John Carmack: The Engineer Behind Doom and Quake on VR, AI, SpaceX, Neuralink, and the Cost of Ambition

John Carmack helped transform video games with Wolfenstein 3D, Doom, and Quake. He then pursued increasingly difficult challenges in virtual reality, artificial intelligence, high-performance computing, and experimental rocketry. Along the way, he developed an unusual understanding of what makes ambitious technology succeed—and what causes it to fail.

In a wide-ranging conversation with Joe Rogan on August 28, 2019, Carmack examined several decisions he would make differently. He questioned the development philosophy that once defined id Software, explained why his rocket company fell short of reaching space, challenged Jeff Bezos's cautious approach to aerospace engineering, and defended his decision to remain a programmer rather than become a conventional technology executive.

He also described a future in which virtual reality replaces ordinary screens, brain implants restore capabilities to disabled people, and artificial intelligence potentially reproduces human reasoning.

Some of his predictions have since encountered measurable reality. Neuralink has advanced into human trials, Blue Origin has reached orbit, and standalone VR hardware has become substantially more capable. Other ambitions remain unresolved.

The interview, published as The Joe Rogan Experience #1342, offers an unusually detailed account of how Carmack approaches engineering, commercial risk, personal ambition, and the limits of technological progress.

The Six-Year Game That Changed Carmack's Philosophy

Why Carmack Abandoned “When It's Done”

During id Software's early years, Carmack became associated with a philosophy of refusing release deadlines. When people asked when the next game would appear, the answer was simple: it would arrive when the developers considered it finished.

That independence mattered. The company could concentrate on technical quality without a publisher forcing it to meet quarterly financial targets.

Experience eventually changed his mind.

Carmack distinguished between delaying a game for several months to improve it and extending development for years. Longer delays introduce consequences that additional engineering cannot necessarily overcome: hardware advances, consumer expectations change, competitors release new products, and the original commercial opportunity may disappear.

“Time has a physicality that you may not appreciate.” — John Carmack

The clearest example was Rage, id Software's technically ambitious first-person shooter released in 2011.

The project consumed approximately six years. Its developers pursued advanced rendering technology, presented impressive demonstrations, and continued improving the game beyond its original schedule.

But the market did not remain stationary.

By release, some technical decisions designed around earlier hardware generations were becoming less appropriate. Meanwhile, Call of Duty and Battlefield had developed into dominant competitors.

Carmack believed Rage should have shipped approximately two years earlier, even if achieving that deadline required abandoning some technological ambitions.

This was not simply hindsight invented for the 2019 conversation. In a 2011 GameSpot interview conducted before Rage's release, Carmack already described the long development period as undesirable. He observed that the iPhone had not existed when work on the game began—an illustration of how much the technology market could change during one project.

The commercial lesson was more complicated than releasing unfinished products. Carmack was arguing that finishing every technically desirable feature could be less valuable than delivering a coherent product while its underlying assumptions remained relevant.

Why Doom Was Better Timed Than Quake

Carmack identified similar problems in the development of Quake.

Although Quake took approximately two years rather than six, its production was difficult because id Software attempted to introduce too many fundamental innovations simultaneously.

The project combined true 3D environments, sophisticated graphics technology, multiplayer networking, internet game servers, and new systems for player modifications.

Carmack later wondered whether the company should have separated these advances into multiple releases. One possibility would have been an improved Doom-style game containing part of the technology, followed by a more ambitious title that developed the remaining ideas.

He still valued Quake's achievements and the community it created. His criticism concerned the development strategy, not the finished game's historical importance.

Doom, by comparison, appeared to him almost perfectly timed: sufficiently advanced to transform expectations while still achievable within the opportunity available.

Why Large Budgets Can Limit Creativity

Carmack recognized that contemporary games often surpass older releases in visual detail, production quality, and technical sophistication.

Yet increasingly large budgets also create incentives to avoid unusual ideas.

A game costing hundreds of millions of dollars needs a large audience to justify its development. Publishers therefore have reasons to favor familiar mechanics, established franchises, extensive market research, and broadly appealing designs.

Older games could sometimes be stranger precisely because less money was at stake.

Quake's unforgiving multiplayer systems, for example, were not entirely intentional. Those systems nevertheless created a distinctive competitive experience that became central to its identity.

Carmack did not argue that old games were inherently superior. He cautioned against nostalgia while recognizing that increased production resources can coexist with reduced willingness to experiment.

The connection between these observations is important: engineering ambition, creative ambition, and commercial timing are separate problems. Excellence in one does not guarantee success in the others.

Armadillo Aerospace: Why Carmack's Rocket Company Stopped Short of Space

Building Rockets by Experimenting Until Something Broke

Carmack's interest in ambitious engineering eventually took him far beyond software.

He founded Armadillo Aerospace, a small experimental rocket company, to investigate technologies that might make private spaceflight cheaper and more practical.

The company developed vehicles capable of vertical takeoff, hovering, sideways movement, and controlled landing. One demonstration involved a rocket-powered vehicle weighing roughly 2,000 pounds performing those maneuvers at an Oklahoma airfield.

The work required Carmack to learn new disciplines, including control theory, embedded software, propulsion, and flight stabilization.

Unlike video game development, aerospace experiments could destroy physical equipment worth substantial amounts of money.

Carmack recalled that Armadillo built and lost approximately two rocket vehicles annually over eight or nine years. Later prototypes could cost more than $100,000 apiece.

That expenditure was not necessarily evidence of incompetence. The company was exploring systems that could not be perfected through computer simulations alone. Real flights exposed interactions among propulsion, hardware reliability, control software, and physical operating conditions.

There were genuine achievements.

In October 2008, Armadillo won $350,000 in NASA's Northrop Grumman Lunar Lander Challenge. Its vehicle rose 50 meters, moved horizontally to a landing platform 100 meters away, remained airborne for at least 90 seconds, landed, and repeated the maneuver.

In 2009, the company also earned a $500,000 second-place award in the challenge's more demanding second level.

These results are documented in the U.S. Department of Commerce's official prize record. They establish that Armadillo had developed substantial reusable-rocket capabilities, even though it never achieved Carmack's larger commercial objective.

The company also performed assignments for NASA, the U.S. Air Force, and the Rocket Racing League, including work involving rocket-powered aircraft. Carmack recalled that Armadillo achieved an operating profit during one year.

The Business Trap of Government Research Contracts

Armadillo eventually encountered a difficult strategic choice.

The company could continue accepting aerospace research contracts. Government agencies had an interest in maintaining small engineering firms capable of developing useful technologies.

Those contracts could pay salaries, support further experiments, and keep the business operating.

But Carmack worried that such work could consume the company's attention without bringing it closer to building commercially useful reusable spacecraft.

He described a pattern in which companies accept enough contracts to survive, then spend their resources completing each assignment and securing the next one.

He believed the available funding could maintain an engineering organization without supplying enough capital or urgency to achieve its original ambitions.

This was Carmack's assessment of the incentives, not proof that government-funded research is inherently unproductive. Armadillo itself benefited from publicly funded prizes and contracts, and many aerospace technologies require long-term research before becoming commercially viable.

The practical question was whether Armadillo should become a sustainable aerospace contractor or continue risking its resources on a more ambitious vehicle.

Carmack chose the latter.

The 100-Kilometer Goal and the Decision to Stop

Carmack committed additional money to one final development push.

The objective was a reusable suborbital rocket capable of reaching approximately 100 kilometers above Earth—the commonly recognized Kármán line.

He characterized the decision as:

“Suborbital or bust.” — John Carmack

The company did not reach its target.

Carmack recalled achieving approximately 92 kilometers, producing photographs of the black sky, sunlight, and the vehicle's return toward Earth. That altitude is his figure from the interview; the independent records examined do not conclusively establish the same maximum flight altitude.

He considered the results technically meaningful but insufficient to justify continuing under the same arrangement.

Rather than indefinitely financing the project or shifting its purpose toward contract-funded survival, he sold most of Armadillo's assets to team members who obtained additional funding.

Former Armadillo personnel subsequently developed related technology through Exos Aerospace, continuing part of the original company's engineering work.

For Carmack, the ending was painful but instructive. A project could produce important achievements, employ capable engineers, and still fail to reach the objective that justified its existence.

Why Carmack Would Never Run Another Rocket Company Part-Time

One of his strongest conclusions concerned divided attention.

During Armadillo's development, Carmack was still devoting approximately 40 hours weekly to id Software and another 20 hours to aerospace engineering.

That amounted to a considerable personal workload. Nevertheless, he concluded that 20 hours was insufficient to lead such a demanding technical venture.

Rocket engineering required rapid learning, careful decisions, sustained experimentation, and close attention to each failure. Splitting his schedule between unrelated projects reduced his ability to understand and improve the aerospace systems.

Carmack was therefore reluctant to restart the company unless he could commit himself full-time.

His conclusion connects directly to his criticism of excessively long game-development schedules: serious projects need not only enough money and technical talent but also concentrated execution.

SpaceX, Blue Origin, and the Economics of Moving Fast

Carmack's Early Encounters With Elon Musk

Armadillo Aerospace began developing rockets around the same period as SpaceX and Blue Origin.

Carmack recalled Elon Musk and an early SpaceX colleague visiting Armadillo in Dallas to discuss experimental rocket technology. Years later, Musk invited Carmack to participate in a SpaceX engineering meeting.

