Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
Reuben Metcalfe Special Guest | AI Biz Hour with Andy and John Ep #147
Recorded: June 3, 2025 Duration: 2:19:30Space Recording
Short Summary
In a dynamic discussion on the AI Biz Hour, industry leaders explored the latest trends in AI, including NVIDIA's strategic partnerships and innovations, the growing importance of power generation for AI systems, and the transformative potential of AI in healthcare and creative industries.
Full Transcription
Thank you. Hello and welcome to the AI Biz Hour with Andy and John.
This is your host, AI John Allen, and Andy Wergedal is here as the co-host.
How are you doing today, Andy?
I'm good. It's Tuesdayuesday we've got a bunch of
really interesting news things happening and uh i don't know if everybody's got that but x on my
account has x chat enabled now so which means you can do encrypted conversations with x chat so
that's pretty interesting too as well so uh nvidia a bunch of stuff that they're announcing
uh them and dell have announced a major collaboration a whole new you know
supercomputer at the berkeley lab i mean it's every single day is new every single day is like
a tidal wave and if you don't get on this know, if you don't get your surfboard and figure out how to surf on that wave, that thing is going to
crush you and your business, maybe even your entire business sector. Because I've got some
friends who were longtime senior consultants who have worked for years and decades doing amazing things that right now,
their particular function can be done by AI better than they could. And so they're struggling to
find work. So it's fascinating, but also quite challenging.
Yeah, I mean, there's a lot with a lot of uncertainty comes opportunities, but you got to be flexible, ready to pivot.
And it helps if you're really curious and want to learn. you've had something that you've counted on for a long time and it's done very good for you,
it must be very jarring and a little unnerving for the rug to get pulled out from under your feet.
Yeah, absolutely. And, you know, folks have been doing stuff for a long time, you know,
for a long time, architecture, design, analysis, data side, infrastructure side, security side.
And now half that stuff can be done in a blink of an eye with an LLM.
I had a really interesting conversation with a business owner that has some significantly
sized businesses. And I asked him, I said,
well, you know, what do you want your executives to learn? And he said, they're just going to have
to come up to speed. I'm not going to teach them anything. I'm just going to give them a tool.
And so his whole perspective is we're building the, you know, super awesome robot answer machine.
And then we're going to turn them loose on it and we'll see if they come up
to speed or not.
I thought that was really interesting perspective from someone who has a
couple of businesses that are worth, you know, a lot.
I won't say how much, but a lot.
I, you know, there's no easy.
So, you know, I mean, the thing is, is people really want easy answers to complicated problems. And I think especially now. And so they're going to, there's going to be a lot of people running to Pied Pipers and who knows what. But, you know, it it is what it is. I feel fortunate to be here at the AI Biz Hour on a daily of, you know, working with outside of the space.
And so it's, you know, for me personally, it helps to be here on a daily basis and hear from other people what they're doing, opportunities, what's going on, learning, you know, Mr. E's here.
I've learned a lot of sophisticated high-level prompting through Mr. E.
Dylan's learned a whole bunch of stuff from
Dylan, from you, Andy. You know, TVD's showing up. There's a lot of people here who show up on
a regular basis and bring a lot to the table and really helps a lot. Today, we have a special guest,
Andy. Yeah, I know. I'm excited to talk to Ruben. I'm fired up to hear all the
stuff that he's doing and what's going on in his world and his perspective because, you know,
you guys have heard me and John talk enough. We'd really rather have other people tell us
what they're doing because it's way more exciting. Ruben, how are you today? How are you doing?
Ruben, how are you today? How are you doing?
Testing, testing.
Ostensibly well so far.
How's the mic? Are we coming in?
It's sounding clear as day.
We do a little bit of news.
Do you want to help us with one or two news stories?
And then we'll get into it as people show up.
Perfect.
Yeah, by all means, yeah.
All right.
So, Andy, if you have any news you want to get to,
I've got three news stories.
Well, yeah, sure.
The first one is that NVIDIA, even though at the beginning of the year
they announced their new thing and then they took that hit,
their stock is still up 24% year to date.
So I don't see it going down.
Just as a kind of Captain Obvious question here.
Broadcom shipped a new data center.
Cool Colagix and Lambda launched
the first Nvidia B200 accelerated clusters.
And they're doing a super, what they call a super micro AI optimized hardware cluster.
So essentially what they're trying to say is, we take small concentrated things and we're running sequenced activities through those small, super micro AI optimized clusters.
I think that's really kind of an interesting thing. Ruben, what do you think about that?
I think speed is hard because of an old problem that we have known as the memory wall.
Y'all familiar with the memory wall?
Go ahead and tell us. Yeah, yeah yes but i don't know if
everybody is well so i wasn't until quite recently uh i was um sort of tapped by a friend to review
a pitch deck of a silicon chip company and i'm not that guy i'm more sort of software and the
abstract you know i'm more software than hardware in general,
but I sort of understand,
I've been in enough things to sort of get the broad strokes,
you know, in a lot of different areas.
But these guys were working on a chip
and it was a useful chip.
It was designed for AI clusters.
And in learning about why their solution was useful,
I had to learn about the problem, which was huge.
And what's happening is, you would have heard of Moore's Law,
which I personally think is a bit of a suggestion,
but that's fine.
It's basically, you know,
compute speeds double over a certain period of time.
And, you know, it appears if we sort of look into the last, you know, 30 years or so, that curve looks elliptic, right?
We get more and more compute. It keeps doubling every so few years and so on, right?
However, memory is not moving at the same speed, right?
same speed, right? So we're getting fast, really, really fast processors, but the ability for those
So we're getting fast, really, really fast processors.
processors to like access the memory and get all the stuff that you want to process into them.
It's like there's, it's like you've got like a giant tank that you might have for like wine or
milk or something, right? Just huge multiple stories tank. And that's all of your data. And then you've got a massive liquid processing engine, right?
We'll call that the processor.
And those engines have been getting really, really big.
And those silos have been getting really, really big.
But the pipes that connect them have been staying the same size.
And because of that, we're starting to get throughput challenges
with some of the larger models,
particularly when you're starting to talk about
like, you know,
however many millions of parameters
for some of the new models, right?
And so it's not that we can't process the data.
It's that we can't get,
we can't organize the data in a way
where we can shove it into the pipe fast enough
for them to be able to actually leverage all of that processing power.
And so they're doing things like taking the memory chips and putting them right next to the processors and like duplicating the data and like loading it up right next to the processor.
And it sounds like that's what's happening here with NVIDIA is it's not so much the compute power as it is the throughput.
And that requires a new kind of chip,
which NVIDIA is bringing to the market.
Yeah.
Bigger models.
Yeah.
Yeah, I tend to state it like this.
It's like you said,
big tank, big processing unit, little pipes.
And instead of they're trying to grow the pipes,
but the pipes, they can't figure
out how to make the the input output of memory faster it's the bottom line yeah and so all
they're doing is they're running parallel processing so in other words they just add
more pipes and so that's why you have so so think of it um you have a CPU. Now we have four core CPUs, which is essentially four CPUs on top of each other, so they can process all 64 channels or 120-bit channels at the same time.
input and output and that's so they instead of making one network channel they make motherboards
with multi-channels in multi-channels out stacking cpus on top of each other and they're just
paralyzing everything meaning meaning parallel processing that's why yeah yeah think think of
the design of colossus it's something like a hundred thousand gpus all series together so that they could all
or parallel together but so they can all operate essentially at the same time because no one's
saying we need faster cpus no one's saying we need bigger memory storage or elastic memory storage
we already have that we already have c CPU power that we can do that.
But it's the loading it into RAM to take advantage of the CUDA libraries
and the NVIDIA GPUs.
That's why we have, you know, the data center design is multiple CPU racks
and huge sets of parallel GPUs.
When in the past it used to be all cpu with a little bit of
ram next to it and huge memory well we fixed the memory problem we fixed the cpu problem and now
the bottleneck is caching into ram and that's that's what the memory wall is right but it was
a great great description huge tank huge huge processing engine, little tiny pipe.
And so either you make the pipes bigger or you have to put more of them in. That's just
the memory wall that you have. It's like lanes and a highway.
Yeah, we've talked about lanes and the highway as the lanes in. So if you have a 64-bit chip,
that means you have 64 lanes. And it can process 64 answers at the same time, but only 64.
So you just, the more input you can send into it, the more it'll go, the faster, it'll go as fast as it can, as long as there's stuff loading it.
So, yeah.
Anyway, okay, cool.
All right, well, the next one that I have, wait, let me flip back to the other one.
The NVIDIA announced at the Red Hat Summit that they have increased acceleration to the inference model.
Okay, so we have 4-bit and 6-bit and 8-bit models.
We have 4-bit and 6-bit and 8-bit models.
But now what they're saying is the NVIDIA Dynamo framework supports GPU auto-scaling with Kubernetes automation and network optimization.
Essentially just trying to solve the problem you just mentioned, which is we put things in a block.
we put things in a block, we run them separately,
We run them separately.
we can then process input and output from that
and then coalesce it some other or aggregate it
or combine it in other places.
And these are all hardware problems that they're fixing
so that because the desire is add more ideas,
add more input, add more details,
and then the throughputs from the different models is, is getting up in the,
you know, seven, seven trillion mil parameter model is pretty,
pretty common, but now they have 27,
47 trillion parameter models. I mean, this stuff is getting crazy.
So where is the infrastructure to address this problem?
Ruben, is it all?
Big giant warehouses full of GPUs, CPUs, fans, and air conditioning and power.
Mostly power.
Power, power, power.
By the way, if you want to go long on any industry right now,
don't go long on people who are building chips.
Don't go long on, by the way, this is not financial advice.
This is Andy's brain.
Sorry, I'm not going to give you advice, but I'm going to tell you,
it seems likely that going long on,
whoever generates power to these big systems is going to be the big, big, big winner.
CPUs and GPUs will become a commodity.
LLMs and their usefulness will become a commodity.
We're going to build stuff and put them on our desk.
It's going to be a commodity.
You're going to build your own system.
That's a commodity.
But your power generation is going to go up.
And you're going to have to manage increased power requirements for wherever your data systems are
or wherever your computers are that are
connecting to it. So either network or power or both seem to be the big, big, big gap, big winners.
Cool. So I'm going to do a little house clean just for one minute. Everybody, please like the
space. We've got a full house, but we've only had three people share the space.
So please share the space.
And also this is recorded so you can give it to friends.
Ruben's done a lot of very interesting things.
Everybody's gonna really enjoy today's episode.
So please show us some love, like, and repost the space.
And also if you tag a friend or two
that you think would be would be appreciative
appreciative of listening today's space please put them in when you repost and like it also
please follow myself and andy and follow ruben turn on notifications so whenever we're doing
the space you'll get to know it i'm gonna put our uh our our uh newsletter in the Purple Pill.
And so please join our newsletter.
What I do every day is we transcribe the space
and then that goes in the newsletter.
So then you have a document you can refer back to
because often we talk a lot about technical stuff
and stuff that can be very helpful to you.
So please sign up to the newsletter.
So you'll have that.
Also, if you miss an episode,
you can go see it on the newsletter often.
And so I think we're going to start, Andy.
Are you good with that?
Or do you want to get to one more news?
Or how are you feeling?
Well, I just want to recognize we have spaces,
Andrew from Spaces Dashboard in the listeners as well.
So I think that's the first time we've had Spaces Dash dashboard. By the way, if you don't know what Spaces
dashboard is, you can click on
there, put your name in. It'll tell you
how many spaces you've been in, spoken,
hosted, co-hosted, whatever.
So welcome, Spaces.
Welcome, Andrew. By the way, great name.
No, let's go for it. Let's
get to Ruben. Ruben's the guest. I mean,
nobody wants to hear me anymore. I'll
jump in as i
see fit but let's go to reuben it's way more important
excellent i just want to let everybody know today is uh sponsored by uh gov bid mike who helps people with businesses um to to get government contracts as well as grants and he's in the
newsletter i'm sure he's going to show up we're. And he's in the newsletter.
