@doshkim
Account based inUnited States
About this account
- Account based in
- United States
- Connected via
- United States App Store
Account-level information from X, not a live location or the device used for a specific post.
Building for humanity in the era of AI @getDelightAI • Backing the inspiring stories of founders @valoncap • Loves brain, space, energy, longevity, complexity
Yet on 🌎
Joined March 2007
- Tweets3.8K
- Following845
- Followers3.3K
- Likes6.5K
Impressive
I reviewed Tobi’s biomarker’s from the car race yesterday. His heart rate, respiration, nervous system, focus, brain state, and more.
It’s insane.
I have some observations below. First, you need to get into the right headspace to understand the insanity.
Imagine this…
> you're breathing at an 8 min/mile (5 min/km) pace
> in a 110°F (43°C) heat chamber
> heart rate of 145 bpm, peaking at 171 bpm
> intermittently body checked 700 lbs (320 kg) of force
> your core body temp near 102°F (39°C), a high fever
> claustrophobically strapped in a six point harness
> helmeted head weighs 65 lbs (30 kg) under g load
> no evaporative cooling
> you’ve lost 4.6% of your body mass through sweat
> you can’t move
> vision narrowing from thermal strain and dehydration
Now, in these circumstances, your job is to…
> drive a tightrope at 180 mph (290 km/h)
> overtaking/defending against ahead, beside, behind
> staying within millimeters of tolerance
> lose concentration for a tenth of a sec, in trouble
Do this for 2.5 hours straight.
A few additional things I find interesting from looking at the data:
1. It’s crazy to me his body can be in such a heightened state while completely stationary. Sitting down, he produced a cardiac load equal to high intensity physical exertion.
2. Thirty minutes before the race, Tobi’s body was already racing. His heart rate elevated to 136 bpm, more than 2x his baseline.
3. In qualifying, the higher his heart rate, the faster his lap. Arousal and performance moved in sync, including his personal best lap.
4. He lost 4.6% of his body mass in sweat. This is severe dehydration. At 2% dehydration your reaction time measurably decreases. He surpassed this in the first hour.
5. His nervous system was switched entirely to fight. There was no chill, anywhere.
6. His incredibly high breathing rate was driven by extreme sympathetic stress and rising core temperature, averaging 41 breaths a minute on his flying laps, moving 93 liters of air a minute.
7. His peak respiratory rate was 59. This is an all out sprint!
8. The only life experience I can map onto Tobi’s experience was doing 27 mg of 5-MeO-DMT, the world’s most powerful psychedelic. It was all consuming, taking me to the cliff of my physical and mental capacity for 45 minutes. Otherwise, it is very hard for me to imagine being able to excel in these extreme conditions for 2.5 hours.
|
I now understand a little more what Tobi was telling me about the flow state he drops into. Under such extreme physiological conditions, the brain ruthlessly narrows focus. His brain downregulates the self-monitoring prefrontal regions (the self narration). The spike in norepinephrine dilates time, creating the perception of time slowing down.
MRIs on professional drivers show they use roughly one brain region for tasks where novices light up fifteen. A novice would be crushed in these circumstances in less than 10 minutes.
I’m impressed. It’s very hard and nearly impossible to actually understand what it’s like to go through something like this.
Tobi and @dhh (Tobi’s teammate) were both my first customers at Braintree and gave my startup life when it wasn’t clear if we had a future. They generously encouraged other startups in the Ruby on Rails community to use us. DHH’s co-founder @jasonfried did as well. They’ve been great friends to me.
As to Tobi’s performance in the race, he did exceptionally well:
> closed within +0.51% of the fastest Bronze driver, putting him neck and neck with the top ranked amateur drivers in North America
> got his fastest lap of the race on Lap 88 (1:15.359), over two hours in and with peak dehydration and thermal strain, showing incredible tenacity
> finished just +2.60% off absolute class best, competing against world class professionals
There were 150 drivers competing in this race. Respect to all of you. I have a whole new appreciation for the sport.
Source: journals.aps.org/prl/abstrac…
Adding a shortcut can still make every commute longer. Once each driver chooses the route that looks fastest, the new road may attract enough traffic to congest links that everyone depends on. No driver can improve by changing routes alone, yet all arrive later.
