Rizwan Virk

Bostrom's trilemma: one real world vs. a billion simulated ones

Nick Bostrom's simulation argument offers three possibilities: no civilization can build ancestor simulations, those that can choose not to, or we are almost certainly in one. If one simulation is possible, billions are, and base-rate odds say you're far likelier to be in a simulated world than the single physical one.

“And the third option, it seems like a jump there, but basically what he's saying is that if they can create one simulation, they can create multiple simulations and they can create a billion simulations. So there's a billion simulated worlds. There's only one physical world. Statistically speaking, which one are you more likely to be in if, and this is the point, the leap that I think most people have a tough time making.”

Steal thisWhen judging 'is this the real one?', count how many copies could exist: if one is possible, base rates favor being a copy.

Two flavors of simulation theory: NPC vs. RPG

Rizwan Virk splits simulation theory into an NPC version, where everyone is an AI character running on some computer, and an RPG version, where a player outside the game inhabits a character. He maps the RPG version onto Eastern and Abrahamic ideas of the soul, illusion (Maya), and a universal 'river of forgetfulness' at birth.

“And so there's actually 2 versions of simulation theory. Uh, it's the NPC version or flavor of simulation theory and the RPG version. So RPG stands for role-playing game. Now, in the NPC version, we're all just AI characters inside, uh, a simulation that's running on a type of computer. In the RPG version, the simulation is also running on computation, but there's a player of the game who exists outside of the game.”

AI agent towns threw birthday parties, ran for mayor, invented religion

Virk cites the Stanford/Google 'Smallville' experiment, where about 100 AI agents left alone began self-organizing (one planned a birthday party, one ran for mayor), and a later 1,000-agent Minecraft simulation where some agents developed their own religion.

“And they let these agents kind of go and watched what they did. And they started to have these self-organizing activities. Like one was organizing a birthday party for another agent. One was running for mayor. Now that was a small number. Then there was another group that had 1,000 agents running in Minecraft, you know, and, and some of them had their own religion that they came up with. And you started to see these emergent behaviors.”

Forget P(doom) — worry about P(hack) from moderately smart AI

Virk argues the bigger near-term risk isn't superintelligence wiping out humanity but moderately intelligent agents that misinterpret directives and, Terminator-like, keep pushing through security holes until the mission is done. He calls the probability of AI hacking systems and causing chaos 'P-hack'.

“I would say we're far away from P-doom. What worries me is not so much that super intelligent AI will try to wipe us out. The problem is moderately intelligent AI, given directives, will go around and misinterpret some of those directives.”

A 10,000-agent swarm brute-forced the Navier-Stokes millennium problem

Virk points to reported progress on the Navier-Stokes fluid-dynamics millennium problem, emphasizing the method: 10,000 non-superintelligent agents grinding on one approach by brute force. To him it shows how swarms of ordinary agents can find their way around obstacles, which is why he rates the hacking risk high.

“The other big AI news was about solving one of the millennium problems, right? You know, uh, which was some fluid dynamics problem called the Navier-Stokes problem. But the specifics of the problem aren't as interesting to me as how they solved it. And they solved it by having 10,000 agents. That were all working on this one particular method through brute force.”

Most people run as NPCs 90% of the time, like an LLM on its training

Virk argues the NPC and RPG views blend: most people react on autopilot to their politics, religion, and past experiences the way an LLM responds from its training data. Only occasionally do mystics and yogis step out and ask what came before.

“This is why I say that the two aren't always necessarily mutually exclusive, uh, is that you can have combinations of the two where most people are acting like NPCs 90% of the time. We're just reacting to our training to our, whether it's our political party, our religion, uh, you know, the experiences that we've had in the same way that an LLM is trained on a certain set of data and then it's reacting to that.”

A VR ping-pong game fooled his body and sparked a decade of research

In 2016, just after selling his last game company, Virk tried a VR table-tennis demo whose physics were so convincing he tried to lean on a table that wasn't there and nearly fell. That moment led him to map out 10 stages to the 'simulation point' and, eventually, his book.

“I started playing this virtual reality ping pong game, and the physics engine was so good that it really felt like I was hitting a real ball. So much so that my body forgot for an instant that I was playing a virtual game of table tennis. And I tried to put the paddle on the table and I tried to lean against the table. And of course, there, there was no table.”