I'm sure he's going to show up.
We're going to put him in the purple pill.
If you have a business and you want to expand,
I highly recommend you get in touch with him.
You mentioned the AI Biz Hour and you'll get a 10% discount.
He's a great guy.
He's totally solid.
And so I really recommend him wholeheartedly.
We have Zoo in the audience.
We have RV.
We have Mr. E.
We have Michael.
We have Nicholas.
We have TVD, Joe, Dylan, Timo, a lot of great people here.
This is a networking space also, so it's good that you get to know each other.
And so, Ruben, can we start at the beginning
when I was born uh yeah uh let's see commercial geolocational but none of just like like what
what what inspired you do you think to get you to where you are today like was there a moment in
time when you know like like what abilities did you recognize
or someone recognizing you that you think got you
to the place that you are currently?
It depends on the year.
Is there an overarching theme?
Probably.
But the lived experience is more just a bunch of different lives that sort of
had start start the names of chapters um and perhaps it might be useful to put that in perspective
um we have how about like um like a notches on the belt type list with like where and the when
and the what sort of mixed in what do you reckon that that sounds that sounds perfect just just so people have context of of oh yeah yes yeah yeah sure so i
grew up in new zealand brew and i spent a little bit of time in australia and it's been about a
year in japan and i've been the last 13 years in amer. And I have done all sorts of things.
I've started six companies.
Most of them died horrible deaths.
A couple of them, thankfully, for different reasons, you know, which is a waste of test.
It's okay to fail, but hopefully you're, you know, learning the lessons, right?
And they ranged from, my first business was importing paint and vehicles from Japan to New Zealand.
We used to go to the Nissan vehicle auctions.
Japan has this cool thing where the longer a vehicle is on the road in Japan, the higher the registration fees are.
And so by the time you've had your vehicle on the road for five or six years, it's actually cheaper to just get a new car and then pay lower registration fees, right? And the reason they do
that is they want to get the domestic market for their vehicles, but then they don't want to keep
them too long because they want a new fleet. And so they ramp up the registration fees to encourage
people to sell it back to the company, swap it out for a new model, and then they sell the older
models overseas, right? And that works great because, you know,
countries like my country in New Zealand,
we don't manufacture cars.
We've got, you know, 5 million people and 25 million sheep
and the sheep are driving.
So someone's got to make them.
And yeah, that's what I did as I went over there
and bought cars and heavy-duty anti-corrosive steel primer paint to be applied at 192 microns
thick. It's the same stuff they paint the Golden Gate Bridge. It's to stop your bridge
rusting, basically, or your ship or whatever. So we did that brick and mortar type stuff.
And then I got into utilities brokerage. So gas, electricity, telecommunications,
group purchasing
agreements between small organizations and large organizations um called free procurement and i
learned about insurance there as well and then i almost accidentally built 26 000 websites for the
national clearing teacher association and i started a thing that was not a space lottery
but looked a little bit like a space lottery where where people would pay $10 for getting a chance to go to space.
Of course, that's illegal if it's a lottery, a sweepstakes, or a raffle, because gambling is bad, and you shouldn't do it.
And they said the government, in which case it's fine.
But we weren't doing that.
We were selling a slightly large piece of paper called a poster.
And we were selling the poster for fair market value with a free chance to go to space as defined by the New Zealand Gambling Act of 2003 as a sales promotion.
So we read a sales promotion, giving people 10 bucks for a chance to go to space on Richard Branson's Bird of Galactic thing.
We're going to get 25,000 people. We're going to buy the ticket for 200 grand, pocket the 50K and then do it again. Right.
we're going to buy the ticket for about 200 grand pocket the 50k and then do it again right
um and we never quite hit the limit so i gave everyone a refund and sort of mothballed the
project but in the meantime it got me to america and i got a bunch of i got to meet a bunch of
real space nerds and uh learned all about space stuff and so you know between that i was uh i was
at samsung for a little while as their guinea pig on for now in residence to work on some data stuff back when we were calling it big data in 2013-14.
And what else?
And I was early crypto.
So like 2009 on Bitcoin, forgot to hodl.
Whoops.
But I was sort of familiar with the ethos and the infrastructure.
And more recently, I've been focusing on the big one.
I built a legal tech thing, which made it easier for people to find out all the class actions they're a part of, file a claim and collect their funds.
And we scraped PASER, the federal docket system, through all the court cases, and extracted out all the settlement agreements,
figured out what the eligibility criteria was,
chucked that in the back of a chat bot.
It was back in 2017.
And the end result was people would talk
to those little bear on the internet glasses like mine.
And he'd ask you questions like,
oh, did you purchase anything
with a lithium-ion battery in it?
Say, for example, a phone between the years 2000 and 2011.
Oh, you did?
Okay, well, you're part of the lithium battery antitrust settlement,
which is a $300 million settlement as a result of lithium-ion battery manufacturers' price fixing
and it knocks on to the consumer and the FTC gets involved and so on and so forth.
But then you've got to try to figure out which, you know,
which of the 300 million people are going to get their $9.50 at the end of it,
which can be a little complex, right?
And so we built a machine to figure out all the class actions,
put them in one spot, figure out what the questions were,
and then build a legally compliant structure to be able to let
them file their claims and collect their funds. And that was the big one. We got up to 113,000
users on that. Before I learned some hard lessons about picking your financial backers,
we didn't go with venture. We did go with angel, but went when it time to get to like venture scale we ended up going with a litigation funder which is like old like almost like a hedge fund style
very very heavy very blunt but yeah it didn't it didn't quite work out unfortunately the business
model was great it was the first time I had started business that didn't fail for business
reasons but I got to learn a lot of lessons I learned a lot about law and that actually has It was the first time I had started a business that didn't fail for business reasons.
But I got to learn a lot of lessons.
I learned a lot about law.
And that actually has led me to my recent work, which is trying to build a kind of, I almost want to call it like an Esperanto, which is like a, actually, that's a terrible way to articulate it. I'm trying to make it so that machines can understand law and so that they can write contracts
as fast as you and I can describe what we want
to have the machine actually make legally executable code
that allows us to have terms and preferences.
Yeah.
Yeah.
Yeah, I mean, that's, it's,
cause I looking, so you it's because I looking.
So you started off as a musician.
I forgot about that part.
Yeah, I was a musician.
I was a drummer and a singer for six years, touring around mostly New Zealand and a little bit in Australia.
In between all the other stuff.
I was also a fire dancer and I was briefly a poet in australia imagine
uh and other sort of quirky little bits uh but uh oh and like the last year and a half i spent
driving around america uh i got to visit i drove from here in california all the way down to new
orleans up to maine and then back to California,
which ended up taking about a year.
So it's not sort of like the traditional arc, I suppose,
but people say that a jack of all trades, a master of none. I don't know if I'm a jack of all trades or a master of none,
but perhaps a jack of some trades or a master of none uh but perhaps a jack of some
trades and master of a couple you know so so with all that you've done you you don't have a formal
education in computer science or like you is your background really business yeah so i i ran away
from home and school when i was 15 years old. And it was my dad's
way of the highway. And I chose the latter. I am my father. So I'm not we're not. Well,
it's it's we're like a 90% crossover. So you know, you know how that goes sometimes.
But I tell you what, I got an early start, first business at like, what, 19, I guess
it was. And yeah, the gamble that I was making with myself, because at the time, Napster had
turned up, right? Napster and Kazana and Morpheus and all the other, LimeWire was a big one.
All the file sharing technologies had just come out.
And you can sort of see the writing on the wall for, you know, the compact disc CD.
And I thought, well, you know, if music's going to have a hard time over the next wee while,
you know, a pirate band and so on, it might be hard for me to make a living as a musician,
but maybe I can figure out this business thing.
Dad's a business guy and he's an idiot, so it can't be that hard.
And then of course I spent the next decade
trying to make it work.
So, you know, c'est la vie.
But, you know, business was accessible to me
as a musician without an education
because mostly people didn't care
about whether or not I had a degree.
They just wanted the product
or the service
that they wanted right and so if you can make something that's useful to people and people are
willing to pay the price you're willing to offer it for well great you're a business person and you
don't need to get an mba although of course it helps right a lot of respect for people to go
through the traditional uh path the discipline i'm not sure I had the discipline, to be honest, to get through academia.
Although I did read section settlements agreements for 18 months, which I wouldn't recommend.
It's a lot of dense language. But yeah, it turns out every book is a children's book if the kid can read, as Mitch Hedberg once said.
So you're basically self-taught, which is incredible.
And, you know, in astrology, there's this thing sometimes they say, you will find your fortunes in foreign lands.
You will find your fortunes in foreign lands.
And it seems like that, you know,
some people really excel when they leave the nest
and they don't just get out of town,
but they get into another country.
And it sounds like you have a little bit
of an understanding of the future
or you can kind of see what's coming down the road.
Do you feel that's an accurate statement?
So I'm not a soothsayer um i am i do follow trend lines and i can i i study history enough to be
able to not predict the future because you know history doesn't repeat but it rhymes and I know how to rhyme. Yo. So, you know, to the extent that, you know,
there's a window before the 2008 crash,
before everything tumbled over and some saw it before others,
you know, I might be in like, I don't know,
maybe chronologically,
maybe like the first to 10th percentile of like figuring things out,
which is not to say that I'm a first to figure things out.
Most people, you know, leave that to the experts.
But I tend to be early,
mostly because I try to collect mentors like Pokemon.
Got to catch them all, right?
And I find that is...
Actually, that is one thing that I do
and have done always
that I don't find other people doing is assuming they know nothing from the get go and assuming they're going to have to work from first principles.
And then finding people who have done all the hard work and then asking them what they would do if they were in your shoes.
And it's an insanely simple way to articulate it.
But I swear the smartest things I've done
have been other people's ideas.
And the smart thing that I did was say,
what would you do?
And then just finding the right people
to ask their question to, right?
And you can go up to them
and you can sort of ask that question upfront.
And some people might give you time
or some others won't.
But if you can frame the
question within their domain of expertise and you can you can articulate
did we just lose ruben
i think ruben, are you?
Well, he's now a listener.
Let's hit him back up to invite to speed.
I was enjoying the conversation.
Absolutely.
Are you back, Ruben back.
He might be traveling.
We're not sure.
Ruben, we lost you for a minute there.
So we're going to see if we can get you back up.
I'm not sure what could be the cause of these things.
But, of course, this is X.
And so there's always the unknown unknowns and the known unknowns.
And then we'll keep on seeing if we we can do this i see reuben down
down there let's see speaker again oh he's back fantastic a thousand apologies uh yeah i'm gonna
for some reason the wi-fi ratter and these parts like to cut out for 30 seconds every eight minutes
uh so that's convenient uh but we're back where were we uh so so the so
finding the right people to ask the right questions and how to frame those questions
to to to get into a position of um of insight prepared yeah yeah and and you know what's
interesting um i found out about halfway
along there's a big difference between asking what should i do and what would you do in my shoes
massive massive difference and it's so subtle but if you ask people and you say hey what should i do
they will often go into the framework of like, this is my client or this is my, you know,
they go into counselor mode, but often people's counselor mode is not the same as their sort
of first party, okay, I'm like putting them, really putting themselves in your shoes and
asking themselves what they would do.
Because often people treat others slightly differently in ways they treat themselves.
Because often people treat others slightly differently in ways they treat themselves.
And so I found, honestly, maybe like a 40% plus quality between the difference of asking like, hey, what would you recommend that I do?
Versus what would you specifically do if you were literally in my shoes?
It's a slight difference on that turn of the question, but I find that once people really sort of inhabit the situation, they start to care a little more and they get into the story and they give better answers, to be honest.
It's a slight difference on that turn of the question.
But I've been very fortunate to have people who I couldn't afford to pay give me the time and attention to do you know, do what I'm trying to do better. Right.
So do you collect people on a constant basis? I don't know if that's the right wording of it.
