Paul Graham's talk, "Before the Startup," outlines six counterintuitive points about starting a startup. He emphasizes that startups are unlike anything most people have experienced and that relying on instinct often leads to mistakes. Key advice includes trusting instincts about people, focusing on user needs instead of startup mechanics, and understanding that gaming the system doesn't work in startups. He advises against starting a startup in college, encouraging students to learn broadly and pursue genuine curiosity, as the best startup ideas often emerge from side projects.
Key Takeaways
1
Startups are counterintuitive, and following your instincts can lead you astray, similar to learning to ski where initial impulses are often incorrect.
2
While startups are counterintuitive, you can and should trust your instincts about people, choosing to work with those you genuinely like and respect.
3
Success in a startup comes from understanding your users deeply and creating something they want, not from expertise in the mechanics of starting a startup or fundraising.
4
Gaming the system, a strategy often learned in traditional education, does not work in startups because there's no 'boss to trick,' only users who care if your product meets their needs.
5
Starting a startup is an all-consuming commitment that irrevocably changes your life, similar to having children, and the problems never get easier, only different.
6
Paul Graham advises against starting a startup in college, suggesting that youth is a better time for exploration and deep learning that may lead to future startup ideas.
7
You cannot predict your success in a startup based on past achievements, as starting a startup is a unique challenge that will change you significantly.
8
The best startup ideas emerge unconsciously from side projects and genuine curiosity, rather than actively trying to brainstorm startup ideas.
9
To develop a mind that generates startup ideas, learn a lot about important things, work on problems that genuinely interest you, and collaborate with people you like and respect.
10
The most effective thing to do in college if you aspire to be a successful founder is to pursue education for its own sake and learn powerful things driven by intellectual curiosity.
No matching segments found
0:01
that short like like long introductions
0:04
are no good um Sam knows H all right
0:08
ready everybody I'm not going to ask if
0:11
the mic is working like in every talk so
0:13
far um I'll just assume it's working I'm
0:17
just no
0:21
[ __ ] all right well make it work
0:23
somebody it works it works all right all
0:25
right this is like some kind of class
0:27
tradition um all right all right
0:31
uh I wrote out my talk um and afterwards
0:34
in a couple days I will like turn it
0:36
from a talk into an essay and put it
0:37
online so you don't have to take notes
0:39
just just listen um all right so one of
0:44
the advantages of having kids is that
0:46
when you have to give advice to people
0:48
you can ask yourself what would I tell
0:49
my own kids and actually you find this
0:51
really focuses you um so even though my
0:54
kids are little uh my 2-year-old today
0:58
went asked what he was going to be after
0:59
two said bat um the correct answer was
1:03
three but a bat is so much more
1:05
interesting um so even though my kids
1:08
are little I already know what I would
1:10
tell them about startups if they were in
1:12
college um and so that is what I'm going
1:14
to tell you so you're literally getting
1:15
what I would give my own kids um since
1:18
most of you are young enough to be my
1:20
own kids all right so startups are very
1:24
counterintuitive and I'm not sure
1:26
exactly why it could be simply because
1:29
knowledge about them has not permeated
1:31
our culture yet but whatever the reason
1:33
this is an area where you cannot trust
1:35
your intuitions all the time it's like
1:38
skiing in that way any of you guys learn
1:40
to ski as adults you know when you're
1:42
skiing when you first try skiing and you
1:44
want to slow down your first impulse is
1:46
to lean back just like in everything
1:47
else but lean back on skis and you fly
1:51
down the hill out of control so as I
1:53
learned um so part of learning to ski is
1:57
learning to suppress that impulse
1:59
eventually you get new habits but in the
2:01
beginning there's this list of things
2:03
you're trying to remember as you start
2:04
down the hill you know like alternate
2:06
feet make S turns do not drag the inside
2:09
foot all this stuff um well startups are
2:12
as unnatural as skiing and there is a
2:15
similar list of stuff you have to
2:16
remember for startups and what I'm going
2:18
to give you today um is the beginning of
2:21
the list the list of the
2:22
counterintuitive stuff you have to
2:24
remember to like prevent your existing
2:26
instincts from leading you
2:28
astray the first thing it is the fact I
2:32
just mentioned that startups are so
2:33
weird that if you follow your instincts
2:36
they will lead you astray if you
2:37
remember nothing more than that uh when
2:40
you're about to make a mistake you may
2:42
at least pause before making it when I
2:45
was running y combinator we used to joke
2:47
that um our function was to tell
2:50
Founders things they would ignore um and
2:53
it's really true batch after batch the
2:55
YC Partners warn Founders about mistakes
2:58
they are about to make and the founder
3:00
ERS ignore them and they come back a
3:02
year later and say I wish we'd listened
3:05
um but that dude is in their cap table
3:07
there's nothing they can do um why do
3:10
founders persistently ignore the
3:13
partner's advice well that's the thing
3:14
about counterintuitive ideas they
3:17
contradict your intuitions so um they
3:20
seem wrong so of course your first
3:22
impulse to ignore is to ignore them and
3:25
in fact that is not just the curse of Y
3:27
combinator but to some extent our risone
3:30
Det
3:30
you don't need people to give you advice
3:32
that doesn't surprise you right um if
3:35
Founders existing intuitions gave them
3:37
the right answers they wouldn't need us
3:39
that's why there are a lot of ski
3:41
instructors and not many running
3:43
instructors right like that you don't
3:45
see those two words together running
3:47
instructor as much as you see ski
3:49
instructor um it's because skiing is
3:52
counterintuitive so sort of what um what
3:56
YC is is like uh business ski
3:59
instructors except for going up slopes
4:01
instead of down them um well ideally uh
4:06
you can however trust your instincts
4:09
about people you your life so far hasn't
4:12
been much like starting a startup but
4:14
all the interactions you've had with
4:15
people are just like the interactions
4:16
you have with people in the business
4:18
