartupNV or the Sierra Angels, but a commentary on agency. Jeff saw a gap between Austin and Nevada’s startup support culture and decided it was a problem worth solving rather than a grievance to bitch about, building from scratch the ecosystem he was lacking.

I walked away from my conversation with Jeff thinking about agency. Real agency, that is, not the motivational poster version. Agency as the willingness to look at circumstances, including the ones that I didn’t choose or might have chosen with limited foresight, and doing something about them instead of being their casualty.

This is part of what draws me to the technical startup ecosystem in the first place. If you strip away the hustle-culture theater, the GPT wrappers with billion dollar valuations, the founders who spend more time on commenting on LinkedIn than working on the best version of their idea…technical entrepreneurship and the systems that support it operate with a genuine “can-do” current. People who see a broken or missing thing and decide to fix it.

Some might call it naive optimism, but I think it’s more of a default assumption that all problems are solvable. I’m talking about real problem-solving, not manufactured problems that commoditize attention or feed rampant consumerism.

There’s a lot of noise right now about agentic AI, positive and negative. What it can do, who it will replace, what it will break. How it’s awesome for humanity and its future, how it’s an existential threat to humanity and its future. Some of the bedlam might be warranted. But I think the most important part of the buzzword is the descriptor itself, agentic. I think it’s curious that we’ve handed that title, of being agentic, over to software.

As humans, we are agents. The human brain, for all its flaws, remains one of the most intricate problem-solving instruments we know of. Nobody knows quite how deep its capacities go. It can hold contradictions, read a room, change its mind, care about outcomes and other beings. It can move to the mountains to get more ski days in and, eventually, stop complaining about doing so.

I’m not here to make an anti-AI argument, I’m instead hoping to make a pro-human one, championing agency and the zeal to alter circumstances. In all the conversations about the coming wave of agentic AI and what it will do to us and for us, it’s worth remembering that we don’t need to outsource our own agency, nor do we need permission to act on it. We can make better conditions instead of waiting for them to show up.

Jeff Saling didn’t wait for Nevada to become Austin, he took steps to fill a gap and give Nevada an ecosystem of its own. That’s the whole thing, really.

Leave a comment

Share Danny Nikolai

There’s something about reading a good book in front of a fireplace. There’s something even deeper about reading a good book in front of a fireplace when it’s 04:30 AM, black as pitch outside, and I’m the only one awake in the house. Something academic, romantic, and monastic that puts me into a headspace to think too deeply about the activity with which I’m engaged. I had just cracked open a new book from my queue the other day, Brad Stulberg’s The Way of Excellence, and it occurred to me that there’s a degree of opulence about being able to start a new book on a whim. It dawned on me that I can read whatever I want, a refreshing, unusual, and unlikely luxury.

I tend to work on a couple of different books at a given time. Right now that’s Stulberg’s latest, Wind, Sand and Stars by Antoine de Saint-Exupery, and Lightbringer, the last release of Pierce Brown’s Red Rising epic. That’s self development, a memoir from a French pilot, and a sci-fi novel all going at the same time. Three completely unrelated worlds, captured in written text, all sitting on my nightstand because I felt like reading them, independent of someone’s assignment or algorithm’s suggestion.

As much as I love reading, I think it is far too easy to take it for granted in the form that I currently understand it. If we zoom out and consider the entirety of human history, there’s a blip on that timeline where written language shows up, probably somewhere in Mesopotamia and the Fertile Crescent around 3000 B.C. For those who want to dive deep into the origins of writing, I’d direct you towards Jared Diamond’s Guns, Germs, and Steel for a nice executive summary of the topic. 3000 B.C. was a long time ago. But on the scale of 300,000 years of homo sapiens, 5,000 years of written language is not all that long in our species’s history. And, as with most things, written language was not conceived with the agenda to someday allow me to write this essay using my MacBook Pro and Google Docs. It began as a record-keeping system for grain ledgers, not inherently designed to tell stories and convey deep ideas.

It’s easy to retroactively look at history and find inevitability in the way things have shaken out. But we didn’t have to have sci-fi epics, biographies, fairy smut, Substack, or any of it. If written language stayed limited to its original purpose, we might only have mile-long CVS receipts and nothing more to read about. The fact that it evolved into something that allows me to take in information about the dinosaurs, the intersection of the Kabbalah and quantum mechanics, and the history of the Irish Republican Army in a single lifetime, let alone a single month, was anything but guaranteed.