Oh, sure. I mean, you know, maybe not like in a, in a, in a knapsack, but like, uh, you know,
I'll, it's just the other thing, actually. By spreading your information diet around different sources, right?
So I've got, like, a friend of mine, Dave Moskovitz, who's, like, really good for, like, interpersonal relationships.
He's sort of a bit of an ambassador back in New Zealand, right?
I might go to him for, you know, some HR stuff or some leadership type stuff.
you know, some HR stuff or some leadership type stuff. I've got another friend who is kind of a
generalist, but he is like on the board of like a meat company and a beer company, just old school,
sort of brick and mortar, but a lot more bricks and a lot more mortar than I ever did. And I turn
up with my complex problems and I'm like, what is this? What about that?
What if we get the JSON,
and we turn it into an articulation,
and we can transpose it across different programming languages,
like, you know, Ruben, Ruben, Ruben,
slow down, slow down.
Look, look, most good businesses are boring.
All right?
I understand you got all your technology and your fancy fan wave,
but, like, it's sounding sounding exciting and that's concerning.
I want it to be boring.
Bring me a boring business.
I want a boring business.
Come back when it's boring.
And it's, you know, a 10-year-old could give that advice,
but it's so, to the extent that it's accessible advice
but uh but it's easy to get wrapped up in the context right whatever it is that you're doing
and um particularly a lot of the inventive people that i've met are so inventive and so uh you know
often often articulate and narratively compelling that they start to believe their own stories. And that can be a little bit of a slippery slope, that one.
So I try to surround myself with people who explicitly think differently than I do
because that's where I find insights that I wouldn't actually land on.
And it also sort of spreads the surface for them a bit as well you
know if you if you call the same person every time for every problem you know um it's it's an
intimate relationship but it's um it's a heavy relationship whereas if i have you know a hundred
people that i call like once every two years man they they're so they're so stoked they're so stoked to hear from you and they want to catch up on what's happened i haven't heard for two years, man, they're so stoked. They're so stoked to hear from you,
and they want to catch up on what's happened.
I haven't heard from you in two years, you know?
And so it's low-touch but high-impact relationships
rather than high-touch, high-impact.
That's just the trait that I like to make,
and I find that it informs perhaps a diversity of views
that I might not have had I sort
of concentrated in a particular field.
Yeah.
I mean, I am really fascinated in terms of, did this come to you naturally?
Entirely accidentally.
Because I didn't learn anything.
And so I had to learn on the fly.
And when I want to get in a new area, and when you're starting as well,
when you've got nothing, right?
Like my second business was me and a phone, right?
I literally had, this is kind of like humble brag slash,
there's other words for this that are not friendly,
but I had a pair of business shoes
and I knocked on so many doors
that I wore a hole in the bottom of my business shoes
and I couldn't afford to get a new pair of
business shoes, but I could afford some cardboard to put into the bottom of the shoes so that I
wouldn't get my feet wet. And, you know, when you're in that kind of sort of place in your life,
right, you better get out there, you better start knocking on some doors and you meet people.
And, you know, some of them are like, well, I don't have time for you. Come back when you're serious. And other people are like, come here, you young buck.
By the way, the collar goes on the outside. And your tie's all messed up. Come here.
And I've been very lucky to have had
some more seasoned types sort of
take me under their wing and straighten my jacket out and so on.
And I think that sort of culture changes from place to place, but that culture is everywhere.
I haven't found a culture where mentorship doesn't exist, and it expresses itself a different way, sometimes familiarly, sometimes in a sort of educational type environment,
often in business environments we have the whole journeyman,
you know, apprenticeship type models in different countries.
But I think, you know, a lot of people are asking for,
they sort of assume they don't have permission to ask
or just nobody's ever given them the time.
To be honest, I think I just lucked out early
and realized how in over my head I was
and had to reach out.
And then I realized it was really useful to do that.
And then I sort of internalized it into a, you know,
a bit of a method, I guess.
Yeah.
But I find extremes are useful for the purposes of illustration.
And so to the extent that you can find people,
not just who are experienced with the thing,
but if you can find a difference of an opinion
in a field,
and you can get two people with differing opinions
to give you, to act as resources
to run things past them. I found that's more
useful than finding a single expert. And I'll give you an example. When I first started the
class action company, I didn't know anything about law, really. In fact, I'm not sure I had
filed my taxes at the time. But I knew that it was broken. And I knew that it was broken because
I had a postcard that said I got $3.80. And I knew the postcard was probably going to cost about as
much to send me that. So there's a bucket of money somewhere and it's not being distributed.
This is a breadcrumb. And at the end of this breadcrumb trail, there's going to be a loaf
or a very fat rat. And I want to find out which, as my friend once articulated it. And so, but I didn't know anything, right? And so I said,
okay, well, I know that lawyers are fighting each other all the time. And I know that the way that
they fight is a way that I can't fight. I don't know how, I don't know how the tools are. So I
can't, I can't really pretend. I can't just really
wing it. You can't wing it in law. However, there are a lot of experienced actors there.
So I reached out to some of the people that sounded like they were saying the kinds of
things that I was trying to say, which is that the consumers are getting screwed over
in class actions and they weren't really working.
And then I asked those people, I said, Hey, who are the, like,
who are the, who's the Elvis of this world?
And actually not just like class action world, but like the major factions, right?
You've got the plaintiff's bar and they're finding all the lawsuits and you
got the defense bar and they're defending all the lawsuits.
So like between the defense bar and the plaintiff's bar,
if we assume that they have different opinions about things, you know, who would you say are like the pillars of those kinds of thought?
Who can I look to for like modeling how this kind of actor might think?
And they, you know, one of them was like, oh, this guy over here.
And he was a very, very fancy class action plaintiff's attorney.
And so I said, okay, well, that's the plaintiff's bar.
Now I've got to try to find out, like,
if that's Batman, then who's Lex Luthor, right?
Or if this is Lex Luthor, who's Batman,
depending on your view in that world.
And I looked on the internet for this guy
and I was looking for somebody who had something negative to say about this person
because I wanted to hear the other side as well, right?
And so I found out that he actually had Alex Luthor
and it was another attorney on the defense bar
who represents Google, Facebook, Walmart, Niantic, Twitter
back in the day, and a very, very fancy class action defense attorney from Cooley, Mark
Rhodes.
Yeah, super crazy intelligence, like hypercarb, very linear kind of thinker, but flexible
enough to be able to go linear.
He's a very interesting brain.
Anyways, so I went to that guy first before I went to the Batman.
And I said, hey, I'm thinking about hitting up Batman because I think I want to build a thing that's going to be cool for everybody.
And he's like, oh, really?
You want to talk to Batman?
Well, I'm Lex Luthor. And, you know, tell me everybody. And he's like, oh, really? You want to talk to Batman? Well, I'm Lex Luthor.
And, you know, tell me what you're doing here.
And yeah, sure, I'll be your advisor.
You know, I'll want 1%.
And, you know, but, you know, yeah, Batman's all right.
You know, he brings some BS to the court sometimes,
but he plays by the rules.
Yeah, sure, you should go speak to Batman.
So I spoke to Batman.
And Batman was like, oh, really?
You've already got Lex Luthor on the board, I see. Interesting. Uh, yeah, sure. Okay.
We can't be in the same room. Uh, but I'm happy to take your calls or whatever, you know, kid.
Uh, and, uh, and, and the result of that was from that point on throughout the rest of the business
and it would go for about four years is I was able to, if I had a question of law,
which is not a question I can answer, right, because I don't have a JD,
legally, but also in spirit, right, it's just a very, very complex domain,
I could call either one of them, or for the really hard ones, I could call both of them.
And when I got the answer back from both of them and it was the same answer, great, I can have a high confidence signal that this is the right way to do the thing,
right? But if I would call them and they would differ, now I have the opportunity to say, okay,
well, maybe this is expressing as a difference of opinion between this kind of actor in the ecosystem
and this other kind of actor in the ecosystem and how that expresses through their biases from their positions, right? And so to the extent extremes are useful for the
purposes of illustration, we can borrow from that in our narrative and in our sort of information
diet to help better contextualize their decision-making, you know, respectively, you know?
their decision making uh you know respectively you know yeah that's um i mean number one i'm
just kind of curious how you got past the first gatekeepers for lex luther and uh so you could
get his attention um i mean you know i've called attorneysorneys are pretty good about picking, if you call them and you have a reasonable question, it's not too hard to get past the gatekeepers.
And a lot of, it might be American, I don't know, but a lot of people have trouble assuming they don't, or being honest with themselves that they're in over their head and they don't know what they're doing did that happen to you real quickly
being that you left home at 15 is is that correct or do you have to kind of check yourself still
to make sure you you keep that goingidence calibrates over time. But I have found even when I feel confident,
I try to at least stress tests with first principles, which is to say, okay, here I try
to at least write down my assumptions, even if I'm high confidence. And then that way,
if something goes wrong, I can run up past those assumptions
and see if, okay, let's say I might have five assumptions about a thing. And one of the
assumptions is that Mary is going to be there. And another one is that people are going to be
lifting wood and not metal. And, you know, so you take the, whatever the thing is that you're
trying to do, you take the assumptions, you bake it in. And if you don't articulate them, then sometimes when something goes wrong, you don't know where to look for problems.
But if you articulate the assumptions, then you can go back across, okay, well, something didn't work and I don't know why it didn't work.
Maybe it's because one of these assumptions was wrong.
was wrong. Okay, let's mess around with this assumption. Would that have any impact? Yes or no.
Okay, let's mess around with this assumption.
Would that have any impact?
Yes or no.
And it sort of allows you to sort of index into a degree of granularity that you might not have
otherwise if you didn't list the assumptions. And it also somewhat protects you from your own bias,
right? Another really, because the other thing is, is that I'm, I think on a hardware level,
I'm an optimistic person.
I like a version of the world where people get along,
and I like the idea that we can not be dicks to each other, man.
So I think I have a predisposition
to hoping that things turn out well.
And because I come up with ideas as well. Um, I even often,
I often bake in optimistic assumptions into things. And that's, that's, that's a cool
kind of lifestyle and a state of mind to be in, but it's a, it can be very dangerous if you're
building tools for other people and suddenly your undersights or oversights, you know, start impacting other people, right?
There's the whole phrase of moving fast and breaking things,
which I'm a fan of, but if you're going to break things,
like, can you break your own things, man?
You know, like, yeah.
So, you know, I think it is prudent for the creator or the builder
not just to come up with the ideas and probably not just to build it as well, but it'll do all the additional work of trying to figure out how that's going to impact people and bring on stakeholders accordingly and hopefully pad things out. right? And so I certainly have a lot to learn when it comes to management. I'm not a particularly
good manager. I can speak one to many, or I can speak one to one. But day to day organizing a
small team, there's a lot of I've met people who are so much better at that. And so I try to, uh, I try to recognize where I can do well and add value.
And I try to recognize, and it's, it's almost more important to recognize where other people
are actually better for this, you know? Um, cause that allows you to sort of, um, contribute in a
more meaningful way to a group. But, uh, yeah, it takes time, man. Yeah.
Did that answer the question, or did I go off the rails there a bit?
No, no, absolutely.
And just I was curious, Lex Luthor,
was there any resistance getting to him directly?
Well, this is an interesting thing, right?
This particular Lex Luthor wasn't so hard to get to because he only turns up once every couple of months in the world.
He only turns up when things get real weird because it's not a normal kind of attorney,
right, where like a regular litigator might have, you know, a dozen cases or something this week.
But class action take class actions. I've seen class actions take over 10
years to settle. And so they'll have some big cases, but they're all long time frames are a
little bit more accessible. And to be honest, it's so niche that nobody really cares. So it's not
like he was, you know, on the front page of the New York Times or something where he has this big
social profile. It was more uh okay well can i
find an insider to this industry and then what does that person think and that's that's that's
and i because because otherwise i wouldn't even know where to look right and so sometimes you
don't need to get the right person you just need to get to a person who would know the right person
and that can get you a step closer to signal rather than just noise on the first pass.