world so in fact one of the big mistakes
4:20
that Founders make is not to trust their
4:22
intuitions about people enough they meet
4:25
someone who seems impressive but about
4:28
whom they feel some misgivings
4:30
um and then later when things blow up
4:33
they tell them they say you know I knew
4:35
there was something wrong about that guy
4:37
but I ignored it because he seemed so
4:39
impressive um and there's this a
4:42
specific subcase in business especially
4:45
if you come from an engineering
4:46
background as I believe you all do you
4:48
think business is supposed to be this
4:50
sort of slightly distasteful thing and
4:52
so when you meet people who seem smart
4:54
but somehow distasteful you think well
4:57
okay this must be normal for business
4:59
but it's not just like pick people the
5:01
way you would pick people if you were um
5:03
picking friends um this is one of those
5:06
rare cases where it works to be
5:08
self-indulgent work with people you
5:10
genuinely like and respect and that you
5:12
have known long enough to be sure
5:15
because there's a lot of people who are
5:17
really good at seeming likable for a
5:18
while um just wait till your interests
5:21
are opposed and then you'll see all
5:23
right um the second counterintuitive
5:26
point is that uh and this will might
5:29
come as a little bit of a disappointment
5:31
um but what you need to succeed in a
5:34
startup is not expertise in startups
5:38
that makes this class different from
5:40
most other classes you take if you take
5:42
a French class at the end of it you will
5:44
learn how to speak French if you do the
5:46
work you may not sound exactly like a
5:48
French person but pretty close right
5:52
um this class can teach you about
5:54
startups but that is not what you need
5:56
to know what you need to know to succeed
5:58
in a startup is not expertise in
6:00
startups what you need is expertise in
6:03
your own users Mark Zuckerberg did not
6:06
succeed in Facebook because he was an
6:09
expert in startups he succeeded despite
6:11
being a complete Noob at startups I mean
6:14
Facebook was first incorporated as a
6:16
Florida
6:17
LLC right even you guys know better than
6:21
that
6:22
um he succeeded despite being a complete
6:26
Noob at startups because he understood
6:28
his users very well
6:31
most of you don't know the mechanics of
6:33
raising an angel round right um and if
6:36
you feel bad about that don't because I
6:39
can tell you Mark Zuckerberg probably
6:40
doesn't know the mechanics of raising an
6:42
angel round either if he was even paying
6:45
attention when Ron Conway wrote him the
6:47
big check um he has probably forgotten
6:50
about it by now in fact I worry it's not
6:53
merely unnecessary for people to learn
6:55
in detail about the mechanics of
6:57
starting a startup but possibly somewhat
6:59
dangerous
7:00
because another of the characteristic
7:02
mistakes of young Founders starting
7:04
startups is to go through the motions of
7:07
starting a startup they come up with
7:09
some plausible sounding idea um they
7:13
raise funding at a nice valuation they
7:16
rent a nice Office in Soma hire a bunch
7:18
of their friends and then the next step
7:22
after uh rent a nice Office in Soma and
7:26
hire a bunch of their friends is
7:28
gradually realize how how completely
7:30
[ __ ] they are because while imitating
7:33
all the outward forms of starting a
7:35
startup they have neglected the one
7:37
thing that's actually essential which is
7:38
to make something people want by the way
7:41
that's the only use of that swear word
7:43
except for the initial one that was
7:44
involuntary um and I did check with Sam
7:47
whether it would be okay um he said he
7:50
had done it several times I mean Ed the
7:52
word uh
8:03
we saw this happen so
8:06
often no I mean people going through the
8:08
motions of starting a startup
8:11
um that we made up a name for it playing
8:15
house eventually I realized why it was
8:18
happening the reason young Founders go
8:20
through the motions of starting a
8:22
startup is because that is what they
8:24
have been trained to do for their whole
8:26
lives up to this point think about what
8:29
it takes to get into college
8:32
extracurricular activities check right
8:37
um even in college classes most of the
8:40
work you do is as artificial as running
8:41
laps and I am not attacking the
8:43
educational system for being this
8:46
way inevitably the work that you do to
8:49
learn something is going to have some
8:50
amount of fakeness to it and if you
8:52
measure people's performance they will
8:54
inevitably exploit the difference to the
8:56
degree that what you're measuring is
8:58
largely an artifact um the fakeness and
9:01
I confess that I did this myself in
9:03
college in fact here's a useful tip on
9:05
getting good grades um I found that in a
9:08
lot of classes there might only be 20 or
9:10
30 ideas that actually had the right
9:12
shape to make good exam questions and so
9:14
the way I studied for exams in these
9:16
classes was not to master the material
9:19
in the class but to try and figure out
9:21
what the exam questions would be and
9:22
work out the answers in advance and like
9:25
for me the test was not like what my
9:27
answers would be on the exam for me the
9:29
test was which of my exam questions
9:31
would turn up on the exam right so I
9:34
would I would get my grade instantly I
9:36
would walk into the exam and look at the
9:38
questions and see how many I got right
9:40
essentially
9:41
um it works in a lot of classes
9:44
especially CS classes I remember automa
9:46
Theory there's only a few things that
9:48
make sense to ask about automa Theory um
9:51
so it's not surprising that after being
9:53
effectively trained for their whole
9:55
lives to play Such games young Founders
9:57
first impulse on starting a startup is
9:59
to figure figure out what the tricks are
10:01
for this new game what are the
10:03
extracurricular activities of startups
10:05
what are the things I have to do um they
10:07
always want to know since since
10:10
apparently the measure of success for a
10:12
startup is fundraising another Noob
10:14
mistake um they always want to know what
10:17
are the tricks for convincing investors
10:19
and we have to tell them the best way to
10:21
convince investors is to start a startup
10:23
that's actually doing well meaning
10:25
growing fast and then simply tell
10:27
investors so um so then they ask okay
10:31
what are the tricks for grilling fast
10:33
right um and this is exacerbated by the
10:35
existence of this term growth hacks
10:38
right whenever you hear anybody talk
10:40
about growth hacks just mentally
10:42
translate it in your mind into [ __ ]
10:45
um because what we tell them is the way
10:48
to make your startup grow is to make
10:51
