Good news: I can agree with some of the
OP.
Much better news: I do believe that it's
fairly obvious that there are good
solutions to the most important problem
mentioned in the OP.
First a remark on scope: I'm talking
about information technology (IT) startups
based heavily on Moore's law, the
Internet, other related hardware,
available infrastructure software, etc.,
and I'm not talking about bio-medical
technology which I suspect is quite
different.
Second, a remark on methodology: When the
OP says "almost certainly" and similar
statements about probability, sure, (A) in
practice he might be quite correct but
(B), still the statement is nearly always
just irrelevant.
Why irrelevant? Because what matters is
not the probability, say, estimated across
all or nearly all the population, or all
of business, or all of startups, or even
all of IT startups. Instead, what is
important, really crucial, really close to
sufficient for accurate investment
decision making, is the conditional
probability given what else we know. When
the probability is quite low, still the
conditional probability -- of success or
failure -- given suitable additional
events, can be quite high, thus, giving
accurate decision making. So, net, what's
key is not the probability but what else
is known so that the conditional
probability of the event we are trying to
evaluate, project success or failure,
given what else we know is quite high.
So, back to the OP. We can start with the
statement:
> The absolute minimum to play the game
even once is about $5-10k, and if that's
all you have then you will almost
certainly lose it.
Here for the "almost certainly" to be true
needs to depend on what else is known.
Sure, if not much more is known, then
"almost certainly lose it" is correct.
But with enough more known, the first
investment can still likely be a big
success.
The big, huge point, first investment or
101, is what else is known.
> There is a small cadre of people who
actually have what it takes to
successfully build an NBT, and experienced
investors are pretty good at recognizing
them.
I agree with the first but not with the
second. From all I can see, there is
hardly a single IT investor in the US who
knows more than even dip squat about how
to evaluate an IT investment. E.g.,
commonly the investors were history or
economics majors and got MBA degrees.
Since I've been a prof in an MBA program,
I have to conclude that a history or
economics major with an MBA has no start
at all evaluating IT projects.
Here is huge point:
We can outline a simple recipe in just
three steps for success as an IT startup:
(1) Find a problem where the first good or
a much better solution will be enough
nearly to guarantee a great business,
e.g., the next big thing.
(2) For the first good or much better
solution, exploit IT. Also exploit
original research in high quality, at
least partly original, pure/applied
mathematics. Why math? Because the IT
solution will be manipulating data; all
data manipulations are necessarily
mathematically something; for more
powerful manipulations for more valuable
results, by far the best approach is to
proceed mathematically, right, typically
with original work based on some advanced
pure/applied math prerequisites.
(3) Write the corresponding software, get
publicity, go live, get users/customers,
get revenue, and grow the revenue to a
significant business.
So, right: Step (2) is a bottleneck: The
fraction of IT entrepreneurs who can do
the math research is tiny. The fraction
of startup investors who could do an
evaluation of that research or even
competently direct such an evaluation is
so small as to be essentially zero.
So, net, the investors in IT are condemned
to miss the power of step (2) and, thus,
flounder around in nearly hopeless mud
wrestling in a swamp of disasters. And,
net, that's much of why angel investors
lose money.
So, the main problem in the OP was losing
money on IT projects. The main solution,
as both an investor and an entrepreneur,
is to proceed as in steps (1)-(3).
For IT venture capitalists (VCs), they
can't use step (2) either, e.g., can't do
such work, can't evaluate such work, and
can't even competently direct evaluations
of such work, but they have a partial
solution: Likely enforced by their LPs,
in evaluating projects they concentrate on
cases of traction and want it to be
significantly high and growing rapidly.
So, with this traction criterion, and some
additional judgment and luck, some of the
VCs get good return on investment (RoI),
but they are condemned to miss out on step
(2).
So, what is the power of step (2)? As we
will see right away, clearly it's
fantastic: Clearly with step (2) we can
do world changing projects relatively
quickly with relatively low risk.
The easiest examples to see of the power
of step (2) are from the US DoD for US
national security. Some of the best
examples are the Manhattan Project, the
SR-71, GPS, the M1A1 tank, and laser
guided rockets and bombs, all relatively
low risk projects with world changing
results. Each of these projects, and many
more, was heavily dependent on step (2)
and met a military version of steps (1)
and (3).
More generally, lots of people and parts
of our society are quite good at
evaluating work such as in step (2) and
proposals for such work, just on paper.
We can commonly find such people as
professors in our best research
universities and editors of leading
journals of original research in the more
mathematical fields.
I started some risky projects, e.g., an
applied math Ph.D. from one of the world's
best research universities. From some
good history, only about one in 15
entering students successfully completes
such a program. The completion rate of
applied math Ph.D. programs makes the Navy
Seals and the Army Rangers look like
fuzzy, bunny play time. With much of my
Ph.D. program at risk, I took on a
research project. Two weeks later I had a
good solution, with some surprising
results, quite publishable. Later I did
publish in a good journal. I could have
used that for my Ph.D. research, but I had
another project I'd pursued independently
in my first summer -- did the original
research then, in six weeks. The rest of
that work was routine and my dissertation.
While working part time, the Navy wanted
an evaluation of the survivability of the
US SSBN fleet under a special scenario of
global nuclear war limited to sea, all in
two weeks. I did the original applied
math and computing, passed a severe
technical review, and was done in the two
weeks. Later I took on a project to
improve on some of our work in AI for
detection of problems never seen before in
server farms and networks. In two days I
had the main ideas, and a few weeks later
I had prototype software, nice results on
both real and simulated data, and a paper
that was publishable -- and was published.
My work made the AI work look silly; it
was. Once in a software house, we were in
a competitive bidding situation. I looked
at what the engineers wanted and saw some
flaws. Mostly on my own, I took out a
week, got good on the J. Tukey work in
power spectral estimation, wrote some
software, and showed the engineers how to
measure power spectra and how to generate
stochastic process sample paths with that
power spectrum. As a result, my company
won sole source on the contract. So,
before I did these projects, they all were
risky, but I completed all of them without
difficulty.
Lesson: Under some circumstances, it's
possible to complete such risky projects,
given the circumstances, with low risk.
But IT VCs can't evaluate the risk before
the projects are attacked or even evaluate
the results after the projects are
successfully done. So IT VCs fall back on
traction.
I confess: It appears that the IT VCs are
not missing out on a lot of really
successful projects. Well, there aren't
many IT startups following steps (1)-(3).
So, for IT success, just borrow from what
the US military has done with steps
(1)-(3).
The problem and the opportunity is that
nearly no IT entrepreneurs and nearly no
IT investors are able to work effectively
with steps (1)-(3), especially with step
(2).
The IT VCs have another problem: The know
that for the next big thing -- Microsoft,
Apple, Cisco, Google, Facebook -- they are
looking for something exceptional. And
they know that those for examples have
very little significant in common. Still
the IT VCs look for patterns for hot
topics at the present or recent past.
That's no way to find the desired
exceptional projects. E.g., when the US
DoD wanted the Manhattan Project, they
didn't go to the best bomb designers of
the previous 20 years; doing so would not
have resulted in the two atomic bombs that
ended WWII. Instead, the US DoD listened
to Einstein, Szilard, Wigner, Fermi,
Teller, etc., none of whom had any
experience in bomb design.