And do you show up with a deck, or do you have your resume, or are you just like,
hey, this is what I want to say, and that's enough to perk their attention?
Well, you know what it's like, right? It's the funnel, right?
So if you turn up with an essay, you might not walk out with an essay.
No, if you turn up with an essay,
you might get the cold shoulder
or just like the overwhelm.
But if you can sort of take
sort of the crux of what you're trying to do
and chuck it into a soundbite, right?
Maybe that's 30 seconds of conversation on a phone
or maybe that's a DM on a social network,
or, I don't know, a postcard, whatever you want to do.
But just try to respect the attention of the person,
give them as much information in a shorter period of time as possible.
Maybe not so much information that they can make a decision
as to whether or not it's interesting or not,
because some people will sort of flick things aside if it sounds like
after they've taken the wrong assumption. But if you can just sort of speak to, you know,
here's the thing, here's roughly what I'm trying to do. I know that this is your area because of
one, two, three. Do you have a moment to, you know, discuss X? And the X, hopefully, is like within their domain
and sort of broad enough that they get to have opinions,
but narrow enough that it makes sense to specifically speak to them.
And that's a fine balance.
But if you can get it right, then you get to be the person
who asks the question they've never heard before.
And that sounds like signal to them.
And if you can start the relationship that way and you can maintain it over time,
then they're always going to be happy to pick up the phone. You know,
that's, that's huge. Um, that I, I love that. And I'm so excited. We, we, you know,
we have a podcast now, so this will be out in the podcast world. And I think that there's
a huge amount of value. You know, often we talk about AI, we haven't really been talking about AI
so far. But, you know, Reuben, I can't thank you enough. We're not over yet. But Andy, I know,
what's, are you have to leave soon?
Yes.
Yeah, I do have to leave.
But I do have a question, Ruben.
And I'm just going to ask my question and listen to the answer, and then I'll just drop off so you don't have to interrupt.
But what have you used AI for, and what do you see its benefit, and how do you think it's going to affect you and the things that you're doing in the future?
How long you got?
Yeah, so it feels like a civilizational scale tool
to the extent that we're able to meaningfully do things
that we couldn't before.
I have spent perhaps a little over a,000 hours in dialogue with LLMs,
which is, you know, not like a full-time job,
but it's a fair amount for sort of self-study, right?
And, you know, that's not...
LLMs are like a part of it.
The more, I think, broader surface there is the neural networks in general, right?
And so whether you're using neural networks for large language models or neural nets for image generation,
or I think we're about to see a new wave of audio technologies that come through.
wave of audio technologies that come through. One of the reasons that audio has been slow to
sort of mature as far as general technologies is because the objects in an audio track are
different to an object in a picture where I can take a million, like 100 million pictures and feed them into a training model set.
And, you know, it can look at the picture and say,
okay, this is where the cat starts,
and this is where the cat stops, right?
And it can do that with the pixels and so on.
With audio, however, everything sort of blends
across the frequency ranges.
And so it's very difficult to tell the difference between,
for example, you know, the
sound of a cat versus the sound of somebody pretending to be a cat, particularly if that
person is good at mimicking things, right? And so, you know, separating the trumpet from the bass
drum, from the guitar, from the piano or whatever, it's a little messier if you're dealing with audio
than if you're trying to do it with pictures of these instruments.
And the maturity of the technology is coming to follow.
But we're starting to see things like voice clones and so on. Eventually, there's a lot of things that we have relied on for trust, such as facial recognition, iris signatures, fingerprints, vocal signatures that we've never had to really examine
over the last few decades as a security service. But because it's getting faster and faster to
model and replicate patterns, whether they be audio patterns or visual patterns or speech
patterns, we're getting into a place where a lot of
the systems that were built on assumptions of, okay, well, this can't be copied, suddenly
can be copied.
And so I think there's going to be, I'm going to try to answer your question.
Okay, yeah, I'll come back to this part on the back.
Actually, no, I'll wrap it up.
I think there's some security issues.
I think we're going to face security issues
in ways that are not fully appreciated by everybody,
but are definitely, like, a lot of the security folks
are definitely aware of them,
and they're working on problems,
and some of them are doing a really good job as well.
So I'm not trying to do the,
this is not a doom and gloom army gimmick situation.
Smart folks are working on the job, but security will change. Um, media will change. Um, it will,
it may, I hope that as a culture, we beget more streetwise a little bit. I think that we have a
lot of work to do and, um, cultures, and generations and just being literate.
Recognize the difference between somebody who's trying to sell you something or teach you something.
And those are pretty squishy concepts, so that can be hard to teach people in a sort of traditional kind of way.
More about exposure.
But I think that the next generation will be a little
bit more streetwise, a little, you know,
have a little more cognitive security, if you will.
I think that
playing the piano might be similar
to, you know,
how we sort of treat finger
painting, right? Anybody can finger paint.
You know, Picasso is going to be better at it, but just because you're not Picasso doesn't mean
you can't finger paint, right? And so I think that a lot of the creative arts will see an influx of
new creators. Not only are they going to come with equivalent skills as the old creators,
but they will come with new ideas, which I think is a net positive.
And of course, you know, more the merrier anyway, right, in my opinion.
That'll change a lot of the scarcity models, right?
We're already seeing 100,000 uploads a day onto Spotify.
We've only written like 130 million songs in the last 4,000 years,
which means that that's going to double every
four years, which is kind of insane. So, you know, there's a lot of media proliferation and
the creator economy seems to be expanding on the supply side. The demand for creative works
is going to shift a little bit on the other side. I don't think there's going to be any less money
that goes into creative arts,
but I imagine that it will be spread
between more creators, right?
So I don't know how that plays out,
but I certainly think that, you know,
I would rather a generation that has better tools
that can be more expressive
than restricting that
so that only a few people
can get by doing that thing, right?
I don't know,
I feel like tariffs or something.
Don't tariff the next generation, bro.
You know,
there's going to be some shifts.
I think we're going to see
a lot more self-taught individuals.
I think we're going to see a lot more self-taught individuals. I think we're going to see,
I think we might see more,
we might see more high caliber products come out of smaller teams than we are
used to.
And I also think we're going to see different kinds of teams that we might
not be used to either.
Yeah.
Do you feel like i think it's going to take maybe 30 years for us to
get it back into some normalized territory where this change doesn't seem to be occurring minute
by minute um do you have any thoughts on that and what does that mean for entrepreneurs, business people, people that create?
So, I mean, a lot of it depends on how things go, right?
There are competing forces right now.
And, you know, they're all the same ones, right?
Like you've got the forces that seek to centralize things or the forces that seek to decentralize things.
And that might be viewed through a lens of we've got some big AI companies that are doing the big flagship models versus everybody else being like, hey, actually, I want my own one.
Or I want to get a bunch of friends together and pool our resources so we can have one that's over here.
Maybe that's for data security.
Maybe that's for data security maybe that's for you know just autonomy or whatever um and so those those sort of collective versus individual
type dynamics are sort of expressing themselves um do i think things will chill the hell out bro
probably not i think we're still i think if, it might go the other way to the extent that, I mean, everything normalizes, right?
I mean, the whole point of audio, and I say this as briefly an audio engineer, we are noise sensing machines.
And so when we receive a signal for a long enough time, it stops being signal and it starts to be noise and we start to filter it out.
That's why listening to a loud tone
at a specific frequency range can damage your hearing
over time, not because it's loud,
but because your body learns that it's a kind of noise.
And suddenly you can't hear birds anymore,
which is what happened to my uncle.
That's an oversimplification.
And so I think it might be similar
with some of the things that we're recognizing, right?
We're seeing a lot of new technologies
in terms of the rate of release
seems to be speeding up,
both from the big players
that are concentrating the resources
and the distributed communities
that are coming up with all these weird experiments.
And I don't think that's going to slow down.
I don't think it's going to taper.
But I do think it will express in different ways
across different domains, right?
So, you know, if you, for example,
figure out how to get peer-to-peer coordination,
which is a thing that I'm super into right now,
then you might not need as many legal agreements, right?
Because you don't need any
third parties. You can just go peer to peer, for example. You know, whether that's, you know,
like an Airbnb type model or an Uber type model. And so whilst it's nothing new, we've been able
to do that for some time. The dynamics of the tooling is new because even though we could do
it, it was kind of like a net negative to actually try to build one
for yourself and your friends.
But if your AI can just handle all the execution
of what you want to happen, well, suddenly, you know,
a lot of the tools that would be negative
are actually a net positive, right?
And so I think a lot of the platform economy
might see a shock over the next 20, 30 years.
And it's not so much about a turf war
as it is much about a dilution, right?
And so the example I give is...
And it actually goes back to that commoditization thing
we were talking about earlier, right?
It's just an extension of commoditization.
We come up with the new technology, and then it's all kind of finicky and specialist and expensive.
And then over time, somebody says, oh, well, instead of a serial bus, which we seem to have several thousand of what about a what about a universal
serial bus that just works with everything yeah yeah let's call it the universal serial bus we'll
call it usb yeah that way instead of having a toshiba sb and a samsung sb and a google sb we
can just have a usb and then you you know, you know how it goes.
And so, you know, I think those are the kinds of flavors of changes that we might see,
is a lot of sort of bespoke stuff might generalize.
And I think the more we generalize, the greater relative access people have to tooling, which, you know, hopefully means the greater relative agency that most people have most of the time.
Yeah.
I like that.
Yeah, I think that at some point maybe building something like a Facebook or X or whatever,
the barrier to entry might be so low that there could be all kinds.
And so I don't know if you know sierra but she talks about
proof of human all the time back to your security yeah is it and it's it's one it's past one o'clock
are you able to stick around what's your schedule like and flexible okay um i'm gonna tell everybody
you know if we don't have any questions, it might slow down.
If you have questions for Ruben, I highly recommend you grab the mic and raise your hand, everybody.
So we have some type of order in the room.
But I think it's a great opportunity to speak to somebody who has an eclectic and rounded background as Ruben.
I've not met anyone like you.
And I really, I think that mentorship is really important,
especially now the way things are going in terms of,
I think more people are coming together in groups
as opposed to just trying to do it on your own.
Getting back to what I said earlier, because I know I asked 10, 15 questions at once,
and it's kind of hard to follow.
What are your thoughts on Proof of Human?
It's hard. It's real hard.
It's so hard.
It's so hard that I am skipping it entirely on my side of the building stack.
I've been trying to build tools for people to declare terms and conditions
in ways that machines can understand and enforce.
So basically like having a pocket attorney, but one that talks to machines.
And of course, if you could do that, then you have to figure out law, and have to figure out jurisdiction and you have to figure out the juridical layer, which is a word I didn't even know before I started and so on, which is basically just the thou shalts kind of the personhood part.
There's a lot of different approaches.
We're seeing people using the iris, right,
body metric stuff.
I think that's a little dangerous
because I think a lot of the sensor equipment
is advancing very quickly.
And it seems like a weird idea
to put your password on the front of your face.
I'm just checking it out there.
Maybe I'm naive, but I feel like physically wearing my password where everyone can see it might not be the best idea.
But, you know, there's a bunch of different approaches, right?
I think, honestly, honestly, I think the least worst solve I can think of for personhood is simply just to not assume that it's a person just just assume that it's
not a person assume that it could be a person but don't assume that it definitely is a person
and let the behavior of that actor on a network um give you sort of uh a trust score if you will
um that you know, you know,
so, you know, I mean, because bots behave in,
well, they behave in bot-like ways, right? And so, if I
get, and the danger of just, like,
giving somebody a green tick
and assuming that they're going to be cool
is that,
well, at any given point, somebody that might
have been a human can just hand over
the keys to the machine and now
if the system still recognizes it as a human but it's behaving like a machine you know that's a
problem right um and so the only way i can think about it is to just um build a reputational layer
that isn't just like oh i trust steve oh i trust mary but is like, oh, these 10 people trust Steve, and those 19 people trust Mary.