something that users really love right
10:53
and then tell them about it h so that's
10:56
what you have to do that's growth hacks
10:58
right there um so so so many of the
11:01
conversations the YC Partners have with
11:04
the founders begin with the founders
11:06
saying a sentence that begins with how
11:07
do I and the partner is answering with
11:10
the sentence that begins with just right
11:13
why do they make things so complicated
11:16
the reason I realized after years of
11:17
being puzzled by this is they're looking
11:19
for the trick they've been trained to
11:21
look for the trick so this is the third
11:24
thing the third counterintuitive thing
11:26
to remember about startups starting a
11:28
startup is where gaming the system stops
11:30
working gaming the system may continue
11:33
to work if you go to work for a big
11:35
company depending on how broken the
11:37
company is you may be able to succeed by
11:40
sucking up to the right person um giving
11:43
the impression of productivity by
11:45
sending emails late at night um or if
11:48
you're smart enough changing the clock
11:50
on your computer um because who's going
11:52
to check the headers right um but I like
11:56
an audience that I can tell jokes like
11:58
that too and they laugh um over in the
12:01
business school
12:02
headers um
12:05
okay God this thing is being
12:09
recorded I just realized
12:13
that all right from now on we are
12:15
sticking Strictly To The Script um all
12:19
right um but in startups that does not
12:22
work there's no boss to trick there's no
12:25
how can you trick people if there's no
12:27
one to trick there's only users and all
12:29
users care about um is that is whether
12:32
your software does what they want right
12:34
they're like sharks sharks are like too
12:37
stupid to fool you can't like wave a red
12:39
flag at a shark and fool it it's like
12:41
meat or no meat um so you have to have
12:44
something people want and you only
12:46
Prosper to the extent that you do the
12:49
dangerous thing is particularly for you
12:51
guys the dangerous thing is that faking
12:53
does work to some extent with investors
12:55
you if you're really good at knowing
12:57
what you're talking about you can fool
12:59
in ERS for one maybe two rounds of
13:01
funding but it's not in your interest to
13:04
I mean you're all doing this for Equity
13:06
you're playing a confidence trick on
13:07
yourself you're wasting your own time
13:09
because the startup is doomed and all
13:11
you're doing is you're just going to
13:13
waste your time writing it down um so
13:15
stop looking for the trick there are
13:17
tricks in startups as there are in any
13:20
domain but they are in order of
13:21
magnitude less important than solving
13:23
the real problem someone who knows zero
13:26
about fundraising but has made something
13:28
users really love will have an easier
13:30
time raising money than someone who
13:32
knows every trick in the book but has a
13:34
flat usage graph um though in a sense
13:37
it's bad news that gaming the startup
13:40
stops work gaming the system stops
13:42
working now um in the sense that you're
13:45
deprived of one of your most powerful
13:46
weapons and after all something you have
13:48
spent 20 years mastering um it is
13:52
something I find it very exciting uh
13:55
that the that the that there even exist
13:57
parts of the world where game the system
14:00
is not how you win I would have been
14:02
really excited in college if I had
14:04
explicitly realized that there are parts
14:06
of the world where gaming the system
14:08
matters less than others and some where
14:10
it hardly matters at all but there are
14:12
and this is one of the most important
14:13
things to think about when you're
14:14
planning your future
14:18
um got that okay
14:22
um how do you win at each type of work
14:25
and what do you want to win by
14:28
doing so
14:30
um that brings us to our fourth
14:32
counterintuitive Point startups are all
14:34
consuming if you start a startup it will
14:37
take over your life to a degree that you
14:40
cannot imagine um and if it succeeds it
14:44
will take over your life for a long time
14:45
for several years at the very least
14:47
maybe a decade maybe the rest of your
14:49
working life so there's a real
14:51
opportunity cost here it may seem to you
14:55
that Larry pagee has an enviable life
14:57
but there are parts of it that are
14:58
definitely unv viable the way the world
15:01
looks to him is that he started running
15:03
as fast as he could at age 25 and he has
15:06
not stopped to catch his breath since
15:09
every day [ __ ] happens within the Google
15:11
Empire that only the emperor can deal
15:14
with and he as the emperor has to deal
15:16
with it if he goes on vacation for even
15:19
a week a whole backlog of [ __ ]
15:22
accumulates and he has to bear this
15:25
uncomplainingly because number one as
15:27
the company's Daddy he can never show
15:29
fear or weakness and number two if
15:32
you're a billionaire you get zero
15:34
actually less than zero sympathy if you
15:36
complain about having a difficult
15:40
life which has the strange side effect
15:43
that the difficulty of being a
15:45
successful startup founder is concealed
15:48
from almost everyone who's done it
15:49
people who win the 100 meters in the
15:51
Olympics like they walk up to them and
15:52
they're going
15:53
like right like Larry pig is doing that
15:56
too but you never get to see it um all
15:59
right where are we ycombinator has now
16:02
funded several companies that could be
16:03
called Big successes um and in every
16:06
single case the founders say the same
16:08
thing it never gets any easier the
16:11
nature of the problems change so you may
16:14
be you may be worrying about more
16:15
glamorous problems like construction
16:17
delays in your New London offices rather
16:20
than the broken air conditioner in your
16:22
studio apartment but the total volume of
16:24
worry never decreases if anything it
16:27
increases starting a successful startup
16:29
is similar to having kids and that it's
16:32
like a button you press that changes
16:34
your life
16:35
irrevocably um and while it is like
16:39
honestly the best thing in the world
16:41
having kids um if you take away one
16:43
thing from this lecture remember that um
16:47
uh there are a lot of things that are
16:50
easier to do after you before you have
16:53
kids then after many of which will make
16:55
you a better parent when you do have
16:57
kids and so in rich countries
17:00
most people delay pushing the button for
17:02
a while and I'm sure you are all
17:04
intimately familiar with that um
17:06
procedure
17:07
um yet when it comes to starting
17:10
startups a lot of people seem to think
17:12
they're supposed to start them in
17:14