Carmack admired the company's accomplishments and believed its progress in reusable rockets represented a major technological achievement.

He also considered the broader commercial space industry remarkable. Orbital launch had historically been dominated by national governments and a small number of major contractors.

SpaceX and emerging companies such as Rocket Lab demonstrated that privately funded engineering organizations could enter a field once considered largely inaccessible.

The results were measurable before the interview. In September 2008, SpaceX's Falcon 1 became the first privately developed liquid-fueled rocket to reach Earth orbit. NASA records that milestone in its SpaceX mission history.

The distinction mattered to Carmack because it demonstrated that private spaceflight was more than a business proposal. Real launch vehicles were achieving orbital missions.

Why Carmack Criticized Blue Origin's Cautious Culture

Carmack regarded Blue Origin's early development philosophy as excessively conservative.

Jeff Bezos's company placed great emphasis on deliberate, methodical progress. Carmack associated that approach with the tortoise imagery used in Blue Origin's identity.

He believed substantial personal wealth could become a disadvantage when it removed pressure to make difficult decisions quickly.

His comparison was SpaceX.

Musk had committed a substantial part of his fortune to his companies. Carmack recalled that SpaceX faced a critical financial situation after three unsuccessful Falcon 1 launches.

A fourth failure could have seriously threatened its survival.

Carmack believed such pressure encouraged rapid experimentation and unusually decisive engineering.

He also defended SpaceX against criticism following launch explosions and failed landings. In experimental aerospace development, destructive tests can produce information unavailable through cautious analysis alone.

That does not make every failure desirable. The relevant distinction is whether a failure produces useful knowledge that improves the next attempt.

Carmack considered SpaceX particularly effective at applying that principle.

What Happened to Blue Origin After the Interview?

Carmack's 2019 criticism must be separated from Blue Origin's subsequent achievements.

He already acknowledged that the company had begun moving beyond its original slower development pattern toward larger orbital systems.

Those plans later produced substantial results.

Blue Origin's New Glenn rocket reached orbit on January 16, 2025. Its first-stage booster was lost during the recovery attempt, but the orbital objective succeeded.

On November 13, 2025, the second New Glenn mission launched NASA's ESCAPADE spacecraft and successfully landed the reusable booster on an Atlantic Ocean platform.

The company conducted a third New Glenn mission on April 19, 2026.

These milestones are documented in Blue Origin's orbital mission records.

They demonstrate that Blue Origin developed genuine orbital launch capability after the interview. They do not, by themselves, resolve whether the company could have reached the same position more quickly through a different development philosophy.

There was another consequential change.

On January 30, 2026, Blue Origin announced a suspension of New Shepard flights for at least two years, redirecting resources toward human lunar exploration.

New Shepard had completed 38 flights and carried 98 people above the Kármán line, according to the company.

The decision partly resembled the strategic shift Carmack had anticipated in 2019: Blue Origin increasingly pursued large aerospace objectives and government-supported lunar programs rather than concentrating exclusively on short suborbital tourist flights.

That is a documented strategic development, not evidence that Carmack's broader criticism of the company was necessarily correct.

Why Carmack Did Not Simply Join SpaceX

Carmack believed he possessed skills useful to SpaceX, and his technical conversations with Musk reinforced his interest in its work.

However, he was not primarily interested in becoming another engineer inside an already successful organization.

He preferred the possibility of creating something different.

He mentioned unconventional rocket concepts, including potentially cheaper but riskier propulsion arrangements involving monopropellants—chemicals capable of producing thrust through decomposition without requiring separate conventional fuel-and-oxidizer feeds.

Such ideas could simplify some hardware arrangements while introducing substantial technical and safety challenges.

Carmack did not present them as proven commercial solutions. They represented engineering possibilities he might explore if he returned to aerospace full-time.

The Surprising Economics of Space Tourism

Why a $200,000 Ticket Looked Better Than a $100,000 Ticket

Carmack's experience with space tourism produced a different kind of lesson: customers sometimes value an unusual experience more highly than entrepreneurs expect.

Early suborbital operators considered approximately $100,000 a plausible ticket price.

The proposed service was not orbital travel. Passengers would rise toward the edge of space, experience several minutes of weightlessness, see Earth from extraordinary altitude, and return without completing an orbit.

Carmack contrasted this with parabolic aircraft flights, where participants generally experience weightlessness for only about 20 seconds at a time.

When Richard Branson's Virgin Galactic announced seats at approximately $200,000, Carmack was surprised that customers continued signing up.

For a company struggling with expensive equipment and development costs, the difference between those prices could substantially change its financial prospects.

At $100,000 per passenger, a mission might barely cover its relevant expenses. At $200,000, the same mission could potentially produce enough contribution margin to support more meaningful commercial operations.

Those figures were Carmack's illustrative assessment, not independently verified profit calculations. Neither price alone establishes profitability without knowing actual flight frequency, aircraft maintenance, insurance, staffing, financing, and development costs.

Higher Prices Did Not Solve the Development Problem

Carmack also worried that Virgin Galactic had spent so long developing its vehicles that tourism revenue would struggle to recover its accumulated investment.

Subsequent events illustrate both sides of that concern.

Demand for expensive seats remained real. But achieving reliable, frequent commercial operations proved much harder than selling reservations.

Virgin Galactic retired its VSS Unity spacecraft from commercial service in June 2024 to concentrate on developing its Delta-class vehicles.

By August 12, 2026, the company had increased the price of a new spaceflight reservation to $750,000. Its second-quarter 2026 financial update reported that the latest ticket allocation was oversubscribed, while its first new commercial flight had been postponed until February 2027.

That combination is revealing.

Customers demonstrated willingness to pay almost four times the approximately $200,000 price Carmack had discussed. Yet the company still faced delays, significant capital requirements, and the challenge of establishing a financially sustainable operating schedule.

High demand and high prices improve a business's revenue potential. They cannot eliminate the engineering and capital costs of getting a spacecraft into regular service.

The broader story reinforces Carmack's central concern about development time: in aerospace, as in video games, achieving the desired technical result too late can profoundly change the economics.

How Carmack Gave Doom and Quake a Life Beyond Their Developers

Wolfenstein Players Changed What Id Software Built

Carmack's commitment to releasing game source code began with the unexpected creativity of players.

Before Doom, Wolfenstein 3D had not been designed for extensive modification. Its data was compressed using specialized techniques intended partly to accommodate the limited storage of floppy disks.

That did not discourage enthusiasts.

Players reverse-engineered file formats, disassembled code, built editing software, and developed modifications. Some replaced the game's enemies with entirely different characters, including Barney the Dinosaur in place of Hitler.

Carmack regarded their efforts as evidence of unusually strong engagement.

People enjoyed the game enough to keep creating new experiences after finishing its original content.

Rather than treating these modifications as a threat, he wanted to make them easier.

Doom therefore incorporated organized modification support through WAD and PWAD files. These formats allowed game data, levels, sounds, and graphics to be replaced or extended without rewriting the complete engine.

Id Software documented important aspects of the system and released supporting tools.

Some original development technology remained difficult for ordinary players to use. Doom's level editor, for example, had been developed on expensive NeXT workstations, so the community needed more accessible PC-compatible alternatives.

Quake Made Entirely New Game Modes Possible

With Quake, Carmack extended the principle beyond changing a game's appearance.

The game included QuakeC, a programming language used to implement gameplay behavior.

This allowed developers outside id Software to modify rules and mechanics instead of merely replacing existing levels and assets.

The resulting community helped popularize modes such as Capture the Flag and Team Fortress.

These creations were not simply decorations added to a finished product. They changed how people played, competed, and organized themselves around the game.

Carmack eventually concluded that even sophisticated modification tools placed unnecessary limits on what enthusiasts could accomplish.

Complete source-code access would allow programmers to make deeper changes, adapt engines to new platforms, and support unusual hardware.

Why Id Software Initially Resisted Open Source

The proposal encountered internal opposition.

Businesspeople, artists, and designers worried that giving away technology developed at considerable expense might benefit competitors.

Carmack developed a compromise: release the previous generation of software only after the company had shipped a newer technological platform.

This reduced the likelihood that competitors would gain immediate access to id Software's most commercially valuable engineering.

The company released older Wolfenstein technology after Doom, Doom's source after Quake, and Quake's code after subsequent generations.

Some colleagues regarded the policy as Carmack's personal initiative rather than a generally accepted company strategy.

He nonetheless used his influence to continue it.

Carmack recalled particular satisfaction when longtime colleague Kevin Cloud later acknowledged that releasing the code had been the right decision.

Why the Source Code Still Matters

Carmack believed that software released in this way could remain useful long after its original commercial lifespan.

Programmers ported Doom to devices far removed from the PCs for which it was designed, adapting it to unusual processors, displays, and embedded systems.

The available code also allowed enthusiasts to integrate experimental equipment without convincing Carmack or another company executive to approve each idea.

The historical consequence is directly observable. Id Software's public source-code repositories contain Doom, Quake, Quake II, Quake III Arena, Doom 3, and other releases.

These repositories demonstrate that the preservation of Carmack's engineering work no longer depends entirely on the continued commercial availability of the original games.

There is an important qualification: releasing an engine's code does not automatically place every commercial game's artwork, music, or other assets under the same license.