Yes, access to reading enables access to objective, empirically verifiable information. But it also provides access to ideas that the reader is free to interpret by their own accord. This makes reading indicative of the freedom infrastructure exemplified by classical liberalism and its champions. As a U.S. citizen, I can read satire, critiques, or political smear pieces from across the spectrum. I effectively have access to any thinker, any philosophy, across the entire duration of recorded human thought. The fact that I can explore as wide a range of ideas as I desire and figure out what resonates to me, not what someone tells me I have to believe, is no small thing.

This classically liberal infrastructure was also by no means inevitable. Societies of history that have held book burnings, book bans, and had strict censorship legislation understand this better than most. Some societies retain this stringency today. Dictatorships, oligarchies, and militant theocracies are quick to control information flow and limit the written word that can be consumed by a population. By recognizing the power of access to ideas, they are quick to shut that access down. Societies that consider themselves bastions of free thought, on the other hand, should keep the keys to reading widely readily available.

In 2026, the act of reading books remains an unmediated input in a world of algorithms. The global average of time spent on social media per capita is 2.5 hours per day at the time of writing. For those indulging in such short-form fare as their main source of information flow, nearly everything they see is algorithmically curated, suggested, and thrust upon them. Their intake, for at least 2.5 hours per day, is subject to the whims of computer code catered to show them advertisements, sponsored content, and recommended posts that leave the user with next to zero predictive capacity of what they might scroll upon next. In other words, information diets are being decided for people, an erosion to the very “freedom of intake” described above.

Books don’t have that problem. Sure, I might get recommended books on Goodreads and can appreciate receiving reading lists in my email inbox. But ultimately, I can still walk into a bookstore or library and choose, not having to worry about content being thrown in my face.

What’s more, books are long-form almost by definition. Authors have time to flesh out their ideas, form a bond with the reader who has chosen to pick up their work, and have space and time to develop their ideas on the order of many pages. Writers attempt to deliver thoughtful ideas in exchange for a reader’s agreement to settle in and hear them out.

This is becoming a rare form of socio-intellectual contract between reader and writer in a time when information diets are mostly the result of an algorithm-to-consumer model, where beliefs and desires are decided for consumers in short-form snippets of advertisements and echo chamber realities. Books demand something from you that algorithmically-catered content doesn’t, asking you to focus on just one world long enough for the author’s words to sink in, remedying the tendency to context switch between app windows and browser tabs. This, too, is well worth protecting and engaging with.

Freedom to read anything does not come free of charge, however. Reading’s socio-intellectual contract includes being at liberty to explore the myriad of humanity’s ideas in exchange for the responsibility to filter, to discern useful information and ideas from less-useful information and ideas.

Discernment. It’s almost a buzzword now, a skill that matters now more than ever and one that seems increasingly reserved for those who can exercise patience and plot development in their thinking patterns. Anyone can write a book, and there are certainly some bad ones out there. As a reader, one must be able to spot the bad apples, but hold faith in the whole bunch. You get to do the picking, but you also have to do the filtering. This is not an easy skill, but it is a cornerstone of free society. Freedom to read widely is inseparable from necessity of judgement.

Leave a comment

Share

What is the point at which engineering and philosophy become overlapping fields?

It’s a question I’ve found interesting recently, and the evolving stage of current affairs makes it progressively more necessary. I won’t be using this platform to recap the latest news cycles. Instead, I’d like to riff on some recent events as a pinpoint example, opening up a conversation and starting a series of essays exploring the overlap mentioned above.

This past weekend was a monumental one in the saga of what we call artificial intelligence. Most everyone is aware of the recent outbreak of conflict between the U.S. and Iran, and most everyone is aware of the recent fallout between the Pentagon and Anthropic. For those who aren’t: Dario Amodei and Anthropic ended up in the Pentagon’s bad graces, barred from all official use for refusing to let their models be used in the context of mass public surveillance and any sort of autonomous system war field projectile strike. Sam Altman and OpenAI saw the gap, and jumped in to secure that work for their own models.

Both chess moves were met with public applause and public scrutiny. Some people I know championed Anthropic’s ability to stick to their value structure in the face of immense pressure from one of the largest hegemonic institutions on planet Earth, posting screenshots of their cancelled ChatGPT subscriptions and new Claude installations. Others backed OpenAI, citing the imperative for evolving frontier technologies to keep pace with changing national security concerns. Then the latest round of conflict in the Middle East escalated, and the rubber of AI-use-in-wartime hit the road.