In a 2008 Physical Review Letters study, Youn, Gastner, and Jeong modeled self-interested routing across several urban road networks. They found a “price of anarchy”: individually rational choices produced worse collective travel times, and blocking selected streets could sometimes improve traffic.
Organizational analogy would be that a new approval path, dashboard, or communication channel may reduce friction for each user while routing more work through a shared bottleneck. When things are not moving fast enough, the missing ingredient may not be more capacity. More often than not, it's due to the same bottlenecks.
Imagine several people privately want to abandon a failing process, but each assumes everyone else supports it. A few visible endorsements could change behavior quickly, even if almost nobody changed their mind that day.
Bacterial quorum sensing offers a useful comparison. As Miller and Bassler explain in their review, bacteria release chemical signals and detect their concentration. When enough signal accumulates, it can trigger changes in gene expression that coordinate activities such as biofilm formation. The mechanism depends on signals in a shared environment; it does not require a leader issuing instructions.
When a team is learning something new, perfect consistency can be the wrong goal. Repeating the same approach reliably is useful once it works. Before that, some variation may help you find what works.
In a 2014 Nature Neuroscience study, Wu and colleagues asked 20 people to make repeated arm movements. Participants couldn't see the path their hand took. Instead, they received a score rewarding how closely it matched a hidden curved path. People whose movements already varied more in the direction the task rewarded learned faster. Follow-up experiments found that variation related to the task predicted learning better than simply being more variable overall. This was movement learning, not a study of companies. The important distinction was between exploring useful alternatives and adding random noise.
A team experiment worth trying: when a sales pitch stalls, keep the facts accurate but let people test different ways of explaining the customer's problem. Compare what happens before standardizing the script. The question isn't "are we consistent?" yet. It's "are we varying something that could teach us how to improve?"
The person you’re talking to gets replaced by someone else. Would you notice?
In Simons and Levin’s 1998 experiment, a researcher asked pedestrians for directions. Two people carrying a door briefly blocked their view, and a different researcher took his place.
Only 7 of 15 pedestrians (less than 50%!) reported noticing the swap. Paying attention to the conversation didn’t guarantee noticing who was having it.
A product-design lesson worth testing: “it’s visible” isn’t the same as “users noticed.”
If a status change matters, don’t just update the screen. Test whether someone focused on their task can tell you what changed.
If a decision needs a fresh insight, don't schedule the first serious look at the problem five minutes before the decision meeting. "Sleep on it" only helps if there's something to sleep on.
In Wagner and colleagues' 2004 Nature study, people practiced a number puzzle with a hidden shortcut. They then spent eight hours asleep, awake overnight, or awake during the day. When they returned to the puzzle, 13 of 22 sleepers discovered the shortcut, compared with 5 of 22 in each awake group. A separate test found no such sleep advantage when people hadn't practiced the puzzle beforehand. The researchers proposed that sleep reorganized what people had learned, making the hidden pattern easier to see. This was one laboratory task, not proof that sleep solves every business problem.
A practical experiment for teams: separate the first look from the final decision. Share the actual problem early enough for people to work through it, then leave a night before discussing solutions. Compare that with meetings where everyone encounters the problem and must solve it on the spot. The useful change isn't simply more rest. It's giving rest something to work with.
“My AI is taking notes” might change what or whether you remember.
In a 2011 lab study, Sparrow, Liu, and Wegner asked people to type 40 facts into a computer. Participants who believed the facts would be saved recalled fewer of them than those who believed they would be erased—even though everyone was asked to remember them.
That doesn’t mean AI notetaking is bad. It does suggest a rule for teams: test retrieval, not just capture. When a decision matters weeks later, will the right person remember it or even know where to find it?
Why does a new practice take hold in one team and stall in another? A broadcast may create awareness; it does not necessarily make trying something feel normal.
In a 2010 Science experiment, Damon Centola assigned people to artificial online networks and tracked whether they registered for a health forum after their network neighbors did. In clustered networks, 53.8% registered, versus 38.3% in random networks. A second neighbor's adoption significantly raised the chance of joining. The study tested forum signup, not durable health behavior.
For workplace change, the useful hypothesis is about repeated local proof, not maximum reach: people may need to see several peers try a practice before they try it themselves.