Nevertheless, Carmack's approach gave older games a technical future beyond their initial sales and provided material for programmers to study, maintain, and extend.

He expressed disappointment that game companies widely adopted modification support without equally embracing full source-code releases.

How John Carmack Helped Bring Virtual Reality Back

The $150,000 Systems That Still Failed to Impress Him

Carmack's interest in virtual reality began long before Oculus.

During the development of Wolfenstein, Doom, and Quake, companies approached id Software about licensing its games for VR headsets.

Carmack repeatedly rejected the proposals because the equipment did not meet his technical expectations.

The problem was not simply that developers lacked interesting software. Headsets suffered from inadequate resolution, poor tracking, excessive latency, and other limitations that prevented the experience from feeling convincing.

Some specialized systems cost tens of thousands of dollars, and Carmack recalled equipment priced around $150,000 that still failed to deliver what he considered satisfactory VR.

After completing Rage, he returned to the subject around 2011, expecting improvements in computing technology to have solved many of the problems he remembered from the 1990s.

He was disappointed to discover how much remained unfinished.

Instead of waiting for the industry, Carmack began experimenting with hardware and software himself.

Palmer Luckey, Oculus, and the E3 Demonstration

Carmack's work brought him into contact with Palmer Luckey, who was developing an inexpensive wide-field-of-view headset.

The collaboration became an important turning point in consumer VR.

Carmack adapted Doom 3 to run on Luckey's prototype, correcting optical distortion through software and demonstrating how a comparatively inexpensive display could create an immersive experience.

At E3 in June 2012, he presented the modified game to journalists.

Contemporary accounts document the technical details. WIRED's history of the Oculus Rift describes how Carmack modified the prototype and developed software to compensate for distortion introduced by its lenses.

The demonstration helped generate industry attention for Luckey's work and the emerging Oculus project.

Carmack later became Oculus's chief technology officer, bringing experience in graphics optimization and high-performance software to the development of consumer headsets.

The challenge was no longer simply proving that VR could work. It was making the experience affordable, convenient, and reliable enough for ordinary consumers.

Why Carmack Preferred the Harder Mobile Problem

High-end gaming computers offered vastly more computing power than the processors available inside lightweight standalone headsets.

Carmack estimated that a powerful gaming PC at the time possessed approximately 50 times the computing capability of the mobile VR hardware under discussion.

That disparity made sophisticated graphics much easier on systems connected to external computers.

Nevertheless, he deliberately concentrated on standalone devices.

He enjoyed difficult optimization problems, but his preference also reflected a commercial judgment.

A headset that requires a powerful computer, cables, and external tracking equipment is less convenient than one containing everything necessary to operate.

By removing those dependencies, Oculus could make VR easier to adopt and use.

That became the guiding idea behind Oculus Go and Quest.

Oculus Quest and Carmack's Vision of Replacing Every Screen

Why People Watched Netflix Instead of Playing Games

When Rogan examined the original Oculus Quest, he was struck by how much technology fitted into a relatively compact headset.

Unlike earlier PC-connected VR systems, Quest contained its own processor, displays, batteries, and tracking equipment.

Carmack estimated that it could operate for more than three hours when watching video, although demanding games might reduce battery life to approximately two hours.

One of his most interesting observations concerned the popularity of ordinary entertainment.

Developers had expected VR to be defined primarily by spectacular games and experiences impossible through conventional screens.

Yet watching Netflix, YouTube, or Amazon Prime had become an important use of the earlier Oculus Go.

Carmack explained why that made sense.

Someone who already owned an excellent home theater had relatively little to gain from watching a virtual television.

But someone living in a tiny Tokyo apartment could put on a headset and experience a spacious ski lodge containing an enormous screen.

The simulation did not increase the size of the apartment. It changed the person's visual and auditory experience of occupying it.

The Headset as a General-Purpose Computer

Carmack's long-term vision went beyond gaming.

He believed virtual displays might eventually replace conventional televisions, tablets, laptops, computer monitors, and even some uses of smartphones.

A person would no longer need separate physical screens of different sizes. Virtual windows could be created, resized, and arranged as required.

The same device might support work, entertainment, communication, and experiences that cannot be reproduced on an ordinary flat display.

This ambition introduced several demanding requirements. A headset intended for occasional gaming can tolerate limitations that become unacceptable during an eight-hour working day.

Display resolution must support comfortable reading. The headset must remain physically comfortable. Batteries must operate long enough, interfaces must be practical, and prolonged use should not produce significant discomfort.

Carmack understood that the hardware available in 2019 was not yet capable of replacing everyone's screens.

What Later Quest Hardware Actually Achieved

Subsequent releases demonstrate considerable progress toward the capabilities Carmack described.

Meta's Quest 3 introduced improved displays and full-color passthrough video, allowing users to see their physical surroundings while interacting with digital windows.

In September 2024, Meta introduced the Quest 3S at a starting price of $299.99, with the same principal mixed-reality computing platform as Quest 3.

In its September 2024 product announcement, Meta reported that more than 60% of active Quest 3 users were using mixed-reality multitasking with panel applications.

That is a company-reported usage figure, not independent evidence that most consumers have replaced their conventional screens.

Still, it directly supports part of Carmack's thesis: standalone headsets were evolving toward ordinary computing activities rather than remaining exclusively game devices.

The difference between demonstrating useful virtual screens and permanently replacing physical monitors remains substantial.

Could VR Make Luxury Experiences Affordable?

Carmack also considered VR a way to distribute experiences normally restricted by wealth and geography.

Not everyone can own an expansive home, visit remote destinations, build a private cinema, or travel across the world to explore museums.

But the visual and auditory components of many such experiences can be simulated.

A person might explore an impressive virtual building, enjoy a simulated private island, or meet distant relatives in a shared room.

Carmack expected the price of VR equipment to fall over time. He compared the approximately $400 Quest with the $200 Oculus Go and noted how inexpensive smartphones had become in markets such as India, mentioning devices priced around $25.

He believed similar reductions in computing costs could eventually make immersive technology widely accessible.

This was part of a larger argument about digital life.

Carmack rejected the assumption that an artificial experience is automatically worthless because it is not physically real.

He compared VR with technologies such as air conditioning, which people use to make their environments more comfortable than they would otherwise be.

That analogy has limits. Simulating a beautiful destination does not reproduce its climate, physical materials, food, or human interactions in full.

Carmack acknowledged those limitations. He was interested in how much meaningful value could be delivered through experiences dominated by sight and sound.

Remote Relationships and Virtual Museums

Carmack imagined relatives living far apart meeting in shared virtual rooms, communicating naturally, and watching entertainment together.

Similarly, people might visit detailed virtual museums without the expense and inconvenience of travelling, waiting in queues, or navigating crowds.

He regarded these experiences as particularly useful to people whose circumstances restrict their physical opportunities.

Virtual reality would not eliminate the value of travelling or meeting someone in person. It could offer a more accessible alternative in circumstances where the physical experience would otherwise be unavailable.

The underlying economic principle was significant: software can reproduce and distribute many visual experiences without reproducing the land, buildings, equipment, and transportation that make their physical equivalents expensive.

Why Perfect Virtual Reality Requires More Than Better Graphics

What Theme Parks Can Still Do Better

Rogan described Disney's Avatar attraction, Flight of Passage, which combines immersive imagery with physical movement and environmental effects.

Carmack connected the example to attractions such as The VOID, where participants enter virtual environments containing real physical structures and props.

Such experiences can simulate heat near virtual lava, move a participant's body, or provide an actual object corresponding to a weapon in the game.

Carmack believed consumer VR headsets could increasingly reproduce the visual and auditory elements of these attractions.

That left specialized physical sensations as the principal justification for expensive location-based experiences.

A commercial theme park could spend millions constructing hardware and environments that ordinary consumers would not want to install at home.

The commercial question became whether the additional sensations justified the expense of visiting a specialized attraction.

Why Smell Is Harder Than Color

Displays reproduce a broad range of visible colors by combining red, green, and blue light.

Smell has no equivalent simple set of three primary substances.

Human olfaction responds to numerous chemical molecules and their combinations. Faithfully reproducing the odors in a natural environment could require a much larger selection of compounds than a consumer headset can conveniently store.

Carmack described experimental scent devices that attached underneath headsets and released small quantities of selected fragrances through cartridges.

He and Rogan also discussed commercial sensory masks offering different scent effects.

The limitation was not necessarily the quantity of fragrance released. A small amount near the user's nose might suffice.

The problem was reproducing a sufficiently varied collection of recognizable smells.

Carmack suggested that a few convincing jungle fragrances might satisfy an ordinary city resident, even though someone deeply familiar with an actual tropical environment would notice many missing odors.

The observation illustrates a recurring principle in his work: a convincing experience does not always require complete physical accuracy. It requires enough relevant information for the experience to feel credible.

Beat Saber and the Importance of Matching Physical Movement

Carmack particularly admired Beat Saber because the player's real actions closely corresponded to virtual events.

Swinging a controller produced a matching lightsaber-like movement inside the game. Small vibrations reinforced the sensation of striking objects.

This was more convincing than pretending to hold a virtual object through movements that did not correspond naturally to its physical behavior.

Rogan described the difference between manipulating conventional controllers and holding actual Stormtrooper blasters at The VOID.

