|
Maybe I'm the only one here, but I was hoping for
more in details, information, examples, and insight
into how to find a good idea for a startup. I really can see that from his experience in Silicon
Valley PG gave a view of what he has actually seen
in how startup ideas are or have been found. I
have to respect most of what he said. Still, from some of what he said, I have a tough time
taking him literally: E.g., as I recall from the
lecture, he indicated that early on Google was just
a little thing, just a small interest, or some such. However, also, as is well-known, even in the early
days of Google even as just an idea, Internet
search was already widely regarded as an important
need and problem. Without checking dates, there
was, what, Lycos and Yahoo!, maybe more? Without means of search, the Internet was like being
in the world's biggest library with all the books,
etc. just in a big pile. For libraries, the
solution has long been the library card catalog,
especially the subject index. So I have to suspect
that, even in the beginnings of the idea for Google,
the founders did understand that what they were
working on was possibly important as a business. Next, the lecture did mention that Ron Conway wrote
Google an early check: I have to believe that
Conway took the future of Google fully seriously. From all I know, Facebook, Microsoft, Cisco, Intel,
etc. all knew early on, at the beginning of the
idea or not long after, that they were trying to
make a successful business. But the lecture seemed to be suggesting that at the
start an idea is hardly more than a toy or
curiosity. Next there was, > The way to get startup ideas is not to think about
having good startups ideas: you’ll get bad and
plausible ideas that will end up wasting your time. > To have a good idea, take a step back and don’t
even try. While I can believe that PG has plenty of examples
where startups followed this advice and were
successful, still I'm shocked and have a tough time
swallowing that this advice is the only way to be
successful or that, in particular, always any effort
to find a good idea > will end up wasting your time. Instead, I'd like to build on PG's > What you really need to learn are your users’
needs. Now I'm on board! But then I want to go farther: Okay, not everyone
has to go along with my next thinking, and here is
some of why: First, there's not just one way to
"skin a cat" or be successful. Second, and
something I regard as very important for successful
startups, as we know too well, a really successful
startup is rare, so rare that we have to suspect
that, not all, but some of the most successful
startups will be taking approaches that are new and
with few or no examples from the past, say, 30 years
in Silicon Valley venture funded startups. Moreover, if a startup looks close to what has been
tried, and even worked, before, then we have to be
suspicions and question if that "what" should be
tried again instead of something new. But, broad strokes, aside, let's try to be more
definite: Step (1) Need. Assume we've seen some users' need. Suppose we use
some judgment here and accept only a genuine need.
Such a need might be, say, a safe, effective, cheap
one pill taken once to cure any cancer. Yup, that's
definitely a need -- no question. We want a need so
serious that we can be quite sure, e.g., the pill,
that the first good or a much better solution will
be a must have and not just a nice to have. So,
we pick a need. Now I narrow the scope: We want to exploit Moore's
law and the Internet and do have a need that, at least
for its solution, is in information technology.
One reason: Can send to the users just bits, and
those sent just over the Internet. Step (2) Solution. With the need, we try to find the first good or a
much better solution. If we fail, as we likely
would for years on a cure for cancer, or, in
information technology, a project were we needed an
algorithm that shows that P = NP, we return to Step
(1) and find another genuine need. Step (3) How to So, how can we find a solution? Well, here's my suggestion: Right, even with PG's
broad experience, the suggestion may so far never
have been seen even by PG in Silicon Valley. So, in
this sense we're talking something new, and we keep
in mind that we should not be reluctant to pursue
something new since being new and different is from
maybe good up to crucial for the rare, big successes
in the future. But for something new, we very much want some
evidence of, say, efficacy, or will that dog hunt? So, we have a problem, and we want a solution, in
information technology, and a solution we have high
confidence in, even before we start writing
software. So, where can we find examples of doing
that? Sure: The US DoD. They've been doing that
with, comparatively, a great batting average for 70+
years, that is, finding solutions that well informed
people can have high confidence in before starting
implementation. So, we have Lesson 1: By example, it's actually possible,
and in the US DoD quite common, nearly standard, to
find solutions that well informed people can have
high confidence in before starting implementation. Now, this lesson, and the fairly obvious DoD
examples, e.g., the SR-71, were often quite
expensive to implement. But, there is also Lesson 2: The implementation work commonly was
routine, that is, low risk, maybe expensive but,
still, low risk. More of my suggestion: It can be good, when we can,
to convert the work of finding the solution to a
mathematical problem where, if the math is correct,
then we are quite sure the solution is correct and,
quite likely, effective as a solution for the need. Advantages here include: (A) The math has a good
chance of being done just on paper by just one
person, with very low "burn rate". (B) The math is
comparatively easy to check for correctness. (C)
Since we have assumed that our startup is in
information technology, a solution based on math has
a big natural advantage; e.g., maybe the
implementation is just software with no metal
bending, team of thousands, big buildings, etc. So (1)-(3) is my suggestion for insight into how to
find a good idea for a startup. So, where is this suggestion for finding a startup
idea unclear, wrong? |