college are you crazy and what are the
17:17
universities thinking they go out of
17:19
their way to ensure that their students
17:21
are well supplied with contraceptives
17:23
and yet they're starting up
17:24
entrepreneurship programs and startup
17:26
incubators left and right
17:29
to be fair the universities have their
17:31
hand forced here a lot of incoming
17:33
students are interested in startups
17:35
universities are at least de facto
17:37
supposed to prepare you for your career
17:40
and so if you're interested in startups
17:41
it seems like universities are supposed
17:43
to teach you about startups and if they
17:44
don't maybe they lose applicants to
17:47
universities who do claim to do that so
17:50
can universities teach you about
17:52
startups well not what are we doing here
17:56
um yes and no as I explained they can
17:59
teach you about startups but this is not
18:01
what you need to know essentially what
18:04
univers if you want to learn French
18:06
universities can teach you Linguistics
18:08
that's what this is this is a
18:09
Linguistics class right we're teaching
18:11
you about how to learn languages and
18:13
what you need to know is like how to
18:14
learn a particular language um what you
18:18
need to know are the needs of your own
18:20
users you can't learn those until you
18:22
actually start the company um and
18:25
starting a company which means that
18:27
starting a startup is something and you
18:28
can intrinsically only learn by doing it
18:31
and you can't do that in college for the
18:32
reason I just explained that startups
18:34
take over your entire life if you start
18:37
a startup in college if you start up
18:38
startup as a student you can't start a
18:40
startup as a student because if you
18:41
start a startup you're not a student
18:43
anymore you may be nominally a student
18:45
but you won't even be that for much
18:46
longer so given this dichotomy which of
18:49
the two paths should you take be a real
18:52
student and not start a startup or start
18:54
a real startup and not be a student well
18:56
I can answer that one for you I'm
18:58
talking to my own kids
19:00
here do not start a startup in college
19:04
uh I hope I'm not disappointing anyone
19:06
seriously honestly um starting a startup
19:10
could be a component of a good life for
19:12
a lot of ambitious people but this is
19:14
just part of a much bigger problem that
19:16
you're trying to solve how to have a
19:18
good life right and though starting a
19:20
startup could be a good thing to do at
19:22
some point 20 is not the optimal time to
19:25
do it there are things that uh that you
19:28
can't there there are things you can do
19:30
in your early 20s that you cannot do as
19:32
well before or after like plunge deeply
19:35
into projects on a whim that seem like
19:37
they'll have no payoff um or travel
19:40
super cheaply with no sense of a
19:42
deadline in fact those are really just
19:44
isomorphic shapes in different domains
19:46
um for unambitious people this sort of
19:49
thing can be the dreaded Failure to
19:51
Launch but for the ambitious ones it's a
19:54
really valuable sort of exploration and
19:57
if you start a start up at 20 and
19:59
sufficiently successful you will never
20:01
get to do it Mark Zuckerberg will never
20:04
get to bum around a foreign country he
20:07
can if he goes to a foreign country it's
20:08
either as a de facto State visit or like
20:11
he's hiding out Incognito at the George
20:13
the George sank in Paris right um but
20:17
he's never going to get to just like
20:18
backpack around Thailand if that's still
20:20
what people do um do people still
20:23
backpack around Thailand Okay well there
20:26
that's the first real sign of enthusiasm
20:28
I've seen from this should have given
20:31
this talk in Thailand
20:35
um all right he can do things that you
20:38
can't do like Charter Jets to fly him to
20:41
foreign countries really big jets um but
20:45
success has taken a lot of the
20:46
Serendipity out of his life he uh
20:49
Facebook is running him as much as he's
20:51
running Facebook and while it can be
20:53
really cool to be in the grip of some
20:56
project you consider your life's work
20:59
um there are advantages to
21:02
Serendipity and uh among other things it
21:06
gives you more options to choose your
21:08
life's work from there's not even a
21:10
trade-off here you're not sacrificing
21:12
anything if you forego starting a start
21:14
up at 20 because you'll be more likely
21:16
to succeed if you wait in the unlikely
21:19
case that you're 20 like astronomically
21:21
unlikely case that you're 20 and you
21:23
have some side project that takes takes
21:26
off like Facebook did then you face a
21:28
choice either running with it or not and
21:30
maybe it's reasonable to run with it but
21:33
usually the way startups take off is for
21:35
the founders to make them take off and
21:36
it's gratuitously stupid to do that at
21:41
20 so should you do it at any age I
21:44
realize I've made starting a startup
21:46
sound kind of hard if I haven't let me
21:49
try again starting a startup is really
21:53
hard H what if it's too hard what if
21:56
you're not up to this challenge the the
21:59
answer is the fifth counterintuitive
22:01
point you can't tell your life so far
22:05
has given you some idea what your
22:07
prospects might be if you wanted to
22:09
become a mathematician or a professional
22:12
football player um and boy it's not
22:14
every audience you could say that to you
22:17
H but unless you have had a very strange
22:19
life indeed you have not done much
22:21
that's like starting a startup meaning
22:23
starting a startup will change you a lot
22:26
if it works out so what you're trying to
22:28
estimate is not not just what you are
22:29
but what you could become and who can do
22:31
that well not me for the last nine years
22:35
it was my job to try and guess whether
22:37
people would guess I wrote predict in
22:40
here and it came out as guess um that's
22:42
a very informative Freudian slip
22:46
um seriously it's easy to tell how smart
22:50
people are in 10 minutes you know hit a
22:52
few tennis balls over the net and do
22:53
they hit them back at you or into the
22:55
net um but the hard part was and the
22:59
most important part was predicting how
23:00
tough and ambitious they would become um
23:03
there may be no one at this point who
23:05
has more experience than me at doing
23:06
this and I can tell you how much an
23:08
expert can know about that the answer is
23:10
not much I learned from experience to
23:13
keep a completely open mind about which
23:15
startup in each batch which startups
23:17
would turn out to be the Stars the
23:19
founders sometimes thought they knew
23:22
some arrived feeling confident that they
23:24