Carmack saw the same problem in haptic technology—devices that provide touch sensations.

During the Doom and Quake years, an entrepreneur had demonstrated a leather jacket containing approximately eight or nine mechanical impact devices. The jacket could strike the wearer when their game character was shot.

Carmack doubted that most consumers would enjoy such painful realism.

Nevertheless, he recognized that stronger physical feedback could improve experiences where ordinary vibration motors failed to produce convincing contact.

He considered mechanisms capable of storing and releasing mechanical energy, including spring-driven systems that could deliver sharper impacts.

Why Virtual Experiences Can Become Real Memories

Carmack described playing Minecraft in VR with his family and later remembering aspects of the experience as if he had physically occupied its environments.

He sometimes recalled being inside a cavern or observing objects from a particular location rather than remembering the experience as a conventional video game.

That subjective difference helped explain why immersive technology interested him.

A conventional game is typically remembered as something viewed on a screen. VR can create a stronger sense of spatial presence.

That does not mean virtual events become indistinguishable from reality in every circumstance. It suggests that immersive experiences can be encoded and recalled differently from ordinary screen-based entertainment.

Motion Sickness: The Problem That Quake Could Not Simply Overcome

Why Traditional Shooters Can Make VR Players Sick

Quake encouraged rapid movement, sharp turns, jumping, and explosive changes in direction.

Those characteristics made it exciting as a conventional shooter but extremely difficult to reproduce comfortably in VR.

The problem involves a mismatch between vision and the vestibular system, which helps the body detect movement and balance.

If the display shows rapid acceleration while the inner ear detects that the user is standing still, the conflicting signals can produce nausea and disorientation.

Rocket jumping—using explosions to propel a player through the air—illustrated the difficulty particularly well.

It might feel spectacular at first, but repeated simulated flight and changes in direction could quickly become intolerable.

Carmack preferred games where virtual movement closely followed actual physical movement, although he acknowledged that some enthusiasts willingly accepted discomfort to enjoy unusual experiences.

Why Cockpits and Real-World References Help

Carmack explained that conventional monitors can produce simulator sickness too.

However, a small monitor may occupy only a fraction of the viewer's actual field of vision. The surrounding room provides stable visual information indicating that the person is not physically moving.

A headset removes much of that external reference.

Some VR applications therefore maintain a stable cockpit or surrounding structure within the virtual scene.

The stationary reference can make simulated movement more tolerable.

Carmack discussed Omega Agent, a jetpack game featuring airborne movement and curved flight paths, as an example of the trade-off between exciting locomotion and comfort.

The experience could be compelling even while producing significant motion-sickness risks.

Guardian and the Risk of Physical Injury

Unlike seated VR, room-scale experiences allow people to walk, crouch, turn, and swing their arms while unable to see their actual surroundings.

That introduced a different safety problem.

A player could strike a table, collide with furniture, or unknowingly approach a staircase.

Oculus addressed this through Guardian, a boundary system established during headset setup.

Users marked a safe physical area, and the headset displayed a warning when they approached its edge.

Carmack described experiments with predictive warnings that accounted for movement speed. A hand moving rapidly toward a boundary could trigger a warning before reaching it.

More aggressive warnings, however, interrupted the experience. Some players and developers preferred weakening or disabling the system to avoid distractions.

Carmack acknowledged injuries among Oculus staff who had struck physical objects during testing.

Guardian therefore represented an engineering compromise between safety and uninterrupted immersion.

Cameras, Passthrough, and Automatic Room Recognition

The original Quest used four outward-facing cameras to identify visual features in its surroundings and calculate headset movement.

The same cameras detected infrared lights on handheld controllers.

This inside-out tracking removed the need for external tracking stations.

Quest also offered black-and-white passthrough video, allowing users to see nearby objects without removing the headset.

Carmack considered this important because people could become disoriented during long sessions, sometimes discovering they were facing an unexpected direction when they removed the device.

Tracking outdoors introduced additional complications. Bright sunlight could overwhelm the infrared signals from controllers, while overcast conditions made large outdoor spaces more practical.

He described enthusiasts experimenting with tennis courts and fields to obtain much greater physical movement space.

Facebook Reality Labs had already demonstrated systems capable of reconstructing complicated rooms in three dimensions, identifying individual objects and potential obstacles.

However, Carmack explained that detailed environmental reconstruction required sensors and processing resources too expensive for the original Quest.

His preferred future involved headsets that automatically understood their surroundings and recognized hazards without requiring users to draw boundaries manually.

Augmented Reality: Useful Glasses or Expensive Demonstrations?

Why Carmack Preferred VR Over Contemporary AR

Carmack distinguished fully immersive virtual reality from augmented reality, where digital information is placed within a view of the real world.

He was particularly skeptical about AR demonstrations that transformed rooms into spectacular virtual scenes without providing compelling everyday functions.

Simulated flooding, elaborate visual effects, and decorative overlays might demonstrate impressive engineering. They did not necessarily establish why people would want to wear the equipment regularly.

Carmack cited Pokémon Go as evidence that compelling applications could succeed without relying on sophisticated AR imagery.

He preferred fully immersive VR because its benefits were more immediately recognizable to him.

His position did not mean he considered AR impossible. He questioned whether the available products justified their compromises.

Magic Leap, HoloLens, and the Value of Specialized Work

Carmack criticized Magic Leap's promotional demonstrations for encouraging expectations that its hardware could not reproduce.

He was more receptive to Microsoft's HoloLens in professional settings.

An expensive headset with a limited field of view might make little sense for ordinary entertainment. But it could be commercially useful for industrial training or highly specialized work, such as guiding a technician servicing a jet engine.

The value of helping a professional perform an expensive or difficult task could justify hardware costing thousands of dollars.

That did not establish a mass-market consumer product.

Carmack believed everyday AR required something closer to ordinary sunglasses than bulky goggles.

He wanted technology comfortable enough to wear for approximately 16–18 hours, rather than a device used for occasional short sessions.

The Case for Glasses That Simply Let People Zoom In

One of Carmack's preferred early applications was straightforward: glasses that could instantly magnify distant objects.

A person might look toward something far away and trigger enhanced vision through an effortless gesture or command.

He considered that capability potentially valuable even without elaborate holograms.

Camera-based passthrough systems were easier to improve through better sensors, displays, and computing. Truly transparent glasses remained more difficult because of their optical requirements.

Carmack also imagined more intuitive controls through eye tracking, wrist movements, and eventually neural interfaces.

The distinction was between equipment that occasionally produces a spectacular effect and equipment that becomes useful enough to function as a natural extension of its owner.

Neuralink: From Brain Implants to Restoring Human Independence

What Carmack Learned During His Neuralink Visit

Shortly before Neuralink's public presentation in 2019, Carmack spent an entire day visiting the company and speaking with its engineers.

He came away impressed by the scope of its ambitions.

Earlier brain-computer interface research had demonstrated that neural activity could be used to control external devices. Neuralink wanted to increase the amount of information exchanged between the brain and computers while developing more practical implantation methods.

The company's proposed system involved thin, flexible electrodes, specialized electronics, and robotic surgery.

Carmack considered the approach significantly more ambitious than incremental improvements to conventional computer interfaces.

The historical technical record helps clarify its capabilities.

In a 2019 Neuralink white paper published in the Journal of Medical Internet Research, the company described experimental arrays containing up to 3,072 electrodes distributed across 96 flexible threads.

Its robotic system was designed to insert these threads with high positional precision, while custom electronics recorded neural signals.

The white paper was published on October 31, 2019, after Carmack's August interview. It provided technical details supporting the company's direction but did not demonstrate a fully developed human brain implant capable of all the functions discussed.

The reported experiments involved animal research and prototype technology.

Helping Disabled People Before Enhancing Healthy Brains

Carmack distinguished between near-term medical applications and more speculative human enhancement.

A person with severe paralysis may retain the ability to formulate movements without being able to transmit those intentions through damaged nerves and muscles.

A brain-computer interface could potentially decode the intended movement and translate it into commands for a computer or assistive device.

This might allow someone to operate a cursor, type, control equipment, or eventually interact with robotic limbs.

Carmack considered such applications more realistic than radically increasing the intelligence of healthy people.

His enthusiasm was tempered by technical concerns.

Implants must remain functional inside living tissue. Biological reactions, material degradation, electrode movement, and other complications can reduce their performance.

A laboratory demonstration lasting weeks or months is not equivalent to a reliable medical device expected to operate for years.

What Neuralink Has Demonstrated Since 2019

The company began implanting its brain-computer interface in human participants in January 2024.

Its initial PRIME study focused on people with paralysis using the N1 implant to control external devices.

According to Neuralink's clinical trial information, the technology remains investigational.

By January 28, 2026, Neuralink reported 21 participants enrolled in its trials worldwide.

In its two-year progress report, the company described participants controlling computers, creating artwork, playing games, studying, and operating assistive robotic equipment.

One participant used a robotic arm to perform activities including feeding himself.

The report also identified continuing technical limitations. Neuralink described variations in signal quality and ongoing efforts to improve electrode retention and performance.

The company reported stronger signals in 18 of the 20 participants who followed its first recipient after changes to its system.

These are company-reported results, not evidence that every implant performs equally well or that the system has completed the clinical testing necessary for general medical use.