I’m not here to discuss who is in the right between Anthropic and OpenAI. My gut tells me that Anthropic made respectable decisions, as I find myself increasingly wary of digital panopticons and surveillance states. My bigger concern, however, is that there exists a growing competitive market landscape for autonomous lethal decision making and public wiretapping.

The concept and associated concerns aren’t exactly new. This flavor of concern came up 30 years ago, when General Atomics came out with the MQ-1 Predator drone. Instead of Air Force fighter pilots being up close and personal to conflict zones, operators were able to pilot strikes in from thousands of miles away with an Xbox controller. Questions of desensitization, accuracy, and ease-of-war were all asked. Psychological and moral distancing of combat have been baked into warfare for some time now. The questions surrounding whether or not LLM’s should be involved in decisions and analysis of killstrikes and surveillance seems to be the next step towards the logical extreme of this same debate. The same vector, just further up the technological innovation curve.

If we think about this from a market-incentive standpoint, these innovations probably seek to answer the question, “How can we make wartime easier for humans?” Fair enough. From an objective first principles standpoint, I don’t know if I can definitely say this is a bad thing. Considering the work of E.O. Wilson and Richard Wrangham, it’s fair to suggest that at least for the foreseeable future, warfare will continue to be a part of the human experience, as ghastly as it is. So shouldn’t we make it as non-detrimental as possible to the least amount of people, and leverage LLM reasoning technology to get there?

This is a version of the trolley problem on steroids. Is it more morally correct to leave a human in the loop to have the final say in pulling the trigger on the kill strike, or leave that decision to what should be, in theory, a neutral and objective judge? Are we more interested in potentially sparing some thousand-ought operators from a lifetime of PTSD by outsourcing the decision making to machines, or more interested in maintaining the humanistic empathy element to potentially spare the negative impacts on some thousand-ought civilians? If wartime operators use these LLM’s, how subject are they to skewing their answers and limiting their neutrality on the basis of a context window and a reinforcement learning architecture that incentivizes user confirmation bias? Do humans operate with confirmation bias skew anyways? Is an algorithm more likely to make a trustworthy decision than a stressed, sleep-deprived, combat-warn operator?

I don’t know the answer to these questions. I’m sure folks smarter than me and with more domain expertise have wondered the same things. This strain of questioning isn’t just limited to LLMs in wartime, and is more of a panindustrial consideration. The issue of whether an engineer at Waymo trains their vehicles to plow over a single pedestrian that steps out in front of it or swerve to potentially maim the passengers and nearby café patrons is a softball example, and we already know that some of Silicon Valley’s smartest minds engineered social media to optimize for engagement and screen time over mental health and sanity. With that said, I’d like to look under the hood of these types of questions with relation to culture and time.

For the vast majority of human history, asking hard questions about existence was a serious profession, and those who practiced it sat at influential seats in societies. Priests, philosophers, shamans, theologians, artists. The specific figures changed across cultures and centuries, but deep inquiry about the ramifications of human action was central to the ways these cultures operated. From ancient cosmology to the Roman Empire through the Renaissance and into the Enlightenment, polymathic “thinkers” were integral to operations and examining those operations. Philosophy permeated culture. Subsistence farmers may not have been reading Aquinas directly, but Aquinas shaped the world those farmers lived in.

Fast forward a century or so from the Enlightenment. The cultural infrastructure of metaphysical analysis, the idea that people should spend ample time asking the deepest questions, changed alongside the advent of the Industrial Revolution and accelerated with the post-WWII consumerist economy. As more and more doohickeys got made, the economies around these doohickeys gave people shinier things to think about. A few decades later came the internet, the worst of the internet and the attention economy, which actively monetized the atrophying of deep thought. If we plot “technological innovation vs. time,” we know that it follows a hockey stick, exponential increase. That curve might have a mirror-image inverse, a collapsing curve of philosophical curiosity and fluency, declining faster and faster at a time when we are building some of the most consequential technologies in human history.

I don’t know what it might mean to reach the convergence point of those two curves, if we haven’t passed it already. Powerful material innovation is a good thing in many ways. But I fear that the innovation is accelerating at a faster clip than reflective examination can keep up. Recent events certainly make for a good example, so I’ll bring it back to the issue with which I started this essay. As a species, I believe we can no longer pretend that hard sciences are a separate discipline from the “softer” fields of ethics and philosophy.

OpenAI is paying big bucks to young, sharp, early-career engineers who can design and train their models and algorithms, the very same models and algorithms now being sold for-profit to guide decision making on killstrikes and mass surveillance. The competitive market for the tech and the skills that build it are rich and fruitful. To me, this begs the question: How are the trainers trained?

