“Civilization advances by extending the number of important operations which we can perform without thinking about them.” A. N. Whitehead
In the previous two newsletters, I outlined the recent lifestyle changes I’ve made to create predictable, high-quality sleep and achieve ideal wellness biomarkers through diet. A process which required firing many unreliable versions of myself.
It’s tempting to think that once you’ve figured out an optimal behavior, it’s done. However, that’s actually the easy part. The hard part is making it effortless; minimizing the amount of time you need to spend thinking about them. This enables you to level up again.
Operating my new routines is expensive because they require my attention. To put this into numbers, operating myself daily requires about 1,555 calories. Managing my new protocols may require up to 20% of my daily energy budget.
Which begs the question: if my sleep and diet algorithms are solved to be high functioning, why should I continue to drain my mental battery managing them?
Our fixed biological energy capacity is a fundamental constraint to what we can do or become in life. To overcome or make optimal use of this limit, we use technology to expand and extend our abilities. For example, ride a bike versus walk. We also work to become more efficient and creative.
But if we want to think really big about the future of being human, we need more advanced ways of managing our biological energy expenditure. Achieving peak performance and improving continuously while also freeing up more energy to explore new frontiers.
Thinking about this led me to the idea of the Autonomous Self (see Appendix for Six Levels of the Autonomous Self).
Energy improvements underlie humanity’s progress:
- hand stitch to loom
- wagons to horses to the combustion engine
- abacus to computers
Humans are next.
Let’s look at how practical needs give rise to the Autonomous Self. I am currently trying to solve for four objectives:
- Automate my health algorithms to avoid having to think about them
- Capture long tail gains, i.e. the small stuff that adds up
- Achieve algorithmic improvements that exceed what my brain can do on its own
- Find new insights that continue to level up performance
(On what things do you spend large portions of your daily biological energy that you wish you could automate?)
Objective #1: Automate my health algorithms to avoid having to think about them
Some elements can be automated through habit formation or permanent physical setups:
- Ninety day cycle of biomarker tests and diet plan creation
- Final meal of the day at 10 am
- In bed by 8 pm
- Blacked-out, temperature-controlled room
Other parts, though, need real-time management. For example: what, when and how much I eat depends on my exercise routine for the day. These variables affect my resting heart rate and heart rate variability (HRV) which I know from my experimentation are strongly correlated with high quality sleep. Keeping this information top of mind while trying to be functional in other aspects of life, is a challenging yet critical variable.
Objective #2: Capture long tail gains; the small stuff that adds up
I’d like my health routines to incorporate everything I’ve learned, not just the 20% of things I focus on to get 80% of the benefit (the “80/20” rule). Implementing everything I’ve learned, every day, feels like being on a unicycle while juggling three bowling pins. It’s hard to do other things at the same time. Yet maybe five small things add up to a 7% performance improvement. Right now, I’m not taking advantage of these long tail insights. It’s too much.
Objective #3: Achieve algorithmic improvements that exceed what my brain can do on its own
Increased capability of measurement offers one of the more promising paths to self-improvement. My Whoop strap, for example, provided me with daily REM, Deep, HRV, and HR measurements which enabled my initial experimentation and resulting insights. And my 90-day diet routine generates a huge amount of data about dozens of biochemical processes in my body. These are both inaccessible to introspection.
With these biomarkers flowing in, I can then focus on a few of these variables and perform an experiment. However, if I want to include a few dozen variables in the experiment to find new relationships between them, I’ll need to use some sort of computer intelligence.
The key here is that the algorithms can also use me as the test subject, telling me what to eat or when to go to sleep or turn off devices because the algorithm is, on the backend, generating hypotheses about my personal, optimal, conditions for thriving.
Objective 4: Find new insights that continue to level up performance
We are in the early days of sleep and biomarker science. For example, why can’t a full night’s sleep be instrumented and artificially accelerated to occur in one hour? Sure, that’s a big leap. But do we know enough about sleep to know whether it’s possible?
Autonomous Self: A Definition
The Autonomous Self can be understood through the frame offered by A. N. Whitehead:
“Civilization advances by extending the number of important operations which we can perform without thinking about them.”
Whitehead’s point applies to human learning and growth too. Humans grow more autonomous by increasing the number of important operations that they can perform without thinking about them (See Appendix for Six Levels of the Autonomous Self).
The Autonomous Self is based on the premise that the Self is more than our conscious awareness and the symbolic terms and ontological primitives we have to represent it today. Skin as the boundary of Self is arbitrary.
Our subconscious and continuous dynamic interaction with the outside world can be thought of as an extended, expansive Self that has gradations of being both autonomous (e.g. digestion, wound healing and driving) and self-directed (e.g. voluntary control such as problem solving).
Sleep and biomarkers are two examples of ontological primitives that we don’t necessarily consider to be part of “Self”, largely because each is also simultaneously self-directed and autonomous, with very little cognitive control over when or why our bodies crave the things they do. Fighting to stay awake, as we all know, is a losing battle.
Mapping out a plan for my Autonomous Self has me focused on converting self-directed activities to run autonomously and for autonomous activities to improve continuously without my attention.
For example, what are you willing to change for a good night’s sleep? Conversely, what would you give up to avoid having a bad night’s sleep? Up until now, answering this kind of question in a way that clearly justifies these sacrifices has been out of reach, forcing us to rely on guesswork. The lack of data on associated costs has resulted in our cultural norms devaluing and deprioritizing sleep. (When was the last time your primary care doctor asked to see your sleep performance data?)