Nevertheless, the progression from laboratory prototypes in 2019 to human participants performing useful tasks in 2026 gives substantial support to Carmack's belief that medical applications could emerge before radical cognitive enhancement.

Why More Neural Connections Would Not Necessarily Defeat AI

Elon Musk had also promoted brain-computer interfaces as a possible way for humans to remain relevant in a world of increasingly capable artificial intelligence.

Carmack was skeptical.

He understood the concern that AI might eventually surpass human intellectual abilities. But he doubted that connecting more electronic channels to a biological brain would necessarily make it competitive with powerful artificial systems.

The distinction was between improving the way humans interact with computers and expanding the brain's underlying capacity to reason.

A person might gain extraordinarily fast access to information or control multiple devices through neural signals. That would not automatically allow the person to think with the computational resources available to a large artificial system.

Carmack regarded human enhancement as a much more difficult proposition than restoring lost movement or communication.

Why Carmack Wanted Programmers to Experiment Directly

Carmack believed development might accelerate if programmers could personally experience and modify neural-interface software.

Someone operating an experimental interface could adjust its behavior, immediately experience the consequences, and repeat the process.

He compared this with the slower process of running occasional laboratory sessions with patients.

However, he acknowledged that brain implantation introduces medical risks far beyond ordinary software experiments.

He recalled a researcher who arranged to have electrodes implanted in his own brain outside conventional medical oversight and subsequently experienced complications involving speech and facial movement.

Carmack also discussed enthusiasts implanting simpler electronic devices, such as an access chip used to unlock a Tesla.

These examples illustrated growing interest in human-computer integration, but they were not equivalent in medical complexity.

He proposed using ordinary mobile computing architecture wherever possible. Neural signals might be handled using interfaces comparable to camera inputs, while processors developed by companies such as Qualcomm could perform some of the required computation.

The objective was to reuse mature technology rather than develop every component from the beginning.

Ultimately, Carmack imagined interfaces capable of supporting much richer interaction than moving a cursor.

Human movements involve many simultaneous degrees of freedom. A skilled musician, for example, controls multiple fingers and subtle movements while receiving continuous sensory feedback.

Brain interfaces capable of similarly complex interaction—and eventually sending useful sensations back into the nervous system—would represent a much greater achievement.

Those capabilities remained speculative in the context of the interview.

Artificial General Intelligence: Carmack's Ten-Year Prediction

Why Carmack Believed Human Intelligence Could Be Simulated

Carmack approached artificial intelligence from a strongly materialist position.

He regarded the human brain as a physical system operating according to discoverable processes.

If the relevant mechanisms of intelligence could be understood, he saw no fundamental reason they could not be reproduced computationally.

He acknowledged that many researchers expected artificial general intelligence, or AGI, to remain decades away.

AGI generally refers to artificial systems capable of performing a broad range of intellectual tasks rather than being limited to a narrow application. There is no universally accepted operational test establishing exactly when that threshold has been reached.

Carmack was more optimistic than many researchers.

During the August 2019 interview, he suggested that recognizable signs of AGI might appear within approximately a decade—around 2029.

He explicitly acknowledged his tendency to underestimate how long difficult projects take.

This was a forecast, not a claim that AGI already existed or that its arrival was inevitable by a particular date.

Why the Human Brain's Size Was Not a Definitive Barrier

Carmack referred to the human brain's roughly 85 billion neurons when considering the computational resources necessary for intelligence.

He questioned whether an artificial system needed to reproduce the brain's physical structure exactly.

Biological mechanisms evolved under constraints very different from those shaping electronic computers.

A computer can perform some functions far more efficiently through algorithms and specialized circuitry than through a literal reproduction of biological components.

Carmack therefore argued that directly comparing neuron counts with transistor counts could be misleading.

The key question was whether an artificial system could reproduce the relevant information-processing capabilities, not whether it contained the same number of physical components.

Why Algorithms and Experiments Might Matter More Than Hardware

Carmack suggested that major supercomputers might already possess enough raw computing power to support significant progress toward AGI.

He distinguished computing capacity from the ability to use it effectively.

Traditional software often relies on sequential operations that cannot all be performed simultaneously. An enormous supercomputer may therefore provide little advantage over a powerful desktop computer for certain single-threaded programs.

Modern machine learning is different.

Many AI workloads involve large quantities of matrix calculations that can be distributed across numerous processors.

GPUs and other parallel computing systems are particularly effective for those operations.

Carmack recalled that id Software once operated substantial computing equipment for building and processing complex game levels. Those workloads had required expensive calculations and considerable optimization.

He emphasized the importance of making experiments cheaper and easier to repeat.

A breakthrough might not come from one researcher developing a perfect theory. It could emerge after many researchers tested different algorithms, discovered failures, and gradually identified promising approaches.

The relationship to his rocket experience is clear. Faster learning often depends on reducing the cost of failed experiments.

Quantum Computing and the Security of Digital Information

Why Carmack Was Unsure About Quantum Computing's Benefits

Carmack was interested in quantum computing but acknowledged that he had not studied it as deeply as other technical subjects.

He had intended to work through quantum programming exercises and gain a stronger practical understanding.

He was therefore cautious about claims that quantum computing would automatically transform AI or ordinary computing.

Quantum computers exploit physical properties that can make certain specialized calculations more efficient.

That does not mean they are universally faster than conventional computers.

Carmack was particularly concerned about the possibility that powerful quantum systems might become excellent at breaking important forms of encryption without offering equally dramatic improvements to everyday software.

The Threat to Public-Key Cryptography

Many digital security systems depend on mathematical problems that conventional computers cannot solve efficiently at the sizes used in practical cryptographic systems.

A sufficiently capable fault-tolerant quantum computer could threaten several widely used public-key techniques.

These include methods used for establishing secure connections and verifying digital signatures.

That would not mean every encryption algorithm suddenly became useless. Different cryptographic systems rely on different mathematical assumptions.

Carmack discussed the possibility of widespread disruption to secure communication and authentication if vulnerable systems remained in use when powerful quantum computers became available.

There has since been an important defensive development.

In August 2024, the U.S. National Institute of Standards and Technology finalized three principal post-quantum cryptography standards:

• FIPS 203: ML-KEM, for establishing shared secret keys.
• FIPS 204: ML-DSA, for digital signatures.
• FIPS 205: SLH-DSA, another digital-signature standard.

NIST has urged organizations to begin transitioning vulnerable systems to quantum-resistant alternatives. Its post-quantum cryptography program documents the standards and migration guidance.

This development provides an important qualification to Carmack's 2019 discussion: quantum threats remain serious, but the technical response is not limited to abandoning modern encryption entirely.

Why One-Time Pads Remained Interesting

Carmack discussed one-time pads as a theoretically secure alternative.

A properly implemented one-time pad uses genuinely random secret key material at least as long as the message. That key material must be shared securely and must never be reused.

Under those conditions, the encryption provides information-theoretic security, meaning an attacker cannot uniquely recover the message from the ciphertext alone.

The drawback is practical.

Communicating parties need to obtain, store, protect, and distribute enormous quantities of random secret information.

Carmack speculated that increasingly inexpensive data storage might make very large one-time pads more manageable, perhaps even through implanted electronic storage.

But storing large amounts of key material does not solve the equally important problem of securely distributing and protecting it.

His proposal was a thought experiment about future hardware, not a demonstrated replacement for modern cryptographic infrastructure.

Why Carmack Believed the Easy Era of Faster Computers Was Ending

The Physical Limits of Semiconductor Manufacturing

For decades, programmers benefited from rapid improvements in computing performance.

Software that initially required expensive hardware could often run on inexpensive consumer devices several years later.

Carmack believed developers were approaching a period in which they could no longer depend on that trend continuing indefinitely.

Semiconductor components were becoming extraordinarily small.

At atomic scales, quantum effects such as electron tunneling can complicate the operation of electronic switches. Particles may cross barriers that would behave differently in classical descriptions of electronics.

When circuits contain enormous numbers of components, individually rare errors become engineering concerns.

Carmack noted that computing and communication systems already use error-detection and correction techniques to compensate for unreliable physical processes.

Such methods can extend reliability but also introduce additional complexity and overhead.

Why Advanced Chip Factories Cost Billions

Manufacturing sophisticated semiconductors required increasingly expensive facilities.

Carmack pointed to the concentration of advanced manufacturing among a small number of companies, including TSMC, Samsung, and Intel.

The cost of developing new fabrication processes and constructing advanced factories was reaching billions of dollars.

That concentration complicated the assumption that each new semiconductor generation would automatically provide enormous performance improvements at progressively lower cost.

Carmack considered alternatives such as carbon nanotubes and photonic computing interesting.

However, he did not regard them as guaranteed successors to conventional silicon technology.

The reliable response, in his view, was renewed attention to software efficiency.

Why Optimization Would Become More Valuable

Carmack had built his reputation on producing sophisticated graphics from limited hardware.

He considered that skill increasingly relevant as conventional semiconductor scaling approached physical and economic constraints.

Developers could not assume that an inefficient program would eventually become acceptable simply because future processors would be much faster.

Careful algorithms, memory management, parallel processing, and efficient software architecture would remain essential.

His focus on standalone VR followed directly from that belief.

A self-contained headset requires high graphical performance within strict limits on battery capacity, heat generation, memory, and processor power.

That makes efficient engineering fundamental to whether the product can exist in a useful form.