Oh, and by the way, these 17 people don't trust either of them. And here's all of the history of
record of, like, why those reasons might be, whether it's like an oral history, like, Dave
took my cookie, or it's a sort of,ifiable history, like Dave definitely was here
or Dave definitely was not here at this particular time.
And so it's a little sort of surveillance capital sounding
when I articulate it like that,
but I think we can keep our privacy
and have reputational graphs.
I think we can have our cake and eat it too,
but it's going to take a long time,
and we're going to have to switch around some of those assumptions. I no longer think we can have our cake and eat it too, but it's going to take a long time. And we're going to have to switch.
We're going to have to switch around some of those assumptions.
I no longer think we can assume somebody is definitely a human or definitely not a human from data alone.
I think that we have to switch up our how we're evaluating trust in the moment.
And we have the tools for it, but it's going to take a while to get around. Yeah. And, you know, I like the all in podcast
and what Freeberg said that, you know, it's helped him with his business because people
immediately feel they know him, they trust him. And part of the reason that I do this is I feel
that there is some currency in showing up on a regular basis, people really get to know you. It would, I mean, um, there's
things that I've said that people then, if it was a bot, good test, maybe a bot would be able to
answer it, but I'm pretty, you know, it helps to be a little weird or something. Do you think
there's any, um, any, any, uh, truth in what I'm saying, or I'm in left field, just hoping that,
because I, yeah, I, I, I buy into things field just hoping that because I yeah I I buy into
things a little bit too much it's hard for me to assume I don't know anything about certain things
and I'm probably way too uh you know the glass is all full uh sometimes I gotcha uh I remember
a rapper once said you can get lost in the, but if you ain't got the sauce, you're lost.
Which I think is perhaps an opaque way of articulating that, you know, you got to have a little spice in there, right?
Like if we are so rigorous and we are so logical and so formulaic and linear, um,
that,
you know,
if you always do what you always done,
you always get what you always got.
Right.
And so I think we really do need a blend of experimentation.
and then over time,
you know,
with rigor and,
and,
standards and peer review and so on and so forth.
but,
but really,
you know,
I think the,
the best call me me old-fashioned, but I think one of the best
ways to balance things out is to just give everybody tools to do whatever the hell they want.
And that includes, but within the limitations of what everybody else wants, right? Because we want
to advocate for freedom to the individual, but to the extent
that those freedoms do not depend upon the freedoms of others to borrow from a bunch of
legal nerds from back in the day. And so, yeah, so I think that, you know, if you give everyone
a pocket knife, there's going to be some stab-ems. But you know what?
They might fix some shit.
And that might be useful.
So, you know, it's, you know,
when do you give your kid the keys to the car?
Right?
Depends on the kid, I guess.
And I think that's true,
not just for the village raising the child,
but also, you know, the nations raising the villages or the village raising the child but also um you know the nations raising
the villages or the the planets raising the nations yeah i mean i've always felt that it's
i'm getting a little philosophical but that it's good to have inefficiency it's good to
for people to fail it's good to have wiggle room where everything's not so buttoned up because that's where creativity
happens. There's always people that are going to come in and do something surprising, you know,
so long as they have an ability to do that and what they do. You know, as Lemmy Kilmeister says,
you know, you put on a flag and raise it up and see if someone salutes.
But that means you have to have a flagpole.
You have to have a village square.
It has to be seen to see whether or not anybody cares about it.
we continue to have at least the ability for people with however crazy their ideas are,
as long as they don't hurt anyone to kind of be seen by the public and see if there's
any quality to it or not.
Yeah.
And I think, you know, there's some cultures that are doing that a little cleaner than
others, right?
Like, I like a lot of what's going on in crypto land because first of all there's a
lot of speciation there's a lot of divergent diversity and divergence a lot of people arguing
with each other in crypto land in fact i would say the crypto land is characterized by people
arguing with each other which is kind of insane um but what are you going to do right they're all
they're all sort of out there on the fringes anyway. What do you expect?
Could we expect no?
Should we expect any less, right?
Because I think it's a net positive there.
But then you also see a lot of, you see different approaches, right?
And this is a little inside baseball,
but you'll understand the broader patterns here.
There's one called Cardano,
and I don't really hang out there a lot,
but they're all like ex-academic types, and you can sort of see it in how they work,
how they're building their languages, how they're building their governance structures.
It's very procedural, right? And I found one just the other day called Bitmap,
which is a different kind of thing. It's on top of Bitcoin and it's able to
sort of create these different kinds of database architectures on top of Bitcoin in a weird way.
And that group has no boss, ostensibly. It's just a bunch of folks that are working with
this primitive together in different ways. And so it's not as procedural and it's not
as rigorous and it's not as tested, but you get different kinds of folks turning up and you get
different approaches of building new tools, right? Which I think is really important if you're trying
to do new stuff. And so I think, um, the part of the innovation itself isn't just what we're doing. It's how we're doing it and who we're doing it with.
And I think that we're seeing more creativity in that part
than perhaps we ever have in history,
largely as a function of interconnectivity and interdependence
through the internets and associated technologies,
including the phone.
Phone was a big one.
Shout out to Alexander.
But yeah, I think we're only going to get more connected.
And I think in particular, I am curious what happens with psychology over the next couple of decades.
Because whether or not we develop a neural interface with Neuralink or other companies,
a lot of how we talk and how we speak in written text and in speech,
you can pick up a lot if you're a large language model with trillions of parameters, right?
And you can infer a lot about the individual, not necessarily just how they talk, but perhaps how they might think.
And that might sound a little creepy, but, and you know what?
It's kind of creepy.
of creepy. But if you, if you, once you get past the creepy part, and you may not, but if you do,
you might find like I did, that it can actually be a really useful tool. Because I can go to a
machine and say, hey, here's like five essays that I've written. Can you tell me the things that I do
when I'm talking that other people don't do? Maybe you could put them on like a graph or something, right? So explicit versus
opaque. Maybe we could do that from like a negative 10 to a positive 10. And, you know,
the degree of metaphors we might use, you know, plus 10, minus 10. Or if you want, we can use
plus duck minus duck. I'm kidding. That's a metaphor joke. You know, basically getting it to quantify whatever it is that it thinks that
makes you different, and just asking those questions based on your, you know, what you've
written. Is there a world where the creepy overlords are extracting this information and
getting into your brain so that they can take your candy? Maybe. Probably not. Maybe though.
so that they can take your candy?
Maybe.
Probably not.
Maybe, though.
Is it, you know, even if that was happening,
is it still useful right now?
Hell yes.
It's really useful.
I found, yeah.
You want to dip in there, Mike?
Yeah, well, I mean, I'm going to let you finish,
but, you know, Scoville, he was talking about this the other day.
And I just told my mother and father-in-law, you know, who are who are big time boomers.
And I'm like, you need to get over that. You need to let these things know everything about you.
And then I joked, I said, hell, China already knows everything about you already anyways.
But, you know, point is, is that, yeah, I think the more these things can understand how you think, you can see yourself in ways you never have before and then use a mirror of yourself.
I know I did.
So anyways, I don't know if I'm adding, Ruben, but, yeah, I think you're on spot or spot on with where you're heading.
yeah well I mean we still you know things are things are happening by the day right
Yeah, well, I mean, we still, you know, things are happening by the day, right?
if I were to like have any have I changed how I'm thinking about doing things before versus now
pre versus post one thing that I've been doing and you may want to do this yourself if you've
got old projects that you've given up, see if you can find the old works
and shove them through your favorite LLM.
And see if you can't answer questions
that you might have hit a wall on 10 years ago.
That's been an eye-opening for me.
I was working on some stuff.
I just hit a wall.
I just hit a wall in 2018.
And so I just mothballed it. And I recently put it back into Claude and GPT. And I'm like, hey, can you answer these questions that would
take me several months to answer? And it's like, yep, done. And so I can't even comprehend the
impacts of that. You know, what could Newton have done if he had GPT?
What could Einstein have come up with?
What could Tesla have done if they had GPT?
What could Pythagoras have done if they had GPT?
And I think that, I don't know, call me crazy,
but I don't think those people are unique.
I think there's a bunch of those people are unique. I think those people,
there's a bunch of those people in every generation and occasionally one of them gets lucky.
And we recognize those people as like the big ones,
but really if they hadn't, somebody else would have.
And it might've taken another 10 years,
but like, you know, humans, you know,
we're a curious bunch and we tend to figure things out, right?
And so what I'm curious about, and this could be, I may be proven right or wrong about this,
but I have a hunch, I have a hunch that the next generation, like the kids who grow up with it,
like my kid grew up with the iPad, as opposed to that kid's grandmother,
who once fixed the, uh, the,
the PC by getting in there with a screwdriver.
Thanks grandma.
just like the polar difference,
right?
Like the,
the first early tech kids,
like the cypherpunk sort of generation who just gets into the hardware,
like,
now here's your problem versus the kid who really can't get past the swiping of
the things,
you know,
like they're not even like mouse or keyboard yet, just polar differences. But I think I,
I suspect that if, you know, the Einsteins and the Newtons are like one in a thousand
or one in 10,000 and then like one in a hundred generations, one of them gets the cool cookie for society.
I suspect that those ratios are probably really similar today,
but the luck part is different
because the kinds of things that would have blocked somebody
from asking questions or getting answers about their questions
are no longer blockers.
And I'm interested to see what that looks like
once you give it to an entire generation.
Yeah, I think we're going to see population shifts
over the next decade, well, probably 50 to 100 years.
And that might put pressure on healthcare systems.
But I also think that AI will help
with a lot of healthcare systems as well.
And actually, that's about time
we gave it an upgrade anyway, right?
And so, you know, it's about time we gave it an upgrade anyway. Right. Um,
and so, you know, I, I am, am I, what's my, what's my thermometer on society? I'm actually,
I'm actually pretty, I'm pretty, I'd give it a, out, out, out forecast looks good. I wouldn't say clear. I wouldn't say sunny, but I would say stable and pretty healthy.
I think there's a lot of, I think there's a lot of, I think there's enough conflicting forces
to prevent like totalitarianism, whether it's in a technological sense or a nation-state sense,
which is cool, which I'm grateful for. It's not to say that it couldn't happen, it's just that it probably won't,
in my personal opinion. But I don't think it's going to be quiet. I think it's going to be bumpy and noisy and there's going to be all sorts of nonsense. Yeah. Is that my opinion?
Yeah, I think that, I mean, I always, I came up with this in one of my darkest hours, but
you must stay optimistic against all reason because the alternative has no future.
It doesn't mean that you go blindly into everything wishing and dreaming.
You got to be practical and keep your eyes open and street smart.
But to sit there and, you know, focus on the most dour in my experience doesn't lead to a good outcome. And who knows, maybe
there are multiverses and we're shifting timelines all the time based on what we think and believe.
And we jump from one cosmos or planet or parallel universe to the other based on what we're
attracting. But I mean, I think there's going to be some pretty wild understandings about like what
is reality especially when we get into the augmented reality with all these glasses
we can clone ourselves we can talk to ourselves um you know it's going to it's it's going to bring
up a lot of new stuff and and yeah i i i'm i'm happy that that like right now we're in a space we're
talking to people all over the world there's people here from all over the world in real time
having a discussion that you know unless this the space glitches out or something can continue and
then it's recorded this is um you know, it changes just like the smartphone for kids.
It changes their their the neural patterns in their brain.
And there's good and bad and everything.
I mean, fire was one of the greatest things that we ever got a hold of.
And then also it is the most destructive thing here on the planet.
So, Michael, I'll let you go.
I mean, I'll let you.
I just wanted to say that.
So you have something you want to say?
Yeah, I don't know if you want to wrap this up, John.
I'm just I'm spurring Reuben.
We talked before and so I'm enjoying this.
So, hey, Reuben.
So question.