Are the engineers that work on models that could eliminate human decision making for launching lethal projectiles well-versed in Thomas Nagel’s “What is it like to be?” frameworks the same way they know linear regression, random forests, and U-net architectures? Are those studying interpretability at the for-profit AI companies fluent in Hannah Arendt’s models for liberal societies the same way they are with token retrieval mechanisms and K-values?

By most Computer Science, Data Engineering, and Artificial Intelligence curricula, they are probably not. Professional ethicists exist, but it seems to be a relatively small field, more strongly represented in the field of genomics but generally more interested in making sure there is demographic equity in research considerations. If my suggestion that we are near a critical convergence of philosophy and engineering is correct, these two formerly different disciplines may now be inextricably linked. Engineers can no longer play the role of indifferent builder, and philosophers can no longer play armchair quarterback.

As an engineer myself, I think we may be due to develop a new field entirely. Call it “Scientethics.” An equal-parts dedicated study of the physics behind our innovations and the second-, third-, fourth-, etc. order philosophical ramifications of what we build. I was in academia as a student for nine years and currently serve on advisory boards for colleges of engineering, and I can tell you firsthand that this concept doesn’t seem to be a key issue in anyone’s “strategic plan for the university.” This essay is the first in a series I plan to dedicate to the subject.

As a society of builders and doers, we should probably all start asking: How are the trainers trained?

Leave a comment

Share

We bought a Snickers bar from one of these small vendor stalls that we all shared while looking out over the Wall’s many watch towers and undulations through the hills. I didn’t think anything of it at first, but eventually a high degree of metacognition set in: We just bought a Snickers Bar on the Great Wall of China, one of the seven wonders of the ancient world. I was eating a hyperengineered dessert at the very spot where some Ming Dynasty soldier and Mongol warlord may have grappled to the death 500 years ago. To me, that thought was trivial, ironic, absurd, and ultimately indicative of an overarching trend I noticed during my Chinese travels.

Over the span of two weeks, I saw the depths of China’s storied past and the intensity and innovation of its present. Firsthand experiences of landmarks (both antique and contemporary) that I’d only ever seen in travel vlogs and Wikipedia was astounding. But for all the beauty, grace, and tradition of China’s monuments and natural spaces, there was a nagging feeling that something was just a little bit off.

That feeling, I think, was a hyperfixation on modernization, a forced fit of novelties into historical landmarks. The Snicker’s bar was a microcosm of a larger pattern. A robot barista was making cappuccinos at the Summer Palace in Beijing. I bypassed the KFC outside of Zhangjaijie’s East Gate, but almost stopped in the Starbucks for a bubble tea and was told that there’s a Burger King near the top of the staircase at Heaven’s Gate. I bought some grapes from the 7-Eleven underneath the Guanyin statue at Mount Xiqiao.

My gut check to buying a Snickers bar on the Great Wall was something along the lines of “This is weird, this is a little bit tacky.” When I describe the plopping of the new (fast food and robots) into the old (ancient temples and national forests) to my friends back stateside in the U.S., the reaction is more or less the same as mine: slight unease. It wasn’t wrong per se, it was just a tad jarring, like two incongruent frequencies playing at the same time.

It’s not that commercialization of historical sites is unique to China. I think it’s the degree of integration that makes the difference. I have a hard time picturing KFC at the base of Half Dome with escalators scattered throughout the Yosemite Valley. I recognize that China thinks on a fundamentally different time scale than the U.S. A country measuring its history on the order of millennia might have a different relationship to what can get deprioritized during a 50-year sprint. But the question still stuck with me: I don’t think much of this would fly at home, so why is it so commonplace in China?

Over the last few decades, China has made a beeline to get to the future first. The intensive modernization and cross-pollination of novelty and antiquity is almost certainly intentional. Fifty years ago, much of the country was subject to deep poverty, existing as subsistence farmers on low-yield agricultural land. With the advent of Dengism and Deng Xiaoping as the “Architect of Modern China,” rapid industrialization and manufacturing prowess became the pinnacle state imperative above all else, a sentiment that continues today. Industrialization-at-all-costs. This imperative has worked. By every economic measure, it has worked spectacularly. Many millions were pulled out of poverty and integrated into the industrial workforce, global supply chains now revolve around China, and the country now boasts multiple cities with tens of millions of people and some of the most impressive skylines in the world.