Why Carmack Thought 5G Was Being Oversold for VR

Carmack recognized the technical benefits of fifth-generation cellular networks.

Higher bandwidth allows more information to be transmitted, while lower latency reduces the time required for information to travel between devices and services.

But he questioned marketing that presented 5G as a revolutionary requirement for virtual reality.

Beat Saber illustrated the problem.

The game runs primarily on the headset's own hardware. Swinging a controller, striking objects, and calculating gameplay results do not require continuous communication with a remote server.

A faster cellular connection therefore does little to improve its central experience.

Carmack identified more plausible beneficiaries.

Stereoscopic 360-degree video requires substantial data transmission, especially at high resolutions.

Cloud gaming could also benefit from reduced latency when games run on remote computers and stream their rendered images to users.

Under suitable conditions, faster networks might reduce delays enough to make remote rendering more convincing.

Nevertheless, those applications depended on their entire technical systems, including servers, routing, processing, and local hardware—not merely the label attached to a wireless network.

Carmack also recognized the commercial incentives behind 5G promotion.

Telecommunications companies faced enormous expenses upgrading infrastructure and therefore had reasons to associate those investments with compelling new consumer experiences.

His criticism concerned exaggerated claims about immediate applications, not the usefulness of faster networks themselves.

Competitive Quake and the Birth of Professional Gaming

The Ferrari That Helped Define Early Esports

In 1997, Carmack provided an extraordinary prize for the Red Annihilation Quake tournament: his personal Ferrari 328 GTS.

The winner was Dennis “Thresh” Fong, already one of the most accomplished competitive Quake players.

Carmack believed the prize was so extravagant that it might remain unmatched for years.

He was surprised when other tournaments began offering substantial cash prizes soon afterward, including approximately $100,000 within the following year.

Fong's victory became an important moment in the history of professional gaming. Guinness World Records recognizes him as an early professional gaming pioneer and documents the Ferrari awarded at Red Annihilation.

Carmack and Rogan discussed how recordings of Fong's gameplay helped ordinary players understand the extraordinary gap between experienced amateurs and elite professionals.

Spectators could watch from Fong's perspective, study his positioning, and observe decisions that would otherwise be difficult to recognize.

The early competitive scene eventually grew into an industry featuring celebrity players, major sponsorships, substantial prizes, and enormous live audiences.

Carmack pointed to games such as Dota and League of Legends, especially their popularity in overseas markets, as evidence that competitive gaming had developed far beyond the scale he imagined during Quake's early years.

South Korea was particularly important to the growth of organized esports.

Carmack was impressed by major arena events and the international visibility of professional players.

Why Professional Quake Was So Difficult

Quake rewarded more than fast reactions and accurate shooting.

Its maps contained weapons, ammunition, armor, and health supplies that appeared at known locations and regenerated according to predictable schedules.

An expert player learned where the resources were located and when they would return.

That knowledge allowed a strong competitor to control the map.

A player could collect a powerful weapon, move toward an armor pickup just as it appeared, replenish health, and prevent an opponent from obtaining equivalent resources.

The result was a compounding advantage.

Carmack explained that a player only slightly better than an opponent could potentially win every engagement.

After a close victory, the stronger player might collect enough health and armor to recover before the next fight.

The weaker player, meanwhile, would return with fewer resources and face an opponent who had already recovered.

Matches could therefore produce extreme score differences, including results such as 20–1.

How Game Design Changes the Meaning of Skill

Carmack recognized that Quake's harshness was partly an unintended consequence of its mechanics.

A game without extensive opportunities to replenish health could produce more balanced encounters. An opponent who failed to win one fight might nevertheless inflict enough damage to affect the next.

Quake's resource system could instead allow the leading player to reset their condition repeatedly.

That created a demanding combination of strategy, memory, movement, timing, and aiming.

Rogan particularly enjoyed the challenge of mastering maps and denying opponents essential equipment.

Carmack understood the appeal, while also recognizing that such systems could discourage beginners.

He contrasted this with team-based games such as Overwatch.

A newcomer might initially contribute very little but still experience victory because stronger teammates compensated for their weaknesses.

Over time, that player could learn to become useful without enduring repeated overwhelming one-on-one defeats.

The comparison illustrates how game mechanics determine not only competitive difficulty but also whether new players enjoy participating.

Why Professionals Sometimes Made Quake Look Worse

Competitive players also manipulated graphics settings to improve performance.

Removing textures or reducing graphical complexity could increase frame rates and lower input latency, particularly on early graphics hardware.

Simplified environments also made moving opponents easier to identify.

A nearly textureless wall offered less visual distraction, allowing a competitor to recognize an enemy immediately.

Carmack understood the advantage but disliked seeing the work of artists and designers stripped away.

He recalled imposing minimum visual-quality requirements at early competitive events.

The objective was partly commercial: spectators encountering Quake for the first time should see an attractive game rather than an environment reduced to the minimum graphics necessary for competitive efficiency.

The tension between player performance and spectator presentation became one of the early challenges of professional gaming.

When Playing Games Becomes Physically Exhausting

Rogan described spending hours playing Quake, particularly during periods of intensive competition.

The experience could leave him physically tense, exhausted, and with an elevated heart rate.

He also described dreaming about Quake maps, movement, and incoming rockets.

Carmack recognized the phenomenon from his own programming experiences.

During periods of learning unfamiliar technologies, he sometimes worked around 13 hours a day and continued processing the problems in his dreams.

He interpreted these experiences as signs of deep mental absorption.

Why Carmack Enjoyed Social Media While Others Hated It

Carmack's view of social media differed sharply from the experiences of people who found platforms such as Twitter consistently upsetting.

He acknowledged problems involving anger, addictive use, confirmation bias, and unhealthy online behavior.

But his own feed exposed him to scientific discoveries, engineering accomplishments, interesting artwork, and people developing new products.

He found those connections inspiring.

Carmack also recognized that his experience was not representative of everyone.

He described himself as an introvert whose interests and social preferences differed from those of many people.

For someone with access to exciting activities, strong local communities, and abundant opportunities, spending more time online might provide limited additional value.

For a person living in an isolated town with few people sharing their interests, digital communities could be transformative.

Carmack believed critics sometimes underestimated those benefits because they judged digital life through their own circumstances.

He defended the broader value of connecting people internationally and believed Facebook executives, including Mark Zuckerberg, genuinely cared about that mission.

These were his personal judgments about the company's leadership and social impact, not independent findings establishing that its practices were always beneficial.

He also admitted that he was not particularly qualified to solve all the psychological and social problems associated with online platforms.

His argument was narrower: the harms of digital connection should not obscure the opportunities it creates for people whose offline circumstances are limited.

Why Carmack Chose Programming Over Becoming a Billionaire

The Work He Actually Wanted to Do

Carmack had spent much of his career leading technically ambitious projects.

He could have concentrated on managing larger organizations, directing engineering teams, and pursuing the executive career associated with exceptional financial rewards.

Instead, he deliberately preferred continuing to write software.

He estimated that only about half his time at Oculus or Facebook involved programming.

The remainder consisted of meetings, strategic discussions, technical persuasion, and attempts to influence organizational decisions.

Carmack did not consider himself particularly talented at conventional people management.

He believed he could demonstrate technical leadership and inspire other engineers through his work, but he was less comfortable evaluating individual motivations and directing employees.

His concern went beyond personal preference.

A programmer who stops working directly with technology can gradually lose familiarity with current tools, techniques, and engineering problems.

Carmack feared that moving permanently into management could eventually undermine the expertise that made his technical judgment valuable.

“I love building things. I like creating things and making forward progress.” — John Carmack

He accepted that this decision might prevent him from becoming a billionaire.

To him, continuing to enjoy challenging engineering work was worth more than maximizing his executive authority.

Zuckerberg, Musk, and the Cost of Leadership

Carmack contrasted his choice with Mark Zuckerberg's career.

He believed Zuckerberg had deliberately moved away from programming to develop the skills required to run Facebook.

Carmack respected that decision without wanting to follow the same path.

He also saw similar tensions in Elon Musk's responsibilities.

Musk remained involved in technical questions but had to coordinate multiple companies and manage enormous organizational demands.

Carmack recalled Musk speaking wistfully about his earlier programming days.

These comparisons helped explain why Carmack approached his own career differently.

He could recognize the potential influence of directing large teams while still preferring the satisfaction of personally understanding and building complex systems.

The 60-Hour Workweek

Carmack described approximately 60 hours of work per week as his preferred schedule.

Family responsibilities sometimes reduced that figure, but he admitted feeling insufficiently productive when his working time fell below 50 hours.

He rejected exaggerated claims that he regularly worked 18-hour days.

In his experience, effective computer work became substantially less productive after approximately 13 hours.

He also believed long days required variety.

Switching between different programming problems could help maintain attention, while struggling continuously with a single difficult issue produced diminishing returns.

Carmack considered approximately eight hours of sleep important for maintaining performance.

Sacrificing sleep might temporarily increase available working hours, but eventually it damaged concentration, learning, and productive output.

He was describing his personal preferences and working habits, not establishing a generally appropriate schedule for every employee.

Why He Occasionally Disappeared Into a Hotel

Carmack periodically arranged week-long programming retreats.

With his family's agreement, he would travel alone to a hotel, sometimes in another state, and remove as many distractions as possible.