would Ace y combinator just as they had
23:27
aced everyone of the few easy artificial
23:30
tests they had faced in life so far
23:33
others arrived wondering what mistake
23:35
had caused them to be admitted and
23:37
hoping that no one would discover it um
23:40
but there is little to no attit no
23:43
correlation between these attitudes and
23:44
how things turn
23:47
out I've read the same is true in the
23:49
military that the swaggering recruits
23:51
are no more likely to turn out to be
23:53
really tough than the quiet ones and
23:55
probably for the same reason that the
23:57
tests are so different from the tests in
23:59
people's previous lives if you're
24:02
absolutely terrified of starting a
24:03
startup you probably shouldn't do it
24:05
unless you're one of those people who
24:07
gets off on doing things you're afraid
24:08
of um otherwise if you're merely unsure
24:11
of whether you're going to be able to do
24:13
it the only way to find out is to
24:16
try just not
24:19
now so if you want to start up one day
24:23
if you want to start a startup one day
24:25
what do you do now in college there are
24:27
only two things you need initially an
24:31
idea and co-founders and the mo
24:34
forgetting them both is the same which
24:36
leads to our sixth Sixth and last
24:39
counterintuitive Point um the way to get
24:42
startup ideas is not to try to think of
24:44
startup ideas I've written a whole essay
24:46
on this and I'm not going to repeat the
24:48
whole thing here um but the short
24:50
version is that if you make a conscious
24:52
effort to try and think of startup ideas
24:54
you will I think of ideas that are not
24:56
only bad but bad and plausible sounding
24:59
meaning you and everybody else will be
25:01
fooled by them and you'll waste a lot of
25:02
time before realizing they're no good
25:05
the way to come up with good startup
25:07
ideas is to take a step
25:09
back instead of trying to make a
25:11
conscious effort to think of startup
25:13
ideas turn your brain into the type that
25:16
has startup ideas
25:18
unconsciously in fact so unconsciously
25:21
that you don't even realize at first
25:23
that there startup ideas this is not
25:25
only possible Yahoo Google
25:29
Facebook and apple all got started this
25:32
way none of these companies were even
25:34
supposed to be companies at first they
25:36
were all just side projects the very
25:39
best ideas almost have to start as side
25:41
projects because they're always such
25:42
outliers that your conscious mind would
25:45
reject them as ideas for
25:47
companies okay so how do you turn your
25:51
mind into the kind that has startup
25:53
ideas unconsciously one learn a lot
25:56
about things that matter two work on
25:59
problems that interest you three with
26:02
people you like and respect that third
26:05
part incidentally is how you get
26:06
co-founders at the same time as the idea
26:10
the first time I wrote that paragraph
26:12
instead of learn a lot about things that
26:14
matter I
26:15
wrote become good at some technology but
26:19
that prescription though sufficient is
26:22
too narrow what was special about Brian
26:25
chesky and Joe gbia from Airbnb was not
26:28
that they were experts in technology
26:30
they went to art school they were
26:32
experts in design and perhaps even more
26:35
importantly they were really good at
26:37
organizing people and getting projects
26:39
done um so you don't have to work on
26:43
technology per se so long as you work on
26:45
things that stretch you what kind of
26:48
things are those now that is very hard
26:50
to answer in the general case history is
26:53
full of examples of young people who
26:55
were working on problems that no one
26:57
else at the time thought were important
26:59
and in particular that their parents
27:01
didn't think they were were important um
27:03
on the other hand history is even
27:06
Fuller of CA of examples of parents who
27:09
thought their kids were wasting their
27:10
time and who were right so how do you
27:14
know if you're working on real
27:17
stuff I mean when Twitch TV switched
27:20
from being uh Justin.tv to Twitch TV and
27:23
they were going to broadcast people
27:25
playing video games I was like what but
27:28
turned out to be a good business um well
27:32
I know I know how I know real problems
27:35
are interesting and I'm self-indulgent I
27:37
like I'm always interested in working on
27:39
interesting things even if no one else
27:40
cares about them and I find it very hard
27:42
to make myself work on boring things
27:44
even if they're supposed to be important
27:46
my life is full of case after case where
27:48
I worked on things just because I was
27:50
interested and they turned out to be
27:52
useful later in some worldly way um why
27:55
combinator itself is something I only
27:56
did because it seemed interesting so I
27:59
seem to have some sort of internal
28:00
compass that helps me out but this is uh
28:03
for you not me and I don't know what you
28:05
have in your heads maybe if I think more
28:07
about it I could come up with some
28:08
heuristics for recognizing genuinely
28:11
interesting ideas but for now all I can
28:14
give you is the hopelessly question
28:17
begging advice um incidentally this is
28:20
the actual meaning of the phrase begging
28:22
the question the hopelessly question
28:24
begging advice that if you're interested
28:26
in genuinely interesting problem
28:28
problems gratifying your interest
28:30
energetically is the best way to prepare
28:32
yourself for a startup and for that
28:34
matter probably the best way to live but
28:37
although I can't explain in the general
28:39
case what counts as an interesting
28:40
problem I can tell you about a large
28:43
subset of them if you think of
28:45
Technology as something that's spreading
28:47
like a sort of fractal stain every point
28:50
on the edge that's moving represents an
28:52
interesting problem steam engines not so
28:55
much although maybe you never know um
28:59
so one guaranteed way to turn your mind
29:01
into the type that startup ideas form in
29:04
unconsciously is to get yourself to the
29:06
Leading Edge of some technology to to as
29:09
Paul buite put it um live in the future
29:12
and when you get there ideas that seem
29:15
uncannily precient to other people will
29:18
seem obvious to you you may not realize
29:21
their startup ideas but you'll know
29:22
there's something that ought to exist
29:24
for example back at Harvard back in the
29:26
mid90s a fellow Brad student of my
29:29
friends Robert and Trevor wrote his own
29:31
Voiceover IP software um it wasn't meant
29:34
to be a startup he never tried to turn
29:36
it into one he just wanted to talk to
29:38
his girlfriend in Taiwan without paying
29:40
for long-distance calls um and since he
29:43
