How to Survive the Next Tech Revolution (w/ Brett Gibson)
Summary
TLDRThis video explores the evolving nature of technology and engineering, drawing parallels between past experiences with startups like Posterous and the current advancements in AI. Brett Gibson, a seasoned engineer and investor, shares insights on embracing unconventional approaches, adapting to new tools and open-source ecosystems, and prioritizing user experience. The discussion delves into the challenges faced by technical founders at the forefront of innovation, emphasizing the importance of generalist skillsets, software testing, and validating ambitious ideas. Gibson's perspectives underscore the art of problem-solving and the iterative process required to create impactful products that drive the future.
Takeaways
- 🔑 Engineering mindset needs to balance precise determinism with probabilistic approaches for tackling novel problems that create user value.
- ⚒️ Building something that doesn't exist yet requires unusual and innovative engineering solutions, but later rewrites should leverage available tools, open source, and APIs.
- 🎯 The goal is to meet user needs and create sticky products, not just master tooling - being a domain expert in tools aids this goal.
- ⏳ As foundational models and infrastructure rapidly evolve, pragmatically rewriting custom code using new capabilities becomes essential.
- 🌱 Investing in startups solving ambitious technical challenges requires validating if the solution is now possible due to emerging technologies.
- 👥 Hiring generalist engineers adept at diverse skillsets is valuable for novel engineering problems.
- ✅ Rigorous software testing, especially for core cases, prevents issues in high-stakes applications like crypto custody and staking.
- 📖 For problems at the edge of human capability, write your own manual through extensive validation and creative problem-solving.
- 🔄 Continuously re-evaluate potential failure modes, even under time pressure, to get to the right solution.
- 💎 Founders and investors can create massive value by being ahead of the curve on ambitious technical frontiers like asteroid mining.
Highlights
Please replace the link and try again.
Transcripts
technology it's changing the world and
it's the people who can build something
from scratch something that never
existed before
who will go on to change that world
but then what happens for v2 the
rewrites always better who wins in that
world it's a question playing out in AI
today so that's why we're sitting down
with my multi-time co-founder and friend
Brett Gibson he runs initialized now
alongside Jen Wolfe now that I'm back at
YC what's the future gonna look like it
won't be the same as the past but it
will rhyme let's get started
[Applause]
all right
[Music]
Brett and I worked on a startup named
posterous together with our other
co-founder Sachin Agarwal
one of the new things we did was a new
kind of authentication by email header
you could post your blog where we would
use your email to figure out who you
were so then you could just send email
to postpostress.com and we'd figure it
out that was the thing that had not
really been done before in a consumer
product and to do it it required a
different kind of engineering mindset a
big part of an engineer's day-to-day
problem is figuring out the right tools
for the job figuring out what they
should be using and figuring out how it
ties back to user value because you know
I think that I come from this you know
this default engineering mentality where
I really want things to be precise and
deterministic and you know oftentimes
the engineering challenges you face to
get to get the product working the way
you want it for your users are much more
probabilistic because it hadn't been
done before we had to do some pretty
unusual things to get it right you know
we ended up writing our own uh parts of
our own SMTP client which I don't I
don't really recommend doing a lot of
analysis on the routing that email went
through so that we could you know have
some probabilistic score about How
likely it was to actually be from you
and then the postaven experience was
night and day Preposterous later sold to
Twitter for 20 million dollars but a few
years later they decided to shut it down
since the real reason why they bought it
was for the team
when it shut down Brett and I got back
together to write it again this was
called post Haven and this time when we
wrote it again this post by email
authentication that we had to write for
the first time had been commodified
there was a product it was mail gun it
was ready it worked off the shelf for 90
of what we needed had we been too
grounded in fighting the last battle we
would have missed that we would have
wasted a lot of time we would have built
a lot of software that we didn't need to
build what we're seeing today in 2023 is
similar to what we found rewriting
posturous again as post-haven even just
four years later first it's a lot easier
to rewrite once you've written it once
but second tooling an open source and
freely available apis gets so much
better in just the course of a few years
for AI now that set of years sometimes
is months or weeks
but know that this isn't a new thing
what is new is how much more powerful
these capabilities in ml and large
language models have become now everyone
has this new primitive of generative Ai
and especially large language models and
they're trying to figure out exactly how
to use them and the moment is actually
very strange because there's a lot of