He avoided unnecessary internet use and concentrated almost entirely on writing software and studying difficult subjects.

After approximately a day of adjustment, he found that he could return to long periods of highly productive technical work.

One retreat shortly before the interview focused on artificial intelligence.

Carmack particularly enjoyed learning technologies he did not yet understand.

He described the satisfaction of working through unfamiliar material until its underlying concepts became clear.

Those retreats also reassured him that his programming abilities had not disappeared despite spending more time on organizational responsibilities.

Long Hours, Crunch, and the Game Industry's Working Conditions

Carmack recognized that game development often offered less attractive employment conditions than mainstream technology companies.

Highly skilled programmers could earn more stable compensation working for organizations such as Google, Facebook, Apple, or Microsoft.

Game studios, by contrast, frequently depended on employees who were personally passionate about the products they created.

That passion could make people willing to accept lower pay, intense workloads, and uncertain job security.

Carmack compared the phenomenon with artistic and musical careers, where many talented people want to participate despite demanding economic conditions.

He believed some extraordinary games existed because developers had willingly devoted enormous amounts of time and attention to them.

However, the distinction between voluntary commitment and organizational pressure is important.

An independent founder choosing to work long hours has freedoms that a junior employee may not possess.

Carmack acknowledged that his own career had been unusual. He had not personally worked inside large development organizations such as Electronic Arts or Activision, and he lacked direct knowledge of every studio's employment practices.

He defended the right of individuals to pursue demanding work schedules while accepting that exploitation and workplace pressure were legitimate concerns.

Michael Abrash provided a useful counterexample.

Abrash, a highly respected engineer who worked closely with Carmack on Quake, maintained a more conventional family-oriented schedule.

His contributions remained valuable and widely respected.

Carmack therefore did not believe exceptional work necessarily required extreme hours.

He questioned whether a company would genuinely dismiss someone producing outstanding results simply because that employee worked fewer hours than colleagues, but acknowledged that he could not speak for every organization.

The broader distinction is between effort freely chosen for meaningful work and expectations imposed through unequal workplace power.

Judo, Brazilian Jiu-Jitsu, and Learning Through Physical Resistance

How Carmack Returned to Martial Arts

Carmack practiced wrestling and recreational judo while growing up in Missouri.

He considered himself competent rather than exceptional, sometimes finishing near second place in competitions.

His teenage participation also left him with knee injuries that affected his later physical training.

In his thirties, he decided to return to judo.

He joined Becerra Judo, where training under a former Cuban Olympian proved considerably more demanding than his earlier recreational experience.

Although Carmack retained some wrestling instincts, more experienced grapplers repeatedly caught him in submissions.

He particularly struggled against armbars.

Rather than accepting these weaknesses, he began studying the techniques and learning how to defend against them.

The Year of Private Lessons With Carlos Machado

Carmack's wife gave him a year of private lessons with Brazilian jiu-jitsu instructor Carlos Machado as a Christmas present.

The arrangement became an important part of his grappling development.

Training sessions took place at Carmack's Armadillo Aerospace facility, which had sufficient space for mats.

Carmack combined the private lessons with additional judo training and practice with other partners.

He became particularly effective at a straight armbar, a technique involving controlled extension of an opponent's elbow.

That improvement allowed him to surprise people at his regular judo club.

The experience reinforced an approach familiar from his engineering work: identify a specific weakness, study the relevant mechanics, practice deliberately, and test the improvement against real resistance.

What the Kodokan Taught Him About Expertise

During a visit to Tokyo's Kodokan, one of judo's most important institutions, Carmack trained with an older Japanese judoka.

The experience revealed another level of technical control.

Carmack recalled being almost completely immobilized before eventually being submitted.

He contrasted his own tense, forceful style with the relaxed movement of highly experienced practitioners such as Machado.

The strongest experts could appear to offer opportunities while subtly controlling balance, leverage, and positioning.

Carmack admired that economy of movement.

He also rejected the stereotype that an intellectually focused computer engineer should have little interest in combat sports.

When a young student expressed surprise that someone like him attended judo, Carmack defended the value of physical resistance.

Grappling exposes people to effort, failure, and the limitations of their own bodies in ways that purely intellectual exercises cannot reproduce.

How VR Could Change Martial Arts, Sports, and Workplace Training

Studying Techniques in Three Dimensions

Carmack's interest in judo connected naturally with his work in virtual reality.

Martial arts instruction often depends on understanding subtle relationships among body position, leverage, timing, and balance.

Ordinary video can make those relationships difficult to see.

A technique may last only a fraction of a second. Important movements can be obscured by the camera angle or the bodies of the participants.

Carmack imagined immersive stereoscopic video allowing students to examine techniques with a stronger sense of depth.

A student might pause an elite competitor's movement, advance it frame by frame, reverse the sequence, or inspect positioning from a different perspective where appropriate recording technology allowed it.

He considered judo particularly suitable because complex throws and transitions can happen too quickly for beginners to understand through ordinary playback.

VR Training Without Physical Contact

Carmack also considered immersive training useful for striking, weapons handling, distance recognition, and reaction exercises.

Some aspects of these activities depend more on movement and visual judgment than on direct physical contact.

Grappling remained harder to reproduce because realistic leverage and resistance require another body or specialized physical equipment.

Augmented reality might eventually provide a different solution.

A student training with a real partner could see visual guidance showing where an arm, leg, or body should move.

That would preserve actual physical resistance while adding digital instruction.

Workplace Training and American Football

Carmack identified Walmart as an example of a major company already using VR for employee training.

He also discussed American football teams using immersive technology to rehearse plays.

A player could study how a particular formation should develop from a realistic perspective, improving recognition before performing the movement on a field.

Carmack believed headset-based training could also improve attention by reducing opportunities to become distracted by unrelated screens and notifications.

These examples illustrate the difference between simulation as entertainment and simulation as preparation.

The value is not necessarily replacing physical practice. It is improving the learner's understanding before entering the real situation.

Beat Saber as Physical Exercise

Carmack regularly played Beat Saber and described finishing demanding sessions drenched in sweat.

The game allowed him to combine entertainment with meaningful physical activity.

He and his son experimented with additional weights of approximately five pounds, which made arm movements more demanding.

Rogan and Carmack also discussed VR boxing and the possibility of developing gloves that provided resistance or physical feedback when punches connected.

Carmack recognized that ordinary VR boxing involved shadowboxing rather than real impact, but still considered it capable of providing strenuous exercise.

The recurring challenge was how to combine convincing virtual action with physical effects without making the equipment cumbersome, uncomfortable, or dangerous.

From Ferraris to Tesla: Carmack's Experiments With Extreme Horsepower

How an Old British Sports Car Introduced Him to Mechanical Engineering

Despite his later collection of exotic cars, Carmack was not particularly interested in automobiles as a child.

Computers mattered far more to him. He suggested that, when younger, he would have preferred an advanced Apple computer to an expensive sports car.

His first car was an ordinary Volkswagen Jetta.

His interest in automotive engineering developed after he acquired an old British MGB.

The car was unreliable. Carmack recalled problems beginning almost immediately, including a failed clutch master cylinder.

Keeping it running encouraged him to learn mechanical maintenance, engine tuning, and the practical consequences of modifying physical systems.

He studied automotive publications and increasingly technical material.

Eventually, he replaced the MGB with a Mazda Miata, appreciating its more modern and reliable version of the lightweight sports-car experience.

But his experiments were only beginning.

The Ferrari 328 That Became an Esports Prize

Carmack bought his first Ferrari while still young and working during id Software's early success.

He chose a Ferrari 328, attracted by its appearance and character.

A conversation at the dealership revealed something that bothered him: despite its reputation, the Ferrari was not necessarily faster than less expensive performance cars such as Chevrolet Corvettes.

Carmack became determined to increase its horsepower.

Most Ferrari specialists resisted the idea of making extensive modifications.

They generally regarded the original engineering as something to preserve, suggesting comparatively minor changes such as exhaust upgrades.

John Romero eventually introduced Carmack indirectly to a different kind of specialist.

Romero brought an automotive magazine into the office containing an article about Bob Norwood, a Dallas performance engineer known for modifying exotic cars.

Norwood was willing to pursue much more ambitious work.

He proposed turbocharging Carmack's Ferrari 328.

The modified car eventually approached approximately 500 horsepower, far beyond its original output, before suffering engine damage.

Carmack was willing to accept mechanical failures as part of the experimentation.

The same vehicle later became the prize Dennis Fong won at Red Annihilation.

It was a remarkable intersection of Carmack's two interests: aggressive mechanical experimentation and competitive video games.

The Ferrari also reflected the disadvantages of extensive modifications. Its unusual engine-management systems, emissions compliance, and maintenance requirements made it more difficult to own than a conventional road car.

The Testarossa With 1,009 Horsepower

Carmack eventually moved to a Ferrari Testarossa, which offered more space and a larger engine for performance modifications.

Working with Norwood, he installed twin turbochargers and pursued increasingly ambitious power levels.

The resulting vehicle eventually produced approximately 1,009 horsepower measured at the rear wheels.

Carmack emphasized that this was wheel horsepower rather than an estimated figure measured at the engine.

The Testarossa's greatest strength was not necessarily launching from a standstill.

Its gearing, turbocharger characteristics, and traction limitations made low-speed acceleration difficult to exploit fully.