was an expert on networks it seemed
29:46
obvious to him that the thing to do was
29:48
turn the sound into packets and ship
29:50
them over the Internet for free why
29:52
didn't everybody do this well because
29:53
they weren't good at writing that kind
29:55
of software um he never did anything
29:59
with this he never tried to turn it into
30:00
a startup but that is how all the best
30:03
startups tend to happen um so strangely
30:06
enough the optimal thing to do in
30:08
college if you want to be a successful
30:09
startup founder is not some sort of new
30:12
vocational version of college focused on
30:16
entrepreneurship it's the classic
30:18
version of college is education for its
30:20
own
30:21
sake if you want to start a startup what
30:23
you should do in college is learn
30:26
powerful things and if if you have
30:28
genuine intellectual curiosity that's
30:31
what you'll naturally tend to do if you
30:33
just follow your own inclinations the
30:35
component of Entrepreneurship I can
30:37
never quite say that word with a
30:39
straight face um that really matters is
30:43
domain
30:44
expertise Larry pagee is Larry Page
30:46
because he was an expert on search and
30:49
the way he became an expert on search
30:50
was because he was genuinely interested
30:52
in it not because of some ulterior
30:54
motive at its best starting a startup is
30:58
merely an alterior motive for curiosity
31:00
and you'll do it best if you introduce
31:02
the ulterior motive at the end of the
31:04
process so here is the ultimate advice
31:08
for young wouldbe startup Founders
31:10
reduced to two words just
31:13
learn all right how much time do we have
31:15
left 18 minutes 18 minutes of questions
31:18
good god um do you guys have the
31:21
questions uh sure we'll start with uh
31:23
two questions online or we can start
31:26
with audience you guys are in charge
31:27
whatever you want want to
31:29
do uh okay um so to start with online
31:34
questions uh the most voted question
31:36
today was do I have to repeat them by
31:38
the way how can a non-technical Founder
31:41
most effectively contribute to a startup
31:43
how can a non-technical Founder most
31:45
effectively contribute to a startup um
31:48
well if the startup is if the startup is
31:52
working in some domain um if it's not a
31:54
pure technology startup but is working
31:56
in some very specific domain
31:59
um like if it's Uber right and and the
32:02
and the non-technical founder was an
32:04
expert in the limo business then
32:06
actually the non-technical founder are
32:08
probably doing most of the work
32:10
recruiting drivers and doing whatever
32:12
else Uber has to do right and the
32:14
technical founder would be just writing
32:15
the iPhone app which is probably less
32:18
well iPhone and Android app um which is
32:21
less than half of it um if it's a purely
32:24
technology startup the the non-technical
32:28
found ER does
32:29
sales and brings coffee and
32:32
cheeseburgers to the
32:34
programmer okay
32:36
nice oh audience okay audience yes do
32:41
you see um any value in business school
32:44
for people who want to pursue
32:46
entrepreneur entrepreneurship do I see
32:49
any value or if so what value I guess
32:51
I'm hoping we probably won't have to get
32:53
to the second
32:55
question which I suppose is the answer
32:57
to your first first question um okay so
33:00
the question is is there any value in
33:02
business school if you're interested in
33:04
entrepreneurship and if so what um so
33:08
basically no I mean it sounds
33:10
undiplomatic but the the point B what
33:12
business school was designed for is to
33:15
teach people management right and
33:18
management is a problem that you only
33:20
have in a startup if you're sufficiently
33:22
successful um so really what you need to
33:26
what you need to know early on to make a
33:28
startup successful is developing
33:30
products you'd be better off going to
33:32
design school um if you want to go to
33:34
some sort of school although frankly the
33:36
way to learn how to do it is just to do
33:37
it you know one of the things I got
33:40
wrong early on uh is that I advised
33:43
people who were interested in starting a
33:44
startup to go work for some other
33:46
company for a few years before starting
33:48
their own but honestly the best way to
33:50
learn how to start a startup is just to
33:51
try and start it you may not be
33:53
successful but you'll learn faster if
33:55
you just do
33:56
it so not
33:59
really um Business Schools are trying
34:02
really hard to do this but honestly like
34:05
they were they were designed to train
34:06
the officer Corp of large companies
34:09
right which is what business seemed to
34:11
be back when it was a choice of either
34:13
the officer Corp of large companies or
34:14
Joe's shoe store and and then then there
34:17
was this new thing Apple which started
34:19
out as small as Joe's shoe store and
34:21
then like turns into this giant mega
34:23
company um but they weren't designed for
34:25
that world and like let they're good at
34:27
what they're good at they should just be
34:28
do that and screw this whole
34:30
entrepreneurship thing just because it's
34:31
cool yes you said management is a
34:35
problem that you have only when you're
34:36
successful so what about those first two
34:39
or three people as a Founder that you're
34:40
trying to manage okay so management is a
34:43
problem only if you're successful what
34:45
about those first two or three people
34:47
well ideally you're successful before
34:50
you even hire two or three people right
34:53
um didn't didn't you say sam that Airbnb
34:55
took five months to hire their first
34:58
employee um so ideally you don't have
35:02
even two or three people for quite a
35:04
while uh when you
35:07
do you can you can sort
35:09
of the the first hires in a startup are
35:13
almost like Founders they should be
35:15
motivated by the same things they can't
35:18
be people they can't be people you have
35:20
to like manage right um this is like
35:23
that this is not like the office or
35:25
something like that these are your these
35:27
have to your peers really um you
35:30
shouldn't have to manage them much so is
35:32
it just a big no no like someone has to
35:34
be manag no way they can be on
35:37
the uh well Never Say Never the qu I
35:40
should repeat all these questions so if
35:41
someone has to manage no way they should
35:43
be on the founding team um in the case
35:47
where you're doing something where you
35:49
need some sort of super Advanced
35:50
technical thing and there's some bofin
35:54
um if you know that word who knows this
35:56
thing and nothing else in the world
35:58
including like how to wipe his mouth um
36:00
then it may be to your advantage to hire
36:03