interesting Tech on how to do it you
know there's a lot of best practices
around stuff like vector databases and
embeddings so that you can get your
prompts correct but it's it's
exacerbated because large language
modules are fundamentally
non-deterministic systems
um that's you know the the lack of
determinism is is in part what makes
them work so well and seems so magic and
so the the class of I guess you know
first generation hacks to get them
working in ways that deliver value to
users is is perhaps more interesting
than some of the things we've seen in
the past along the same lines this alien
intelligence we have access to now it's
a strange one and we're just in the
first Innings of the tooling the
foundational models and especially the
open source it's interesting because I
see a lot of parallels in what's
happening in AI right now and what
happened in crypto over the last few
years because there was a moment you
know there's so much infrastructure
around wallets and building up
you know ability to be consumer facing
in crypto that if you were building in
2018 you just had to build from scratch
and the underlying ecosystem has come a
long way and you're probably having to
throw away some of that custom code and
in AI it's it's the same thing but
perhaps even moving faster because not
only do you have the day-to-day building
out of of the of the types of tooling
that would make your life easier you we
have these step function improvements in
the underlying foundational models the
change the requirements I think that the
North Star is always backing it out from
what what experience you want to give to
the user because if you make sure you
get that right then you can change your
technology behind the scenes
um but you know trying to try to have
the trade-off of uh how long it takes to
deliver that versus competitors that is
the lesson from software engineering and
it's the same one Brett and I learned in
2008 2013 and we're learning again today
in 2023 whether it was just some new
capability in the early days of social
to what's happening today in AI first
you have to be willing to build
something that doesn't exist yet and do
some pretty unusual things to do it but
then later you need to be able to do
rewrites and stay current with open
source and new Stacks capabilities and
apis all while being connected to the
needs of your customers and users make
stuff people want and that is what makes
products sticky and what makes a durable
business being a domain expert in
tooling
really helps gives you a lot of Leverage
toward that goal but it's not the goal
in its own right
thank you Brett's one of the best
Engineers I've ever had a chance to work
with across our Year's writing code for
Preposterous y combinator post Haven and
initialized and he's always been super
pragmatic and I've learned something new
from him regularly he's got a new
podcast called the high bit where he
sits down with some of the best
technical Founders to talk Tech
engineering and the idea maze kind of
what we just did in this episode one of
those companies was bison Trails which
was acquired by coinbase and post IPO
the notional value of that acquisition
was nearly a billion dollars that's
because it became coinbase Cloud one of
the key SAS Revenue generators for
coinbase
Brett invested in bison Trails when it
was just a few people very early when
proof of stake was little more than an
idea something that did not actively run
the world's largest blockchains like it
does now but that's precisely how you
should invest when you want to create
the future when I met bison Trails there
was there were little or no major
networks running proof of stake and and
I still had doubts in my mind as to the
viability of widespread shift to proof
of stake networks especially for already
running networks like ethereum now it's
very clear that it's it's demodually
possible if not optimal in many contexts
with the Bryson Trails Founders
recognized was that relative to the
current proof of work mining that was
happening running validators on proof of
stake networks was a novel and Rel and
difficult problem they were able to
build something that had never been
built before because of two things first
they were generalist focusing on a wide
skill set here's Aaron Henshaw
co-founder of Bison Trails on how hiring
was a big piece of their success the
best thing that we did was we've
actually brought on like generalists at
the beginning more people who could both
like right JavaScript and do terraform
that's a really wide band between those
two things but that's what we needed at
the beginning because we had JavaScript
right we actually had like a front end
we had an API to build so we had a
middle and then we had this whole back
end you're lucky if you go far enough to
be able to hire the people that you need
to fix your problems that you've created
for yourself and those people you know
they come in wide open but they like
they're so good at it they're so they're
so good at these specific problem
solving and like you're just like in awe
that they can come in and just like
change it all and make it work like 10x
better second you've really got to
invest into software testing so that
when the chips are on the line and they
always are in crypto custody and staking
you have the right engineering infra
some of the things that we did well
early was like we did not try to over
engineer things there was probably some
opportunities to use like slightly more