Do you follow Moe Gaudot educate
me Moe's an ex Google executive I'm not sure Google DeepMind he was I can't remember the
exact title but he was in Google in 2017 and 16 and earlier. Yeah, formative.
He's got an interesting perspective.
So that's the reason I was asking,
because I kind of believe in some of what he's saying is that we're embarking upon a time of abundance.
I think that that is a general belief
among most of the futurists here.
I don't bet against Ray Kurzweil's curve of the future.
We're in an exponential growth factor,
and AI has the potential to change our lives and our world forever.
So that's the general underlying belief.
However, there's an interim.
There's that current state we're in now,
where as AI starts to replace things,
starts to create this world of abundance,
we are still an entire society
that is living in a world of scarcity.
And I don't think we know how to live
with a scarcity mindset in a world of abundance.
That was basically the statement.
And that's kind of Mo's position.
Yeah, I tend to agree.
I tend to agree.
The question is in the how, right?
You know, our friends, Mr. Marx and Lenin,
back in the day, had very strong opinions
about how things should be shared with one another.
And they came from a place of resource asymmetry, right?
But not everybody didn't work out how they planned per se either, right?
And so I think that one of the challenges that we face is if we had built our economy and culture on a sort of assumption of scarcity,
on a sort of assumption of scarcity.
And if we are no longer in a scarcity scenario,
thanks to technology and things like refrigeration
and, you know, whatever it was they got,
they figured out to get the phosphorites out of the air
so we could stop digging up, talk a lot.
Like all the new technologies, right?
Agricultural technology and so on.
If we are in a sort of post-scarcity scenario from the resource base but we are in a
scarcity defined economic and trade infrastructure then we may have a problem because the only way
for people who uh organizations that rely on scarcity to maintain their position or power or influence is to perpetuate the scarcity.
And if the resource in question is their property, their legal property, then you got to make a call.
Either take it from them, like the Bolsheviks did back in the, you know, or the Haitians or the French or the English or the Americans for that matter, right?
Just to be real.
And the Kiwis, I should say.
A lot of the Kiwis.
Or you negotiate over time and you build other systems and you, like you know, maybe, maybe you don't take their
stuff. Maybe you build your own stuff, right? Like there's all sorts of ways to go about
it. But, but it does really take time because you're talking about feeding 8 billion people
and figuring out how to get them everything that you find in Maslow's hierarchy. I do
think that we are, to the vista of my understanding, we've got about 8 billion people and we're making about 14
billion people's worth of food, but only
6 of the billions are getting it, which seems inefficient
to me. I don't think we should seize the means of
production. That seems like a pretty blunt force move, but we
are going to have to figure out how to work together better.
And that might take some time because it's not just technology.
It's people.
Yeah.
I read Gerald Diamond's book Collapse.
And, you know, it goes through all these various cultures that had, you know, massive population collapse.
And, you know, there were these five elements that he found.
But one thing that's interesting is all these were at the zenith of their powers.
And it just seems that, you know, I mean, I always say, you know, we are in Eden.
It's just managed poorly.
But I have faith.
We've gotten through a bottleneck where we've got down to what, 250 or 25,000 people.
We seem to kind of keep on trudging along.
So and and I think it's it's this is a time where people have to take responsibility.
You know, you can't look to others to solve your problems and put, you know, people in some kind of category that, you know, you're going to get everything solved by one or small group.
It's, you know, I mean, we are all responsible to a certain degree for things that go on, whether we're even conscious of it or not. So hopefully, I mean, the crypto space,
it's such a wild west and so crazy and so unmanaged
that it might be the holy grail that gets us through this.
Do you have any thoughts on that, Ruben?
And then if you want, I think,
unless some people got some hands and some questions,
we'll go on a little bit.
But if there is no more, I think
we're going to wind down a little bit. But do you think the Web3 space might come up with a lot of
answers to these questions? I think Web3 is the least worst path that I've seen to solving a lot
of governance problems, actually. It's good good at money but it's so much better at
governance and that's a weird thing to say in the crypto community because most people associate
crypto with just money and yeah sure it's it's currency but like it's it's also identity right
through the wallet it's also governance through know, smart contracts in a way.
It's definitely coordination.
And I think the coordination is the common theme, right?
It's been resonating with an old statement by a dead Prussian general dude called Karl von Klausowicz,
who wrote a book called On War, and it was the second treatise on war since
Sun Tzu and Carl
said war
is a continuation of politics
by other means
which is
pretty edgy thing for someone to say but it's true
it's been occurring to me that
if you look at it in a certain light
both politics and war
are a form of coordination
everybody's not always happy all the time
but I think
I think the challenges of the next century
are less likely to be resource challenges and more likely to be coordination challenges.
And so if we can keep our eye on the prize and use AI for what it's good at and use peer-to-peer technologies like cryptography and cryptocurrencies in general.
And, you know, I'm playing it down.
I think crypto is going to be a big deal.
I do.
I really do think so. I don't think it's going to be a big deal in the way that a lot of people are treating it right now, though. I don't think it's going to be like number go up as much as it is like, oh, we can just do the things that we're already doing. And in a way that's safer and more transparent, which is which is pretty boring, which is which is signal. Good businesses is abhorring, right? Yeah.
I mean, I think that people are talking about a vacuum in leadership, and vacuums always get
filled. Absolutely. And so do you think that those that may be unassuming leaders?
Where do you think leadership's going to come from?
During this time, it's going to get crazier and crazier,
and are we going to have leaders,
or is it going to be groups of cohorts?
Yes and yes.
I think it's going to be the whole spectrum, man.
And I don't know if that's ever going to change
I don't see
a utopia or a dystopia
I do see a
mish happening
where you've got a bunch of people
who want to do this thing in this way
and maybe that's for religious
reasons or
ethnographic reasons
or, you know,
other kinds of
reasons, economic reasons, certainly.
they're going to want to do it in different ways
and people are going to be able to, you know,
operate in certain ways with
certain degrees of authority and jurisdiction,
if you will. And then
other people will do things a different way.
And they definitely won't want to do it the same way as the other people.
And they will have conflicts.
And then other people will build bridges in different ways between those communities.
And which is basically a continuation of everything that we've already done throughout history.
Perhaps with better tools, right?
And, you know, if we make those tools a little bit more
modular and, you know, composable, right, like Lego, then, you know, maybe we can come up with
bounded structures, right? So instead of relying on this party to make all of the decisions about
all of the important things, maybe we can split it up a little bit.
Right. You know, and America did that with, you know, the branches of the government.
Right. The, you know, the courts versus the executive branch and versus the legislative branch and so on.
Separation of powers. I think it's a great idea to separate powers.
Often, often when things get weird, it's because they forgot to separate the powers, right?
And so I think that we might be able to do that not just in a sort of institutional sense,
but perhaps really like in a hyper-contextual, hyper-localized way,
which is less of a top-down or a bottom-up necessarily,
and it's more like just a mesh of different things
that are working together in different ways
with degrees of efficiency,
depending as to where you're going to
and where you're coming from, right?
And some of that's going to be a bunch of machines
doing the conversations,
possibly human-to-machine,
possibly machine-to-machine,
certainly human-to-human.
And yeah, it's hard to say how we'll
express, but I do feel like we are getting closer to an environment where we can have
common schemas for how we do things together. And I think the stronger those systems are,
right, whether it's an illegal legal sense or a, you know,
technological system sense or a cultural sense, I think the stronger those bridges are,
the more resilient the overall mesh gets, right? And people can sort of step in where they're
needed. But some of the challenges that we face are there are those who benefit not from cohesion
but from division.
I won't drop any names there because I'm being a hippie over here.
I'm trying to sit in the middle.
I'm sure everybody has their own opinions about who's to visit and who isn't.
I think the sooner we learn
how to recognize that as a culture and how to build tools that encourage and inspire and incentivize working together, right?
And also, you know, looking for yourself as well.
We don't want to subjugate the individual for the collective.
But I think we can have collective outcomes and individual rights.
I know that sounds crazy, but I think we can have collective outcomes and individual rights. I know that
sounds crazy, but I think it's possible. We just need really, we just need really good tools.
I mean, I think going back to the music conversation in the beginning, like harmony
is a big deal. When things are out of harmony in terms of music, you know it right away.
terms of music you know it right away and um you know to to try to focus on a harmonious uh uh
experience with uh within a group is something that you don't hear talked about too much um but
i think that uh it's it's uh really important and you know maybe there, maybe instead of leaders, finally things will be different and it'll be ideas that will lead us forward instead of individuals pushing ideas.
So who knows?
I don't see any hands up, Ruben.
Is there any, are there, would you be, so where are you right now in terms of, are you think you're
going to stay in the States for a while?
Do you, are you looking for opportunities?
Are you just kind of enjoying the playground of the LLM so we can kind of, you know, look
for the landing field for this, our conversation today?
What's my deal, man?
That's a good
question um yeah i have been working on a hard problem for a long time i've been trying to figure
out how to take all the things that we do with law and to do them with computers instead uh
yeah not instead uh to do them with computers as well well so that we can take away all the complexity
that we might have if we wanted to write a contract for ourselves
and instead we could just point and click
and maybe borrow a template off somebody else
and if you don't like it, you can fork it or whatever.
But I think one of the imbalances that I see
that I think I can help with is there's an asymmetry between people who know how to articulate their terms and conditions and people who don't know how to articulate their terms and conditions.
Because it imbalances negotiations between parties, right?
You want to make sure that if there's two parties hanging out trying to make a deal with each other, that one of them them isn't goliath and one of them isn't david right um actually probably for example but like
they played a different kind of way but you know it's supposed to be like a meeting of peers um
and i think historically we've trusted third-party organizations to do well to deal with that
complexity whether it be your council or whether it be,
you know, Airbnb for figuring out the trust and security part or whether it's Uber for the
pricing part. Surprise. Sometimes their opinion on price is different to my opinion on price.
God damn it. But, you know, I think if we build better tools for consensus and for coordination,
tools for consensus and for coordination, then, you know, I think that's where a lot
of interesting stuff is going to come out over the next decade or so.
And so what I've been trying to do is sort of, what I've been trying to do, man, honestly,
is I've been trying to take 50 billion contracts every 24 hours,
which is about what we're doing, right?
Not just like your rent and your car,
but all the terms and conditions and the license agreements and so on.
And it appears as though all the 50 billion contracts
are really just like 10,000 boilerplate contracts,
just like jiggered around.
And if you look at those 10,000 boilerplate contracts just like jiggered around and if you look at those 10,000 boilerplate contracts
there's really only about
like 100 to 200
like distinct
legal Lego parts
that pretty much make up
all of those 10,000 contracts
which make up all of the 50 billion contracts
so it's kind of like
trying to reverse engineer
like if law was a language in and of itself,
like a functional language,
what would the alphabet look like, right?
So I've kind of been backing into the alphabet
and trying to figure out,
okay, well, let's just deal with the NDA today.
How do we build an NDA that machines can read and write
and we can put into code that works
without overwhelming grandma?
And then how do we do that with this other part, right?
How do we do it with, like, if you buy this,
then you also agree to give me feedback afterwards
and the feedback is in this way, for example.
Like, just taking that and making that
machine rewritable and machine
enforceable, that could be
used for a million things, but
if you try to build
the whole stack every time, you're
going to get 50 billion documents. That's how
we got law.
And so I think if we sort of
almost decompile
what we were trying to do with this legal thing,
and we sort of refactor it so that it does all the same thing, still compliant in the relevant jurisdictions,
there's 192 of them.
Actually, there's a lot more than that, but, you know, states and so on.
But I think if we can do that, then that's going to be useful.
And I'm hoping to get something out the door inside of six months.
So probably inside of, possibly inside of a few weeks.
But I always like to get a little patting in the back.
I see we've got a Wes up on stage.
How are we doing, Wes?
I'm good.
Can you all hear me all right?
Yes, sir?
Yes.
Awesome.
Yeah, I have a question.