Late last year, at Kernel, I acquired data on myself that gave pretty compelling evidence that an aspect of my impulse control is determined by the amount of deep and total sleep the night before. Every day I make hundreds of decisions in my roles as CEO and founder. For the first time, I gained a glimpse into the question, “How valuable is my getting a good night sleep to customers and my co-workers?”
I’ve never had the numbers before, but now I do.
Why I Care: The Future of Intelligent Life
Feeling great is alone worth the effort but my greater interest in the Autonomous Self is in trying to figure out a path to the future of being human. My primary hypothesis: Our future existence requires that we level ourselves up as a species, and at the fastest evolutionary speed in history. To do this, we need to free ourselves of the costly metabolic things we do today, such as rote or biased decision making and logistics management around solvable things such as sleep and biomarker-based diet, exercise, or lifestyle. Leveling us up to spend our precious time and energy to explore the frontiers of being human rather than things we know how to do efficiently. What will happen?
It’s hard to imagine what our minds will do with a new abundance of energy, but we have a precedent: Fire. Fire freed our ancestors from certain caloric and dietary restrictions, which opened up energy — i.e. metabolism/time — for little things like language and society as we know it to develop. I believe a fully Autonomous Self will open up, again, just as much energy. One can only dare imagine what we will do with it. We will have the opportunity to develop new industries, discover original uses of the mind, make iterations of governance and economics, and explore the goal alignment problem within ourselves, between each other, and with AI.
How far away is this? It’s already begun.
Inner Space Exploration
In the 1966 movie Fantastic Voyage, a submarine crew is shrunk to microscopic size and ventures into the body of an injured scientist to repair damage to his brain. A similar adventure happens in Parasites Lost (in the Futurama series) and Rick and Morty Anatomy Park.
These cartoonish depictions are now starting to become reality, with intelligent sensors in place of fictional humans. The sensors themselves have thus become the pioneering explorers of our time, heading into the information rich terrain inside ourselves. In the near future, with these sensors on and in my body, my health routines will be autonomously managed by closed loop systems. Analogous systems already exist today in the form of pacemakers, brain implants to control seizures, and insulin pumps with real-time glucose monitoring.
The key to all these systems is that the measurements happen and actions are taken without conscious awareness. Next-generation sleep technology is already moving in this direction, too, by detecting sleep stages and modulating the acoustics and lighting of the environment for optimality.
From Startup R&D to Scaling
Taken all together, this is the manual version of “startup”, Autonomous Bryan Johnson. My sleep and diet algorithms are needy. It’s time to move past this R&D and into the scaling phase. Not only to automate the manual processes that are happening today and consuming so much of my attention, but to start the next levels of the game, which are going to be even more interesting.
At the most basic level, if your body and mind are well-governed and have a healthy economy on the inside, your chances of being a part of a healthy society increase. If everybody does it, society itself levels up.
Appendix: The Six Levels of Autonomous Self
One way to think about the Autonomous Self is through a familiar lens of autonomous airplanes or cars. Most of what we do, most of what we think, most of what our body processes and manages, is done entirely without our knowledge or awareness.
My ideal Autonomous Self would be at a Level 4 for most things and a Level 5 for the truly unnecessary decision making:
Level 0, NO AUTOMATION
The automated system issues warnings and may momentarily intervene but has no sustained bodily or cognitive control. The Self is in charge of full-time performance of all living tasks, even if “enhanced by warning or intervention systems”.
- Car examples: Seat belt alarm; dashboard warnings; collision/swerve detection.
- Body/Cognition examples: Basic sensory abilities e.g. light vs. dark detection, sound vs. no sound, odor vs. no odor.
Level 1, ASSISTANCE (“hands on”)
The automated system shares control by using information about the environment, with the expectation that the Self performs all remaining aspects of task.
- Car examples: Cruise control; adaptive braking; parking assistance.
- Body/Cognition examples: Phototropy; chemotropy; pupil responses; fight or flight; regulation of respiratory rate and pulse; thermoregulation.
Level 2, PARTIAL AUTOMATION (“hands off”)
The automated system takes full control of movement and cognitive basics. The Self must monitor and be prepared to intervene immediately at any time if the automated system fails to respond properly.
- Car examples: Car has full control of accelerating, braking, and steering but driver must keep watch and hands on the steering wheel at all times and driver does everything else.
- Body/Cognition examples: Wound response/repair where Self is expected to use pain signals to avoid further damage or use; metabolic/dietary cravings generated by automated system where Self is expected to seek out necessary food/nutrients.
Level 3, CONDITIONAL AUTOMATION (“eyes off”)
The Self can safely turn their attention away from most cognitive tasks with the assumption that the Self will and can intervene when requested.
- Car examples: Driver must be vigilant and physically prepared for emergencies, rain, parking lots, etc but can mostly turn attention elsewhere.
- Body/Cognition examples: Skilled/learned movements; walking; language generation
Level 4, HIGH AUTOMATION (“mind off”)
Similar to Level 3, but attention is recommended but not required for safety, optimality, or health. Autonomous Self will fix errors if Self responds improperly.
- Car examples: Robotic taxi or delivery service which does not need human intervention; car has the ability to stop itself.
- Body/Cognition examples: Digestive system; all of perception.
Level 5, FULL AUTOMATION (“body optional” / “steering wheel optional”):
No intervention is required at all. A Self goes about one’s day optimally guided for all the operational cognitive tasks and decisions that are thought undeserving of attention, freeing up metabolic capacity to explore higher order cognitive discovery.
- Car examples: N/A
- Body/Cognition examples: N/A