off-the-shelf CI CD and bake in a little
bit more testing early on that I think
like might have slightly slowed us down
but actually sped us up this may be like
one of the biggest takeaways from the
whole thing was
you should invest heavily in tests
especially even early like don't go
crazy right like you can't be writing
tests for ever but you should have core
tests that cover base cases and that
like help your whatever it is that
you're deploying Founders and investors
both make money and impact the world
when they believe something nobody else
believes yet but are right another
company Brett's invested in is a very
ambitious one called astroforge they're
literally doing mining for high value or
on asteroids at no point was anyone
really worried about the market
opportunity you know we're always
balancing Market risk with technical
risk in in the startups we fund and
there's just no question that if astral
Forge can go to asteroids and mine Rare
Earth elements and bring them back you
know the the only real risk in the
market is that they do it so
successfully that those metals are worth
a lot less and so really all we spent
our time was in was validating is this
possible and I think specifically if it
is in fact possible why is it possible
now suddenly why why did why are we
seeing this company today because on
some timeline it seems like something
that's well within the ability of humans
to produce but you need to have a view
that it's going to be it's going to
happen in time for this company to make
money off it Hosea Kane co-founder of
Aster Forge is trying to solve
engineering problems on literally the
edge of human capability and the thing
is you don't have a manual for that
you've got to write the manual yourself
the gut reaction was asteroid mining is
difficult the refinery is difficult it
couldn't be awkward it couldn't be a
chord it couldn't be it couldn't be the
simple pass through the length that we
went through to solve this problem
baffles me to this day you're focusing
on this problem that's difficult glaze
using over something so simple and and I
talked about this fishbone diagram
because these are things that you should
really actually check you should write
out great the graphs before agreement do
we check everything in there and not
just say like yeah I checked the voltage
which is just one part of it
um and we just checked it off like okay
cool power supply works hard this works
great even if you are under heavy time
pressure take a minute and pause slow
down it's okay to do that and really
assess the situation and really validate
all these potential failures to get to
that Solution that's just a taste of
what you'll find in Brett's new podcast
I'm subscribing and I recommend that you
do too you know the intent is less a
technical Deep dive and more about this
this the the story about the art of
problem solving and what goes into
day-to-day Engineering in these
technical domains and how it can be you
know just interesting in its own right
you can find high bit by Brett Gibson on
anywhere you listen to podcasts or
initialize
Link in the description below
initialized is also hiring a new
investment partner you can't apply to
jobs you don't know about so that's why
they're running an open hiring process
for this role initialize is currently
hiring a partner role and you know we we
really prize uh backgrounds in founding
and working in startups and ability to
advise startups in in some area of deep
domain expertise so uh you know high
level I think we're looking for someone
who's who's been a Founder before you
know hopefully gotten a company uh
through perhaps a series B round and and
but also it's very helpful if they have
some investing experience whether that's
as an angel or within or as part of a a
venture firm
um you know a big a big part of the
applied knowledge of going from founder
to investor is just getting reps and and
seeing pitches being an investor to me
is just an absolute blessing and if you
ever wanted to join a team doing it
right and you have some investment
experience already this is a big
opportunity
that's it for this week Brett's an old
friend of mine and I'm so excited to see
where he takes initialized Capital links
in the description to apply to be an
investor at initialize and to subscribe
to Brett's new podcast hi bit technology
is truly reshaping the world and its
technologists the engineers the PMS the
designers and the builders themselves
they're the ones doing it it if you
watched to the end you must believe this
too and I'm here to tell you you're on
the right track see you next time
[Music]
thank you
[Music]
Browse More Related Video
![](https://i.ytimg.com/vi/bjrTAng0Bq8/hq720.jpg)
Understanding the role of a distinguished engineer
![](https://i.ytimg.com/vi/cIW_FeJhrbA/hq720.jpg)
Creativity in the Age of AI: Generative AI Issues in Art Copyright & Open Source
![](https://i.ytimg.com/vi/ccA_nYX4Xys/hq720.jpg)
Design Thinking Q&A with Clark Kellogg and Stephane Matsushita
![](https://i.ytimg.com/vi/rkDKIJHFtMM/hqdefault.jpg?sqp=-oaymwExCJADEOABSFryq4qpAyMIARUAAIhCGAHwAQH4Af4JgALQBYoCDAgAEAEYciBSKEUwDw==&rs=AOn4CLCQ2ie1oU3k0Zk1Vka-70u3i9XS2A)
ひふみ目論見倶楽部【準備運動編】安宅和人さんと一緒に未来について考えました。
![](https://i.ytimg.com/vi/jNNatjruXx8/hq720.jpg)
Discover Prompt Engineering | Google AI Essentials
![](https://i.ytimg.com/vi/Ou2Uo_F7XBw/hq720.jpg?sqp=-oaymwEmCIAKENAF8quKqQMa8AEB-AH-CYAC0AWKAgwIABABGEEgJyh_MA8=&rs=AOn4CLA9iF1ePMreoA6hrTgFN48vlnrCAQ)
Digital Marketing Institute | AI with DMI | GA4 and AI
5.0 / 5 (0 votes)