I'd love to get your thoughts, Ruben, so I can give a quick intro and then a question Ruben and John can chime in, too.
By all means, yeah. Sounds good. uh software engineer aws um about four years now before that various startups uh it's kind of odd i i spoke back then i swore to myself i'd never work for the man or big tech and uh here we are
um anyway so i kind of work on um mcp servers um and kind of add those to kind of our offerings. I'm so happy to talk about that too.
But to my question, I saw this video two weeks ago on YC,
and I thought it was really interesting,
and I'd love to kind of get your take.
And it's essentially that, so Gary and Jared from Y Combinator,
they had this roundtable talk on their YouTube channel,
and they talked about how
the lean startup is dead. And the lean startup being, you know, this idea of finding product
market fit first, even if it's not scalable, you know, looking for customers and really kind of
honing that customer experience. And they said, well, you know, you can, you can scalp market value and you can have good
success by just using the thing, adding agents to your existing things.
And, and you're going to be fine with that.
So if you just use interesting tech, you'll succeed and, and no longer have to do the
hard thing, the, the finding customers and finding product market fit.
So yeah, I thought that was fascinating
and could be controversial.
So just wanted to-
Yeah, absolutely.
Hey, John, what was I saying earlier
about good businesses?
Boring.
Good businesses are boring.
Look, don't get me wrong.
Don't get me wrong.
There are things that we can do with LLM agents and we can scale in ways
that we can't
we just can't with humans
we can do automated testing now
we can do
scenario role modeling now
we can do things like
round table
customer development processes we can do with simulated humans
with degrees of accuracy, which is not 100%, by the way.
Chad GPT's ability to pretend to be Terry Pratchett is a lot better than his ability to talk like Samuel L. Jackson
Samuel L. Jackson, because Samuel L. Jackson tends to say one thing always.
because Samuel L. Jackson tends to say one thing always,
And so it's very hard for the machine to pretend to be Samuel L. Jackson without saying that
one thing again and again.
It's contextual, but, you know, there is a...
We are getting greater relative agency.
I think we're going to see a lot of...
I think we're going to see more solo founder types.
I think we're going to see a hell of a I think we're going to see more solo founder types. I think we're going to see a hell of a lot more, um, like small teams in
general doing things that you wouldn't expect to come from a small team. Um, but I also think
we're going to find these weird meshes that don't fit with our current standards of what a company
is or what a team is either. Um, and it's, it's hard to articulate, but, um, you, you, you sort
of seeing it in its formative stages
on platforms like discord and again in the crypto space we're seeing these little sort of
loose associations and groups and communities and meshes and you know people pick up hammers and
other people pick up spanners and then somebody turns up and gets both of them high right like
like it's it's gonna be i think it's going to be a lot of chaos.
But I do think that, you know,
experimentation with the how
as much as the what
is going to be where it's at.
And I'm excited to see
what comes down the pipeline.
I don't think the lean startup is dead at all,
in principle.
I really don't.
I really think you can only model so much
if nothing else
then the fact that society is faster in general
and what was true yesterday may not be true tomorrow
right
and so you can predict what you want
but like
I don't think
if you're an economics nerd
you might have read a book
by a dude called Neil
not Neil, Nassim Taleb.
He wrote a book called The Black Swan.
Yeah, I love that book.
And he wrote another one called Anti-Fragile, which is a banger.
But he talks about black swan events where people follow the trade lines,
and then suddenly something new turns up, and it doesn't look like anything that we've seen before and it behaves in different ways.
And suddenly all our prediction models are off. Right.
And he used the example of Uber and taxis where, you know, if you wanted a taxi medallion in New York, you had to pay millions of bucks just to drive a cab because there was a regulation saying only people with one of these can drive a cab.
And suddenly Uber turns up and, well, actually, they're not taxis.
They are ride-sharing.
And as you can see in the law, ride-sharing is different, and it's totally legal.
And if you want to ban ride-sharing, you've got to ban carpools.
Good luck banning carpools, you know?
So it's...
Copy that.
Yes, sir. All right. Sorry about that. you know, so it's copy that yes sir, alright
sorry about that
yeah, I think
things are evolving, I think
to the extent Lean Startup
is iterative development
I think we're going to
see some iterations, I don't think iterations are going away
the shapes are definitely going to change.
And I think there are going to be a lot more
mature products that
we would have expected to turn up
very immature. I think we're going to
see a lot more polish on some of the new stuff
coming out, which would be interesting as well, in terms of
how that interplays.
Does that sort of
resonate there, Wes? or what do you reckon?
That's perfect.
And to add to one data point about the change of teams, we're kind of seeing this already
internally.
Traditionally, our teams, we call them two pizza teams.
So your team should only be as large to feed with two pizzas.
And yeah, I think there's going to be a shift it seems like the future of the software engineering team is actually going
to end up being one engineer and one product manager and the product manager is actually
the only one writing code the engineer is basically being like a quality control, making sure the architecture, sound, security, encryption.
But the product manager can essentially iterate as fast as he can vibe code.
So I think it's going to be super interesting in the next few years.
But anyways.
So I totally agree.
Here's my question.
Let's double click on this for a second because you're spot on.
So what do you think the other engineers do after they leave the company?
You said after they leave the company or the team?
I assume that if we're going to have one engineer and one product manager,
I assume that some people are going to have some free time on their hands, right?
My hunch is that they start new things.
And they start their own PM slash engineer combo.
And I have one data point on this.
In the first year of COVID, EIN registration doubled. For those unfamiliar or
perhaps out of town, EIN is the employment identification number. It's the equivalent
to your tax code, right? So if you're American, you have a social security number, that's your
tax code. If you're a company, you have an EIN. And the rate at which EINs were being registered,
that is the rate at which people were starting new companies, doubled in the first year of COVID,
right after a lot of people got laid off.
And if we see a similar shock to labor, which we may,
I don't think it goes away,
but I do think it changes the landscape.
I think that, you know, if Google fires half its staff,
it might have 20 times as many competitors as it did yesterday, you know if Google fires half its staff it might have 20 times as many competitors as it did
yesterday you know and so I think some of those are gonna like be bounding factors you know we
may end up in a in a world where machines are taking care of everything but I don't think it
is I don't think the conclusion is that
all those machines are owned by one person, per se.
But I do think we can take the hard parts
out of being a human and get machines to help us out
and figure out a way to not be dicks each other as well.
But it very quickly turns up,
not just in the tech,
but also the agreements and the coordination,
the politics and private property.
And Maslow's hierarchy kicks in pretty quick
in terms of food and shelter and so on.
So you can only get so concentrated, right,
before things get weird.
Ask the French.
But yeah, it'll be an interesting,
it'll be a wild ride, man.
It'll be really interesting, yeah.
It's going to be fun. Thanks for the answering my question.
Yeah, no worries, man. Good question.
Just on what you were saying,
I did some research in California and something like 40 or 50 or 50 plus
percent of the businesses in California are one person.
So I'm not sure exactly if everybody's a consultant or I don't know exactly what that meant.
Look, you know what?
Full disclosure, I'm cheating a little bit with that double number.
I also know that about 20% of that was Uber who forced all of their workers to be independent contractors
because they didn't want to pay them any, they didn't want them to have labor predictions.
So yeah, sure. It jumped. It wasn't double, but it was close. But I really do think though,
that we're going to see more people, you know, cycling out of institutions and cycling into owner-operators and small groups doing cool shit, for lack of a better term.
Yeah, and time will tell, right?
Okay, here's a good question.
If this prediction was right, how would we measure it?
Would we measure EINs?
Would we measure average headcount probably a
headcount of the company right i i think that it has to be you know um just just because you have
an ein doesn't mean that you're doing anything you could just file it and be sitting in your room all
day um but if those are active i mean i believe it's a lot easier to sell to a business now than to maybe get through the hoops of getting employed.
So hopefully what will happen is the laws will start to reflect that reality.
And if the laws start to reflect the reality where everybody, where the majority of people, you know, forget about working for 50 years and get a pension, that's gone.
But that the people are out there solving their own problems and coming up with solutions that other people will pay for, that's going to change things dramatically, I would imagine.
Yeah, yeah, law is a tough one because it's a very solid, it's designed to be slow, right?
You don't want somebody getting, let's say, in an executive role with executive privileges,
just switching up the game, right, too quickly.
switching up the game, right, too quickly. But I imagine that the shape of law may be
less legislation and more contracts. And so, yeah, and it's not to say that legislation
still won't be a thing, right? Like, you can't, if dad doesn't turn up to pick the kid at 5 p.m., the kid doesn't sue dad, right?
There's no breach of contract.
That's a different kind of law.
It's juridical stuff, right?
You've got the parent and the child, and they are necessarily tied to each other.
They're not actors in a free market.
And so there are some parts of law that will have to be dealt with procedurally
rather than cryptographically, or at least cryptographically alone. And that may take
a long time because there will need to be shifts, right? And law is not designed to
shift well, right? I think we're seeing it with copyright now, where suddenly,
copyright is jaywalking, basically, at this point. Jaywalking is illegal, you're not allowed to do it,
everybody does it, nobody stops you. Copyright infringement is illegal, you're not supposed to do it. A lot of people are doing it, nobody's stopping them. And I think we're going to see,
And I think we're going to see, you know, when the ability to to infringe upon the law is so easy that the enforcement mechanism to prevent people from doing that cannot possibly keep up, then the law breaks before the behavior does.
And it doesn't happen often.
Jaywalking has been around for a minute, but the copyright thing is actually pretty recent.
You know, it started with Pirate Bay and so on,
but it's sort of suddenly got a hell of a lot more hardcore
post-generative technologies, for sure.
And so we'll see how labor laws develop.
Labor laws are particularly sticky
because you're dealing with every member of society.
You're dealing with half a nation, for a start.
About 320 million Americans, 160 of them are working from every walk of life.
And so making rules for that many people in a way that's enforceable and durable
and by consent, not mandate, is going to be challenging.
But if we are able to build tools for us to work with each other
that produce the same results in terms of security and confidence and predictability
that we might have historically have relied upon a government to produce, then we might
not only reduce pressure on the government, but we might be able to work together in a
more, in a way that is defined by the parties,
the acting parties, as opposed to the governing body, right?
And, you know, we see that happening right now
with things like waivers, right?
Where if I go to a festival,
they get me to sign a document up front.
And the document says,
hey, if you punch yourself in the face, we're not liable.
And that's kind of a weird
thing to say when somebody walks into your house. But like, well, I mean, what happened was people
tripped over some stuff and they punched themselves in the face. And now we're going to have a law
about it. So, you know, it's those kinds of events. Somebody once described policy as a form of
behavioral scar tissue. And I think we've got a lot of scars.
And I don't think it makes sense to throw out all the scars
because we got them for a reason.
But I do think it makes sense for us to examine
to see if we can't, you know,
chill the hell out on some of the restrictions
and hopefully open up agency for people
without restricting their freedoms.
Or the freedoms of other people.
Wes, were you going to say something?
I was going to raise my hand, but you actually somehow predicted I was going to, which is very nice.
So this just kind of came to me, Ruben, I was wondering if you had any thoughts on this.
So this just kind of came to me.
Ruben, I was wondering if you had any thoughts on this.
I recently saw how pervasive forced arbitrations are.
And us as consumers, you know, we have an Android or an iPhone.
We've signed on to, you know, probably a dozen forced arbitrations.
All the products we use in our house, you know, probably adds up to hundreds.
With what you're kind of talking about with LLMs and the law, is there opportunity there for at least to get a better sense of like, you know, what's our exposure?
So picture, if you will, I have a local LLM and I say, hey, can you go through my email addresses and just count how many terms of
services I've agreed to thanks oh this many awesome hey can you figure out which ones
have access to my data oh cool hey can you find all of the other or competing services
that don't take my data awesome hey can you just quickly switch out all of the services
and products that I currently use that do the thing that I don't like and just find alternatives
to those versions that are as good or better, but don't take my things? Can you do that for me, LLM?