At higher speeds, however, the car produced extraordinary acceleration.

Carmack described its performance between approximately 50 and 150 mph as exceptional, sometimes exceeding that of powerful motorcycles in rolling-acceleration situations.

The additional power demanded corresponding changes to tires, suspension, and traction management.

It could also create dangerous handling behavior when boost arrived suddenly.

Carmack eventually reduced its output to approximately 600–700 horsepower before selling it.

He warned the buyer about potential mechanical problems and later understood that the car experienced additional failures.

Why the Ferrari F40 Was Different

Carmack also owned a Ferrari F40, a car designed around a much more explicitly race-oriented philosophy.

Unlike his heavily modified Testarossa, the F40 remained comparatively close to its original specification.

He made adjustments to the turbocharger wastegate system but did not attempt the same transformation.

The F40 required aggressive driving to access its performance.

At low engine speeds, it could feel comparatively weak. Once the turbochargers produced substantial boost, its character changed dramatically.

Carmack also recalled the car's unconventional interior details, including simple door-release cords that could confuse people unfamiliar with the design.

Its appeal was not only raw speed. It offered a particular mechanical experience that more practical cars did not reproduce.

Why Carmack Eventually Preferred Tesla

By the time of the interview, Carmack had become an enthusiastic Tesla owner.

He had previously owned an early Tesla Roadster and was driving a Model S P100D.

He regarded the Model S as the best car he had owned.

Its appeal was partly technical.

Electric motors provide rapid torque response without conventional engine gear changes, turbocharger delays, or the elaborate tuning required by his old Ferraris.

Sophisticated traction control made the available power easier to use.

The Model S also offered everyday comfort and reliability while delivering remarkable acceleration.

Carmack still recognized the distinctive violence of his thousand-horsepower Testarossa during high-speed acceleration.

But Tesla demonstrated that extraordinary performance did not require an uncomfortable, mechanically fragile exotic car.

Carmack's 2019 Roadster Speculation Became a 2026 Demonstration

Why More Horsepower Eventually Stops Helping

During the interview, Carmack discussed the proposed next-generation Tesla Roadster, which he had reserved.

He was especially interested in Elon Musk's suggestion that the car might use propulsion technology associated with SpaceX.

Carmack explained that extremely powerful cars eventually encounter a fundamental traction limitation.

A tire can transmit only a certain amount of force to the road before slipping.

Once that limit is reached, adding more engine or motor power does not automatically improve acceleration.

Compressed-gas thrusters could theoretically provide additional force without relying entirely on wheel traction.

Carmack discussed storing gas at pressures around 10,000 psi and releasing it through controlled nozzles.

By carefully directing those nozzles, an experimental vehicle might generate forward thrust or additional downward force.

The latter could increase tire grip and allow the electric motors to transfer more power to the road.

Could the Roadster Actually Jump?

Carmack extended the idea further.

If thrusters were positioned and controlled appropriately, a car might briefly lift itself off the ground.

He imagined using several nozzles to control the motion and potentially produce a short jump.

He distinguished this from sustained flight, which would demand much more propellant and energy.

At the time, this was an engineering discussion about what might be physically possible, not evidence of a working vehicle.

Seven years later, the proposal became significantly more concrete.

On October 15, 2026, Tesla demonstrated a redesigned Roadster developed with SpaceX technology.

According to Reuters' October 15–16 reporting, the demonstration included a vehicle using thrusters to jump over obstacles and maneuver in the air.

The company presented versions incorporating acceleration thrusters and more elaborate flight-related capabilities.

Tesla said production would begin in 2028, following years of delays since its earlier Roadster announcement.

The distinction remains important: a controlled demonstration is not proof that an affordable, road-approved, commercially delivered vehicle can perform those maneuvers routinely and safely.

Nevertheless, the demonstration established that the basic direction of Carmack's 2019 speculation had become a real engineering project.

It also connected two of his longstanding interests: extreme automotive performance and computer-controlled propulsion.

Rocket Belts, Personal Flight, and the Limits of Human Control

Carmack's aerospace experience also informed his thoughts about personal flying devices.

Traditional rocket belts offered extremely short flight times.

He discussed hydrogen-peroxide-powered systems capable of keeping a person airborne for only about 20 seconds.

Such devices require the operator to carry propellant and manage a rapidly changing, unstable physical system.

Carmack recalled that early rocket-belt pilots experienced serious injuries, including broken legs.

The problem was not simply insufficient thrust.

Personal flight demands precise control of position, direction, and rotation. Small errors can quickly become dangerous when the entire body is suspended by propulsion.

Jet-powered systems offer different possibilities.

A jet engine uses oxygen from the surrounding atmosphere, unlike a rocket that must carry its own oxidizer.

That can permit longer operation for certain personal flight devices, potentially extending flight time from seconds to minutes.

Carmack believed modern computerized stabilization could significantly improve safety.

Small drones demonstrated how rapidly control software could manage complex movement, compensate for disturbances, and keep an unstable flying machine oriented.

Applying comparable control technology to personal propulsion could make experiences possible that would be exceptionally difficult for a human pilot to manage alone.

However, practical safety would still depend on propulsion reliability, available energy, control authority, emergency recovery, and the consequences of failure.

Carmack's interest reflected the same principle visible in his game engines and rockets: software can make difficult physical systems more capable, but it cannot eliminate their underlying constraints.

Why Carmack Still Wanted Humanity to Live Beyond Earth

Space Exploration as More Than a Commercial Industry

Despite stepping away from Armadillo Aerospace, Carmack remained deeply interested in space exploration.

His ambition extended beyond launching satellites, developing reusable rockets, or selling suborbital tourist flights.

He wanted humanity to establish permanent settlements outside Earth.

That included the Moon, Mars, and potentially artificial habitats in space.

Carmack saw practical reasons for becoming a spacefaring civilization.

Permanent settlements elsewhere could provide resilience against catastrophic events affecting Earth, including major asteroid impacts.

But his motivation was not solely about protecting humanity from disaster.

He considered exploration and expansion valuable in their own right.

The Influence of Science Fiction

Carmack drew inspiration from optimistic twentieth-century science fiction, particularly writers such as Robert Heinlein.

Such stories imagined people using engineering to build new communities, explore unfamiliar environments, and expand the range of human experience.

He contrasted those visions with a modern world in which geographical exploration often seems complete.

Satellites, maps, and GPS have made even remote places increasingly familiar and measurable.

Space offered a different frontier.

Carmack wanted people to grow up regarding the Moon as somewhere human beings actually lived and worked.

He imagined lunar settlements becoming recognizable destinations rather than remaining distant abstractions associated only with government exploration programs.

That vision also explained his enthusiasm for private launch companies.

SpaceX, Blue Origin, Rocket Lab, and other ventures suggested that technological progress did not depend entirely on the political priorities of national governments.

Private investment introduced alternative ways to develop aerospace technology, although large government contracts and public infrastructure continued to play important roles.

Virtual Worlds and Real Frontiers

There is an apparent contradiction in Carmack's ambitions.

Virtual reality allows people to experience extraordinary environments without physically travelling anywhere.

Space exploration aims to make entirely new physical destinations accessible.

One reduces the need for travel. The other expands the places people can travel to.

Carmack did not consider these goals incompatible.

Virtual environments can make valuable experiences affordable and accessible to people who might otherwise never encounter them.

Physical exploration creates new opportunities that do not yet exist.

A convincing simulation of the Moon might provide education, entertainment, or an extraordinary sense of presence.

A permanent settlement on the Moon would create something fundamentally different: a new place for human life, scientific work, and economic activity.

Both ambitions depended on overcoming difficult engineering constraints.

Both required making sophisticated technologies cheaper and more reliable.

And both offered ways to expand what people could do beyond the limits of their immediate circumstances.

The Engineering Lessons Connecting Carmack's Career

Carmack's work across games, virtual reality, artificial intelligence, automotive engineering, and rocketry reveals several important distinctions.

His experience with Rage showed that improving a product indefinitely can eventually make it less commercially useful.

Armadillo Aerospace demonstrated that technical achievements and expensive experimentation do not guarantee a sustainable business.

SpaceX and Blue Origin illustrated competing approaches to development speed, financing, and acceptable risk. Their subsequent achievements also show why engineering philosophies must be judged against results rather than personalities.

The source-code releases from id Software demonstrated a different route to lasting value: allowing other people to continue improving technology after its original commercial purpose has faded.

Oculus showed why removing technical inconvenience can matter as much as improving maximum performance.

Neuralink raised questions about whether new interfaces could restore human abilities—and how far biological enhancement might realistically extend.

His work habits reflected a personal decision to remain close to difficult technical problems rather than exchange hands-on engineering for greater organizational authority.

Throughout the conversation, Carmack repeatedly distinguished what can be built, what should be built, and what can be delivered before circumstances change.

Those are different questions, and his career contained successes and failures arising from all three.

The lasting significance of his reflections is not that every prediction was correct. Some remain untested, and others have developed in ways that neither Carmack nor the wider industry could establish in 2019.

It is that he treated extraordinary ambitions as engineering problems that must eventually confront time, money, physical limits, human behavior, and measurable results.

For the programmer who helped make virtual worlds possible, those constraints were never a reason to stop imagining new ones—or to abandon the possibility of reaching real worlds beyond Earth.