Quantum superposition looks like a game rendering the world on demand

Explaining the observer effect through Schrödinger's cat, Virk notes quantum mechanics says a particle stays in superposition until measured. To him that mirrors how video games only render the part of the world a player is looking at, to save compute.

“But what quantum mechanics is telling us in a weird sort of way is that the cat is in a state of superposition. It's both alive and dead. Until that information is needed, right? So it gets measured and then it's needed. And so to me, that sounded a lot like rendering the world on demand in video games.”

Autobiography of a Yogi — Steve Jobs' favorite book

Virk recommends Paramahansa Yogananda's Autobiography of a Yogi, noting it was Steve Jobs' favorite book and was handed out at his funeral. Yogananda explained the world as a film in which we are both actors and audience, a precursor to today's video-game metaphor.

“There was a, one of the first Indian yogis to come over to the U.S. and live here and make an impact was a guy named Swami Yogananda. He wrote a book called Autobiography of a Yogi, which was Steve Jobs', you know, favorite book.”

Unfalsifiable isn't unknowable: science once called meteorites peasant myths

Even if the simulation hypothesis can't be falsified, Virk argues you can still look for evidence. He points to meteorites: scientists dismissed reports of rocks falling from the sky because their model said it was impossible, until the model turned out to be wrong.

“And that, you know, what happened was science used to believe that all the stories that people were telling about these rocks falling from the sky were just a bunch of ignorant peasants who who, you know, were just making things up. Turns out they weren't. But the scientific model of the time said there are no rocks in the sky, therefore they can't be falling, therefore you guys are mistaken, hallucinating, uh, you know, lying, whatever else you might be doing. But turns out we had the wrong model of reality.”

A simulation answer to the Fermi Paradox: distant planets render on demand

One explanation Virk raises for the Fermi Paradox (where are all the aliens?) is that we're in a simulation and the far-off universe is only rendered as our telescopes observe it, which he links to ever-growing estimates of the universe's age and galaxy count. He says he doesn't fully subscribe to it.

“one mode of thought is an explanation for the Fermi Paradox is we are actually in a simulation and we think there are all these planets out there. But there really aren't because those planets get rendered for us, uh, as we observe them and the telescopes look out. And you notice how they keep increasing the distance, um, the age of the universe.”

The UFO that sliced through trees without cutting them

Ufologist Jacques Vallée told Virk about a Pacific Northwest landing where witnesses saw a craft descend at 45 degrees through tall trees that were left completely intact. Virk reads it as a rendering procedure, like objects passing through not-yet-solid walls in a video game.

“He said, look, you guys said it came down at a 45-degree angle. That would mean it would have to cut through the trees, but the trees aren't cut. I mean, the trees are still there. They said, yeah, it was so weird. It was like it was slicing through the trees, but not really cutting them. And then it came and, and landed. And I'm looking at that and said, you know, that looks like a rendering procedure.”

He dreamed of a vanished competitor the morning IBM killed his product

At Virk's first startup, an enterprise tool connecting Microsoft and IBM products, he dreamed one night of a competitor he hadn't heard from in 18 months. That morning IBM called to say it had quietly acquired that competitor, Edge Research, and was announcing a product that day that would kill his.

“He goes, oh, do you remember that competitor of yours, Edge Research? We acquired them about a year and a half ago, and they've been working on this product in stealth for the last year and a half. And today we're announcing it. And I'm just like, okay, wait, I had the dream in the morning. It wasn't the other way around.”

Treat failed startups as quests: you keep the XP and unlock the next one

With 90-95% of startups failing financially, Virk reframes those failures as video-game quests: even an unfinished quest earns experience, and one quest often unlocks the next. The experiences and relationships along the way are the point of the journey.

“And yet 99, let's say 95% of startups fail. So, but yet there are other things you've learned along the way. There are other experiences you have. with other people along the way. And those experiences are the fabric of the journey itself. And so if you view those as quests, what happens in a video game if you don't achieve a quest? Well, I mean, you still have gained some valuable experience, and you go off and you do the next quest. And sometimes one quest ends up unlocking another quest.”

Steal thisLog what each failed venture taught you and which next opportunity it unlocked, instead of scoring it only on financial outcome.