said bofin and wipe his mouth for him um
36:06
but
36:08
uh as a general rule you want people who
36:11
are um sort of selfmotivated early on
36:15
they should be just like
36:17
Founders yes questions in the far back
36:21
do you think we're currently in a bubble
36:23
do I think we're currently in a bubble
36:25
um okay so I'll give you two answers to
36:28
this question uh uh one ask me questions
36:33
that are useful to this audience because
36:35
these people are here to learn how to
36:36
start startups and I have in my head
36:38
like more data about that than anybody
36:39
else and you're asking me the kind of
36:41
questions a reporter does because they
36:42
can't think of anything interesting to
36:44
ask um but I will answer your question
36:47
um there is a difference between prices
36:50
merely being high and a bubble a bubble
36:53
is a very specific form of prices being
36:55
high where people knowingly pay high
36:57
prices for something in the hope that
37:00
they will be able to unload it later on
37:01
some greater fool right and that's what
37:04
happened in the late '90s like in VCS
37:06
knowingly invested in [ __ ] startups
37:09
thinking that they would be able to take
37:11
those things public and unload them on
37:12
the retail investors before everything
37:14
blew up and I was there I was there for
37:17
that um at the epicenter of it all and
37:21
that is not what's happening today
37:22
prices are high valuations are high but
37:25
valuations being high does not mean a
37:27
bubble every every commodity has prices
37:29
that go up and down in some sort of sine
37:31
wave um definitely prices are high uh
37:34
and so we tell people if you raise money
37:36
don't think the next time you raise
37:38
money it's going to be so easy who knows
37:40
maybe between now and then the Chinese
37:42
economy will have exploded and there's
37:43
another giant disastrous recession who
37:45
knows assume the worst um but bubble
37:50
no yes I'm noticing a trend among um
37:54
young people and successful
37:56
entrepreneurs um where they don't want
37:58
to start one more great company they
38:00
want to start like 20 and so you're
38:03
starting to see the rise of these Labs
38:05
attempts where they're going to try to
38:07
launch a bunch of stuff I don't know a
38:09
really Stellar example yet but like
38:12
ideal like no like like like IDE lab
38:17
like gar Camp's new one there's North oh
38:20
yeah yeah okay so there's this new thing
38:23
where people start Labs that are
38:25
supposed to spin-off startups um it
38:28
might work that's how Twitter started in
38:31
fact it I meant ideal lab not Ideo that
38:33
was another Freudian slip um because
38:36
Twitter was not Twitter at first Twitter
38:38
was the side
38:40
project side project at a company called
38:43
Ideo That was supposed to be in the
38:45
podcasting business you're like
38:47
podcasting business do those words even
38:49
grammatically fit together um the answer
38:51
turned out to be no as Evan discovered
38:54
um but like as a side project they spun
38:55
off Twitter and boy was that uh dog
38:58
wagging tail um so people are starting
39:01
these things that are supposed to
39:02
spin-off startups will it work quite
39:04
possibly quite possibly if the right
39:06
people do it um you can't do it yet
39:09
though because you have to do it with
39:10
your own money um yes far back I think
39:13
so much um I hate to step into sort of a
39:15
gendered pitfall but what advice do you
39:17
have for female co-founders as you're
39:19
pursuing funding female co-founders when
39:22
you're pursuing funding well it probably
39:24
is true that women have a harder time
39:26
raising money right um I've noticed this
39:29
empirically and um Jessica is just about
39:33
to publish a bunch of interviews with
39:34
female Founders and a lot of them said
39:37
that they thought they had had a hard
39:38
time a harder time raising money too um
39:42
so what I would you remember I said like
39:44
the way to raise money make your startup
39:47
actually do well well that's just
39:49
especially true in that case if you have
39:52
any if you if you miss the ideal Target
39:55
from the VC's point of view in any
39:56
respect the way to solve that problem is
39:59
make the startup do really well um so in
40:01
fact there was a point like a year or
40:03
two ago when I tweeted this growth graph
40:05
of this company and I didn't say who
40:07
they were um but I knew it would start
40:09
people asking and it was actually a
40:10
female founded startup that was having
40:12
trouble raising money but their growth
40:14
graph were stupendous and so I tweeted
40:16
it knowing that like all these VCS would
40:18
start asking me who who is that right
40:21
and like growth graphs have no gender
40:23
right so if they see the growth graph
40:24
first let them fall in love with that
40:26
right um
40:29
so do well which is good advice for
40:31
startups generally yes what would you
40:33
learn in college right now what what
40:35
what would you learn in college right
40:36
now what would I learn in
40:39
college
40:41
H literary Theory no just kidding um
40:46
let's see
40:48
um well you know honestly I think I
40:51
might try and study Physics I feel
40:53
that's the thing I feel like I missed
40:55
for some reason like I was all excited
40:58
when I was a kid computers were the
41:00
thing maybe they still are right so I
41:01
sort of got very excited about writing
41:04
code um and you could do things with you
41:06
could do you could write real programs
41:08
in your bedroom you can't like build
41:09
real in your accelerators maybe you can
41:12
um so maybe maybe physics I feel kind of
41:15
I always I have I sort of look long I
41:17
notice I sort of look longingly at
41:19
physics um so maybe but I don't know if
41:23
that's what am I saying I'm saying I was
41:25
about to say I don't know if that's
41:26
going to be helpful starting a start up
41:28
and I've just told you to follow your
41:29
own curiosity so who cares if it's
41:31
helpful it'll turn out to be helpful
41:34
questions uh another question online uh
41:36
what are your recurring systems in your
41:38
work and personal life that make you
41:40
efficient ah boy okay so what are my
41:43
recurring systems in my work and
41:45
personal life that make me
41:47
efficient um well having kids is a good
41:50
way to be efficient because you have no
41:52
time left um so if you want to get
41:54
anything done um the amount of done you
41:56
do per time is high hi um so uh actually
42:01
many parents many startup Founders who
42:04
have kids have made that point
42:05
explicitly it causes you to focus um
42:08
because you have no choice uh let's see
42:11
but that's not I wouldn't actually
42:12
recommend having kids just to make you
42:14
more