Sure. I'm not saying that that's a reality in all circumstances. PG&E is probably going to have some opinions about where I get my
gas and electricity from. Something something Enron. But you know it's contextual and I think
that law like any form of code is neutral right? It can cut both ways. And so if we use the
arbitration clause example and for those unfamiliar there's a law in America that says that you have rights to go to the courts and speak to the people that sometimes wear wigs and say, I have a problem.
And they'll say, oh, what's your problem?
And they talk about it and they figure something out.
And that's you have a right as an individual just by being here.
And that's you have a right as an individual just by being here.
An arbitration clause says, hey, as a part of this contract, we agree that if one of us has a problem, if you have a problem, then instead of going to the courts, we are instead going to go to an arbitration, which is a private organization.
an arbitration, which is a private organization.
There's a few big ones, AAA or JAMS,
which are usually retired judges or fancy attorneys
that put themselves out to pasture kind of thing.
And super smart people.
But they'll look at your case and say,
hey, don't take this to the courts.
I'll be your designated neutral. I'll try to figure it out.
If you can't figure it out here, then maybe you can go to the courts after that.
And originally it was designed, it was sort of pitched to the courts as,
hey, here's a way where we can take all of this noise of the courts with all this BS nonsense
and we can put it into this, you know, attachment to the legal system, which is not law.
But, I mean, it's not like legislative law.
It's like contract law.
But if we all agree that these people are cool enough to figure it out, then maybe we can give the court a break, right?
right um the reality is is that it was being used as a divide and conquer strategy where
if you are a large company like an AT&T or an Uber and Uber has received a lot of these
um they have more lawsuits against them than just about any other company we've ever seen it's wild
uh across the board too like civil and yeah it's gross um but uh gross at all costs right um but uh yeah if if a large
company has let's say 5 000 disgruntled customers well that's it's a big problem if if if an attorney
could turn up and represent them a class action that might be a that might be a 5 000 person
problem but if we force them all to go through this little nonsense arbitration thing,
then we can isolate these 5,000 fires.
And instead of being one big 5,000-person fires,
we now have 5,000 little small fires,
which would be a problem except that we've made it so annoying and clunky
that most people never actually turn up, right?
And so instead of dealing with a 5,000-person fire, we're dealing with like, you know, 100 fires,
which are sort of all separate.
We can like deal with them and pay them out if we lose, you know, beat them up if we win or whatever, right?
But the sword cuts both ways because hypothetically, let's say that somebody built a browser extension.
And that browser extension lives in your Chrome or your Mozilla, Firefox, or whatever.
And it just counts all of the arbitration agreements that you have and measures all the products that you use.
And it's all kept privately, locally, so nobody can stick your stuff.
it privately locally so nobody can stick your stuff. And if you have a problem or if you think
that a company has been not cool to you, then you can click a button and say, I have a complaint.
And that browser feeds into a mesh network in the back where if over, say, a thousand people
have the same kind of problem with the same company, then it triggers a litigation funding contract which pays for
the arbitration process and files all of the lawsuits at the same time. So what that means is
that suddenly it's free for your customers to beat you up. Well, not to beat you up,
let me be fair. It's free for your customers to complain in a legal setting with third-party awareness, like accountability.
It's free for accountability now.
And now, instead of having one fire that's 5,000 people large, you are right back to having 5,000 individual fires.
And now they're all happening at the same time.
It's a nightmare.
What was started as divide and conquer is now an actually heal right and i don't i'm not going to build that but i'm pretty sure
someone is i don't know who but it seems inevitable um but like the arbitration clause in particular
which has been contentious for decades california hates it um you knowC. is okay with it. Florida loves it. The banking sector in particular is very enthusiastic about forced arbitration.
And in fact, because that's what hit Wells Fargo, right? And nobody wants that for themselves. So, yeah, it's an ongoing conversation in the courts. Somebody once said,
you know, legislation Fs around, litigation finds out, which I think is an interesting way to view
that. But yeah, I think a lot of the progress will be as much through the courts as it is through legislative efforts.
I'm already thinking about the technical design of this Chrome extension.
Maybe we should just, all of us just do a vibe coding session this weekend and, you know, have a launch next week.
You know, it's funny.
I have been considering setting up like a legal tech uh hackathon uh i don't know what city yet um i used to run hackathons back in the day and i
reckon a lot of interesting folks will pop out of the woodworks yeah
well cool thanks for the context yeah i just i saw the uh i just saw this recently and it was
just kind of like crazy numbers like you, you know, only 6% of consumers win arbitration and the judge or I guess whoever presides over it doesn't actually have to have had a law degree and like some kind of insane rules.
And I hadn't realized that all of us have probably signed on to hundreds, if not thousands of forced arbitration clauses from the products we use.
Yeah. Yeah. Well, I mean, that's the thing, right?
Is that law is gameable almost by definition.
And so, and it's always, it's always lagging, right?
It's, it's laws reactive.
Something happens, there's a new problem and somebody makes a law about it and then it takes a year and then maybe we can enforce it
like three years later uh and yeah so it's it's challenging um and you know legal notice right
that is getting somebody to click a button is not the same as the spirit of notice,
which is to inform the party of what it is that they're doing
when they do the thing, right?
If I put a contract term in that said,
hey, if you use my app, I get to eat your babies.
Fortunately, there are rules in place to prevent me from,
to discourage me from eating your baby.
But the contract,
like nobody would bat an eye at the contract.
No one's written that contract.
It's not really notice.
It's the appearance of notice.
And yeah, and I think LMs are going to be a lot better
at recognizing what kind of agreements we really are entering into and what kind
of alternative options we might have, which I think should balance those scales.
If not through legislation, then perhaps through market behavior.
Actually, that sounded real smart.
I'm going to give it up there.
Try to go be an adult while I still can.
Is that, yeah, anything else while we're here?
John, you reckon?
I don't see any requests for anyone to speak.
I really appreciate your time, Ruben.
Is there anything that you would like to part?
Any thoughts that you'd like to leave everybody with?
Words of encouragement, how to handle uncertainty, a little bit of wisdom?
Yeah.
I'll give you the only thing that I can think of that makes sense now that didn't used to make sense that might be useful, which is if you are trying to do a big thing, see if there's a way that you can chop it up and do smaller bits of it instead, because the world's changing quickly.
And if you are working on like a level 10 problem and it might take you five years to do,
you might find that by the time you get four years in, your level 10 problem is like a level 2 problem.
But if you chop it up into little tiny problems, then not only are you getting stuff out,
but if things change, you can pivot along the way, right?
And so the value of like doing big, long work, deep work,
versus short, snappy things is shifting a little bit.
And I think it lends itself to more almost like a lean starter.
Which Witter did, so ignore me entirely.
But I do think that this sort of execution cycles
are going to speed up a little bit
in the decision-making cycles.
And I think we're going to see a lot of
sort of scrap experimentation.
But not saying that you shouldn't work on big stuff
or that you shouldn't work on deep stuff,
but to the extent possible,
if you can break it down in a way
where you can knock stuff out in smaller pieces, it preserves your optionality over time and it builds credibility through the execution of actually making stuff happen.
This is me giving other people advice that I fail to follow.
So, you know, grain of salt.
So, you know, grain of salt.
Yeah, that's the I.
Yeah, I've kind of, you know, concerned in the sense of these things are moving so fast with LLMs.
There's a lot of people who said, well, just, you know, do whatever you can to get it out there as fast as possible, because that's the, you know, the name main uh uh differentiator at this point and i've kind of
seen a little bit out there i don't know if you have any thoughts on that i know i gave you a last
word but just it brought up one other thing we're referring to the question one more time um that
it's that it's important to to be speed is important in terms of getting things out, but also the, the idea of I'm working on something will, can an, and we're talking about this earlier of just assuming
that I don't know the answers is even more powerful now that we have LLMs because, uh,
I'll, I'll hit a wall and it's like, Oh, you need to like build a Python script to do the
blah, blah, blah, which I don't know how to do.
Um, I go as far as whip hooks, hooks. More of a nurse than a doctor.
But, you know, I can go to the machine and say,
hey, what does that mean?
Can you break that down for me?
Pretend, you know, I'm a six-year-old.
How do I put something into a, quote, recycle bin, right?
Like, whatever this, like, don't be afraid to sound dumb
to the ai because guess what you are uh and so just just roll with it you know and uh and it
very much wants to help you figure out whatever it is that you want to figure out it's very very
patient it's infinite patience uh if you're a, you may be familiar with the concept of walking the room,
which is perhaps in response to an unresponsive crowd. You sort of go rogue and try to be as
offensive as possible and see how many people you can get to walk the room, right? That's called
walking the room. If you ever see a comedian going rogue and everybody booing, they could be
trying to walk the room. Walk the room. They'll love you for it.
But you can't do that with an AI because AI has infinite patience
for all of your dad jokes.
I could tell infinite dad jokes
and it would groan every time
and it would smile and look for more.
So leverage that, right?
Because, you know,
I remember when I was 25,
I got married to a lady who didn't take any of my BS, which of course is perfect for me.
And yeah, I would tell her about all my ideas.
And about a year in, I realized that I'm kind of treating this person like an ideological whetstone.
an ideological whetstone.
And that kind of sucks, you know?
You know, like,
not everybody thinks about the same things
and is working on the same problems.
And if you're leaning on people
with these really hard questions and complexes,
it can fatigue.
So, but finally, we have this tool
which isn't a mind in the traditional sense,
but it can feel like a mind.
And you can test ideas
and you can ask really hard questions
and you can do deep work and preserve the moments
that we have with each other for things that are more mutual
rather than just sort of opportunities to sort of sharpen the mind
or sharpen whatever it is that you're trying to sharpen, I guess.
And so, yeah, don't be afraid.
It's the let me Google that for you 2.0 if you will
right whereas um you know i mean everybody knows let me google that for you but we're basically
getting to the point where it's like let me just ask gpt or claude claude in particular i like
because it's a strong narrative strong writer writer, strong editor, medium writer, very strong editor.
But yeah.
Excellent. Well, thank you, Ruben, so much for taking your time out to talk with us.
Please let us know how your project goes.
You always have an open invitation here at the AI biz hour with Andy and John to
discuss anything you'd like. I got a huge amount out of today's show. I hope that everybody else
did. I know everybody is anybody who's listening that's got half a brain has gotten a lot out of
it. And for those who are listening to us in Spotify, please tell your friends about our show. If you made it to the end,
then you're a real diehard fan. And we are on X if you want to find us to come in and ask questions
directly. This is a pretty unique space in the sense that it's focused on AI. We've got real
people who are really doing it in real time in the world at all levels of expertise.
And people come in and they share what they've learned.
They share what they're doing.
And I think that at this time, while everything's moving so fast, it's really important not just to document day to day what's going on,
but also to have a place where you could come in and ask questions and share what you're doing. It's a big, big, big thing that is a one-time event in the history of the human experience.
And we should all pay attention to what's going on.
And if we can, slightly influence its direction for the betterment of ourselves and the people that we care about.
for the betterment of ourselves
and the people that we care about.
So I think that it's really nice
to see people coming back here again and again.
I'm always humbled by the room.
There's, you know, this space is only as good
as the people that show up and share.
And there's always great people showing up here
again and again and again.
I see Dallas has asked for the mic.
Dallas, I'm going to give you the last word.
If it's a question, I don't want to just keep on dripping and dripping
with new questions for Ruben because I know that he has a life
as everybody else does.
And I think actually he left the – or no, he moved down.
But Dallas, you're a good guy.
You want to say the last words, and I'm just going to just say something and
then we'll close up the space.
Okay. Well, then I'll just, I guess, close out the room. Yeah, no,
it was a great chat. Really appreciated all the insight,
all the alpha being dropped on the panel,
and excited to be able to share some of the stuff
that I've been learning and implementing
through participating in the social audio experiences.
And yeah, no, AI John, and kudos to you
and Andy, who's not here.
I'm looking forward to the next one. Thank you.