My Snickers bar on the Great Wall got me thinking that this multi-decade sprint into the future left some residue behind. The escalators and AI photo booths tucked into the stunning mountainsides of national forests weren’t random, but a byproduct of society moving at a speed where certain questions might not get asked. China did a conventionally excellent job “catching up” to modernity. They placed a heavy bet on hyperindustrialization, and it paid big dividends. But I got the impression that some historical and natural character may have been misplaced in the process. Fast forward to today, and China is still moving, and moving quickly. In the game of geopolitical chess, the U.S. and the West are still moving too.

What is our current version of that bet, when the East and West are now playing a constant game of catch-up, technological one-upmanship, and are making “XYZ-imperative-at-all-costs” decisions every day? The whole world seems to be sprinting towards something, with the looming risk that the “adversary will get there first.” It could be AI dominance, quantum and compute supremacy, interplanetary colonization and the monopolization of the space economy.

The first-order tradeoffs of any great industrial sprint imperative are visible, debated, and documented, including elements like natural resources, labor, and capital. What I’m curious about are the second-order ones that sneak up on us because they don’t even have a name yet, let alone a constituency or lobby group. In the 1980s, China wasn’t having a conversation about what hyper-industrialization might do to the cultural character of Zhangjiajie National Forest. That conversation only came later, after the fact, when the escalators were already bolted on to the mountainsides. I don’t know what today’s equivalent of that is, which unrecognized second-order effects might come out of today’s technological arms race, but I’m willing to bet someone visiting a historical landmark in 2065 might have a better clue.

Leave a comment

Share

By and large, it seems like many people have stopped thinking critically.

Discernment has almost become an afterthought, an inconvenience, something that takes too much energy and mental bandwidth. So, as is the case with other energy drains, people have become perfectly content to offload it. I witness this phenomenon across my professional networks and in my closer friend and familial relationships. The Western, Internet-aged ecosystem has normalized outsourcing critical thinking to podcasts, bots in comment sections on social media, talking heads on news networks, so-called leaders crowing down from ivory towers, short-form sound bites, and AI-generated slop. Our attention and opinions are being increasingly shaped by the highest (loudest?) bidders, and the trajectory isn’t changing.

I turned 18 in 2015, and much of my adult life has been shaped by global cultural oddities. Trump 1.0, COVID, Trump 2.0, the algorithmic radicalization of basically every platform, and a general sense that the cultural center cannot hold. I’m not complaining, just zooming out and noting that this is the water I’ve been swimming in. Many other people my age have been too.

I’ve traveled to countries and cultures in the east and west, shared meals with locals in the depths of jungles and on mountain tops, all while experiencing these “unprecedented historical moments” through my own lens. I’ve gone deep into academia and have built an entrepreneurial career. My growing aggregate of experiences, in its own way, has shown me that true intelligence is the ability to hold multiple perspectives at once. That is the basis for critical thinking. The art of nuance seems to be disappearing, and that disturbs me. I’ve gathered a perspective shaped by a mix of experiences, and I want to do something with it.

As attention fragmentation and divisive outrage proliferate, I find myself craving longer-form thinking. I don’t believe I’m alone in this. But I’m not here to teach, sell you my e-book, build a brand, or inspire anyone. I’m just here to develop my own practice in resisting and remedying what I see in the world around me. This is a long-form outlet to distill my own thoughts from interests and experiences I’m exploring, be that philosophy, technological innovation, Renaissance architecture, fitness, travel stories, political theory, etc. Some intersections and throughlines might pop up, or not. I’m writing because the alternative is outsourcing my critical thinking, and I’m not interested in that.

This is my attempt to make a public field journal of my lived experience. The goal is to avoid philosophizing from an ivory tower and stick to exploring what’s on my radar in a given week. Some deep dives might be deeper than others. Some entries might be more topical than others, but these won’t be response pieces to every current event that captures the world’s attention. These will certainly not be AI-generated bullet point lists of emojis, em-dashes, and clichés about what I’m “quietly building” masquerading as original thought leadership. I have no interest in becoming another voice barking opinions into the void.

I don’t have a niche or a target audience. I just have experiences and a perspective, and I’d like to use writing to investigate how they interact. I’m writing as a sort of “ode to polymathy,” encouraging myself to investigate anything and everything that’s on my mind rather than having to cater it to a predefined audience. I’ll be working things out on the fly, probably generating more questions than answers. But this Substack is just my public journal, a way for me to share my musings with anyone who’s interested (not that they asked).

Leave a comment

Share