42:15
focused so
42:18
uh let's see um you know I don't think
42:24
I'm very efficient I have two ways of
42:25
getting work done one is like
42:27
ycombinator the way I worked on YC
42:29
combinator is I was forced to right like
42:31
there there was an I had to set the
42:33
application deadline and then people
42:35
would apply and then there were all
42:36
these applications that I had to respond
42:38
to by a certain time so I had to read
42:39
them and I knew if I read them badly we
42:42
would get bad startups so I tried really
42:43
hard to read them well right so I set up
42:45
this situation that forced me to work um
42:48
the other kind of work I do is like
42:49
writing essays and I do that
42:51
involuntarily I'm like walking down the
42:53
street and the essay starts writing
42:55
itself in my head um and so really I
42:59
either force myself to work on less
43:02
exciting things um or the I can't help
43:06
working on um on exciting things and I
43:10
don't I don't have any useful techniques
43:12
for making myself efficient sorry um if
43:16
you work on things you like you don't
43:17
have to force yourself to be efficient
43:19
yes I'm think it's a good time to turn a
43:21
side project into a real startup when is
43:24
a good time to turn a side project into
43:26
a startup you will know
43:31
right it starts you you when you okay so
43:34
the question is when do you turn a
43:35
startup into a side project into a real
43:38
startup you will know that it's becoming
43:40
a real startup when it takes over an
43:42
alarming percentage of your
43:43
life right um like when you find like my
43:46
god I've just spent all day working on
43:48
this thing that's supposed to be a side
43:49
project I'm going to fail all my classes
43:51
what am I going to do right um then
43:54
maybe it's turning into a startup yes
43:56
wait I already answered your question I
43:58
should ask I should let somebody else
43:59
ask I may get back to you yes so I I
44:02
know you talked a lot earlier about
44:03
you'll know when you're started was
44:05
doing extremely well but I feel like in
44:06
a lot of cases there's a bit of a gry
44:08
line where it's like you know you have
44:09
some users but maybe not it's explosive
44:11
go chart isn't up to the right how what
44:14
would you do or what would you recommend
44:15
doing in those situations considering
44:16
you're allocating time and resources how
44:18
do you balance and okay when to Startup
44:21
is sort of growing but not much um
44:24
didn't you tell them they were supposed
44:25
to redo things that don't scale you sir
44:28
have not done the readings and you are
44:29
busted therefore because I wrote a whole
44:31
essay and answer to that question and
44:33
that is it do things that don't scale so
44:35
just go read that um because I can't
44:37
remember everything I said but it's
44:40
about exactly that problem yes back
44:43
there what kind of startups should not
44:45
go through incubation in your opinion
44:47
what kind of you mean do you mean
44:48
incubation LC no yeah YC or what kind of
44:52
startup should not go through YC um
44:55
definitely any that will fail
45:00
um and or if you'll succeed but you're
45:03
an intolerable intolerable person um
45:06
that also Sam would probably sooner do
45:08
without um short of that I I cannot
45:13
think of any because a large percentage
45:16
Founders are often surprised by how
45:18
large a percentage of the problems that
45:20
startups have are the same regardless of
45:22
what type of thing they're working on
45:24
and those tend to be the kind of
45:26
problems that YC helps with the most not
45:27
the ones that are domain specific so I
45:30
can't is there do you can you think of a
45:32
class of startups that YC wouldn't work
45:34
for I can't I mean we had like fision
45:37
and fusion startups in the last batch or
45:41
they had um yes um you mentioned that
45:44
it's good advice to learn a lot about
45:46
something that matters what are some
45:47
good strategies to figure out what
45:49
matters well if you think of okay so how
45:52
do you figure out what matters um if you
45:55
think of Technology as something that's
45:57
spreading as a sort of fractal stain
46:00
anything on the edge represents an
46:02
interesting idea sounds familiar um so
46:06
how do you figure out um like I said
46:09
that was the problem you have correctly
46:11
identified um the thing where I didn't
46:13
really answer the question or I gave
46:14
this question begging answer I said like
46:17
that I just like I'm interested in
46:18
interesting things if you're interested
46:19
in interesting things work on them and
46:21
things will all work out right um but
46:23
how do you tell what is a real problem I
46:27
don't know that's like important enough
46:28
to write a whole essay about and I don't
46:30
know the answer I probably should write
46:31
something about that but I don't know I
46:33
don't know um I figured out a technique
46:36
for detecting whether you have a taste
46:38
for genuinely interesting problems um
46:41
which is whether you find working on
46:42
boring things intolerable right and
46:45
there are known boring things like
46:47
literary Theory and working in middle
46:49
management in some large company so if
46:51
you could tolerate those things then you
46:52
must not you must either have stupendous
46:54
self-discipline or um you don't have a
46:57
taste for genuinely interesting problems
46:59
um and vice versa yes one more question
47:02
okay one more question it better be good
47:04
I was just wondering how like Snapchat
47:06
like makes money refs to do so Snapchat
47:09
what do I know about
47:11
Snapchat we didn't fund
47:13
them can I have about another
47:16
question all right go
47:19
ahead talking about hiring people that
47:23
like but it seems like that could lead
47:25
to a monoculture and there a lot of
47:26
benefits culture but how do you deal
47:28
with the blind spots that arise okay so
47:31
if you hire people you like you might
47:32
get a monoculture and how do you deal
47:34
with the blind spots that arise um
47:36
starting a startup is something where
47:38
many things will be going wrong you
47:39
can't expect it to be perfect and the
47:41
advantage is of of hiring people that
47:43
you know and like are far greater than
47:46
the you know the small disadvantage of
47:48
disadvantages of having some monoculture
47:50
you look at it empirically all the most
47:52
successful startup someone just like
47:54
hires all their Pals out of college all
47:56
right you guys thank you
48:00
[Applause]
Start a conversation
Ask anything about Before the Startup with Paul Graham
100/100 free messages remaining
AI responses may be inaccurate. Verify important information.
Before the Startup with Paul Graham
Y Combinator