Notes from Lex Fridman’s 2.5 hour interview of Elon Musk

Crew Dragon
Elon couldn’t sleep the night before the first Crew Dragon launch; “it was a great relief”
highest orbit in 30-40 years
would be tragic if Apollo was highest mark for humanity
**next goal is moonbase, then get people to Mars
he’s chief engineer of SpaceX, signed off on all design decisions

SpaceX Starship
“really hard problem”
If you could magically solve one engineering problem with it? “Engine production”
**prototypes are easy, production is hard
most advanced rocket engine ever designed (the Raptor)
why so hard to manufacture at scale?
“complexity + unique materials”
**holy grail is fully reusable orbital rocket – reduces cost by 100x, imagine if you had to buy a new car every time you drove it
in theory get down to $1-2M cost per launch, put 100 tons in orbit
like karate kid catching fly w/ chopsticks, but much bigger
“we’ll get Starship to work” – the physics pencil out
it’s important to get done, we should keep doing it or die trying, “fuck that we’re gonna get it done”
**physics is a law but everything else is a recommendation

First principles can be applied to any walk of life
example: cost of raw materials, versus cost of the finished product, how to reduce that delta
manufacturing is an underrated problem – easier to design, harder at scale
what’s theoretical perfect product look like? what tools / methods / materials to get there

Mars
“least inhospitable planet”
**timeline for human on Mars – best case 5 years, worst case 10 years
obstacles: cost per time to orbit and cost per time to surface of Mars
right now it costs $1T, we need to get that way down ($1B per ton, need to improve by 1000x – a self sustaining city needs a million tons)
need to be multi planetary species, eventually sun will get too hot (500M years)
earth 4.5B years, first time in its life we have possibility of expanding life beyond earth, maybe window of opportunity could be short, we should act quickly
**“life insurance for life itself”
great filter is Mars city can survive even if Earth spaceships stop coming (eg, if you lack vitamin C)
not sure if it’ll happen in his lifetime, but at least give it momentum
Lex: imagining a civilization on Mars gives people hope

Rules / regs / society on Mars
**would prefer direct democracy instead of representative
laws should be short enough that people can understand them
Hasn’t been world wars recently, operates as a cleansing function (garbage collection) to rules and regulations, because they’re immortal (humans die but rules don’t)
If they accumulate every year, eventually you can’t do anything
Special interests depend on them
Lex: self-dying laws (like C v Java)
Rules & regs are like software for operating a civilization, becomes bloatware
usually biz deals are over lawyered, over complex

Wormholes / space travel
**you can’t move faster than speed of light, but if you can make space itself move… (eg, Big Bang, space expanded faster than speed of light)
energy required to warp space is way too high

Internet cookies
**Always trepidation when you click “accept cookies” – small chance of opening “portal to hell”

Crypto / doge
“Doesn’t understand this whole smart contract thing”
Mars will need its own currency, can’t accurately synchronize with Earth (4 to 20 light minutes away)
Crypto is interesting approach to error that is the database of money
Money literally is heterogeneous mainframes running old Cobol in batch mode
Gov has editing privileges on editing database, can use whenever they want to make more money
**Money should be viewed as info theory – bandwidth, latency, packet drop, etc
Crypto is attempt to reduce error in that information
**“Money is database for resource allocation across time and space”
If running economy, need efficient value ratios between goods & services – ratio of exchange, and shift obligations across time (eg, debt)
**Dogecoin’s merits – higher transaction vol capability than bitcoin, cost per transaction is very low
Bitcoin’s throughput made sense when created 10 years ago, but now is comically low
If anyone, Satoshi is Nick Szabo given his role in the evolution of bitcoin’s core ideas, but “what is a name anyway?”

Tesla autopilot
Lex + Elon first connected over this
“people give me too much credit”
humans drive with optical sensors (eyes) and biological neural nets
full self driving must recreate in digital form – silicon neural nets + cameras – “only way, I don’t think there’s any other way”
Lex: “what’s even an open car door, man?”
control logic isn’t hard part, hard part is “a lot of freaking software”
must convert massive bitstream into vector space
**challenge is building this accurate vector space
after that, control problem is like a video game
your eyes paint color in corners, and in blind spot
brain does crazy post processing on visual signals from eyes
**brain constantly trying to compress and forget as much as possible
Lex: human mind seems to go beyond vector space, into concepts (eg, “this is a school zone”)
full array of neural nets in the cars boggles the mind
wrote own C compiler for max efficiency
lots of bare metal programming
computer can see in dark incredibly well, tiny differences in photon counts
Lex: when solve level 4 FSD?
“looking like next year”
rate of self engagements (switching back to manual driving) is dropping dramatically – probability of accident on FSD < human driving likely to happen 2023 – but should get to 2-3x lower probability

Tesla bot
Lex: does it have role in the home?
yes, possibilities endless
not in Tesla’s primary mission
work will become optional in future
if it’s dangerous or boring, humanoid robots will add most value
pair with some kind of UBI
over 2M tesla cars
Lex: could there be more tesla bots than cars?
haven’t thought that far…code named Optimus subprime
AI in one computer, in real world, much harder than AI in room with many servers
robots – need good real world AI + good at manufacturing
bots could be very good companion
wabi sabi – subtle imperfections
robot that has wabi sabi – map to their human counterpart’s imperfections
focus today is “make it useful” – decent prototype by end of 2022

Dan Carlin’s Hardcore History podcast
Elon thinks greatest podcast ever
should be titled “engineer wars” – engineering plays pivotal role in battle
**in war, what matters is “pace of invention” and access to high quality fuels, raw materials
fuel needs consistent mixture + high octane – germany never had good access to oil, but US had awesome fuel
talked to dan about all this
**most of history is just people “getting on with their lives”
but wars tend to be written about
Stalin book by Montefiore – had to stop reading because too dark / rough
life was really tough for most of history
a good year is not that many people in your village died of eg disease war starvation
now food plentiful and we have obesity problem

Lex wish to interview Putin
Lex: if happened, would Elon join briefly?
“sure i would do that”
Soviets had impressive rocket tech
friendly competition is good
**govts slow, only thing slower is collection of govts
both rec Everyday Astronaut by Tim Dodd

nuclear power / radiation
nuclear power is great way to generate electricity, we shouldn’t shut them down
radiation sounds scary, people can’t calibrate it
ate locally grown veggies on TV in fukushima
impact (of fallout from fukushima disaster) is greatly exaggerated
“everything’s radiating all the time”
**if you wanna know nuclear fire, go outside – sun is gigantic nuclear reactor

**where there’s hunger today, almost always due to war or something – the world has enough food / money to solve hunger

on balance british museum is net good (re: how they took items from our countries through war)
if you judge them, must judge what everyone else was doing at the time too

Lex: would elon do standup?
could try…15 minutes of material

advice to the young
do things that are useful, very hard to be useful
net positive contribution to society
use the mental tools of physics and apply broadly in life – they’re the best tools
read a lot of books
ingest as much info as you can
develop good general knowledge, learn a little about a lot of things
man’s search for meaning
find what you’re good at + like doing
**as kid, he read through encyclopedia – really helpful
lot of respect for honest day’s work, doing useful things
**lot of morally questionable behavior is because they have a zero sum mindset – they may not realize it

what’s role of love
really perplexing question
what is love…baby don’t hurt me
foundationally i love humanity…i wish to see it prosper

what is 42, meaning of life
what author (douglas adams) was saying: universe is the answer, we need to figure out what question to ask – that’s the hard part, framing the question