Subject: Systems Software is Relevant
From: Thomas Lord <>
Date: Sun, 21 Oct 2007 13:29:24 -0700

Stephen J. Turnbull wrote:
>  What I'm saying is that an alternative plausible hypothesis
> about the behavior of VCs in software markets is that it is in fact
> not possible to split up the risk.  Ie, a VC can only swallow projects
> whole.  The VC can sell shares, and thus securitize the risk in that
> way.  However, at the level of "what VCs do", somebody has to spend a
> big chunk of time managing the funded project, much more than
> diversified investors can do.  (I repeat, this is an hypothesis.)

Of course VCs want substantial amounts of control over funded
companies.   They are flippers: buy company, make improvements,
sell company.   The investment targets they specialize in are firms
that can grow non-linearly with scale but that lack internal cash and
almost always lack internal expertise to manage the cash needed
to grow.   So, VC comes up and buys shares and control, exercises
control to make the needed improvements, then exits.


VCs are limited by the range of investment offers they receive,
consider, understand, know what to do with, etc.  The funded
firms lack the expertise to build-out.   A correlary is that VC
firms live and die by developing and filling in that missing
expertise.   In some sense, software VCs compete to be the
the leading experts in "what's coming up, in software".

The professional literature and letters around all of this are amazing
stuff ranging from Gartner reports that at least establish clear
vocabulary for emerging fields, to the conferences, press releases,
op-ed pieces, and many, many conversations.

I'll let you in on a secret:  socially, all of the serious VC firms
are surrounded by a thick fog of executives-in-waiting.  This
sub-culture has ever-shifting membership but evolves slowly enough
to be stable.   If you get anywhere near this big fog you'll know it
because people start brokering introductions and group meetings
left and right (real and virtual) and typical conversations very often
revolve around the emerging consensus view of the industry in the
VC world:  what technologies are hot?  what tactics for bootstrapping
a start-up are hot?  who is having dinner with whom?

And a lot of conversations exploring whether nascent start-up idea
A is better than worse than or should be combined with start-up
idea B.

So, that fog of young executives in waiting (with its little annexes
like the world of campus researchers, a subset of politicians, etc)
is basically the matrix out of which VC deals are born.

Today, at least in software, that matrix runs almost entirely on
nepotism and nobody involved would seriously deny this.  In fact,
the few who are winning that game and the many who think they
will win any day now tend to like it this way, because nepotism is
a game that (by definition) everybody knows how to play -- it's

They're half right, too.  The rapid build-out that is required of
a funded start-up requires very careful ego management and
a team of people who share mutual trust and respect under stressful
conditions requiring physically and mentally demanding work for
very long hours.   If you aren't using nepotism to build those kinds of
teams your results are going to suck.

They're half wrong, though:

The fog crowd around VCs consists, appropriately enough, of people
who think that if the right idea comes along they can muster enough
seed funding and management experts to bootstrap a start-up.  It's
silly to try to put a precise dollar value on it but you could say that the
fog crowd folks have $100K lines of credit for a venture of their
choosing and a list of the most promising management candidates in
their neighborhood.   If you don't have those advantages, you aren't
in the fog crowd, period.    So what, though?

Well, that kind of potential credit and social connections isn't cheap
to maintain and the fog crowd spends a great deal of their time and
resources trying to outdo one another on those two fronts.  They're
*not* spending a lot of time thinking about what's possible with software.
Almost to a one, when it comes down the actual technology, they
are looking for plays that are:

~ technically easy
~ cost somewhere in the range $50K...$250K to bootstrap
~ have a "hook" -- are unexpected in some way (to leverage stealth)

In other words, they are *all* looking at the systems software
they inherited, looking for hacks worthy of a clever freshman
monkey, and then hyping how impressive it was that they alone
were the first to do it (and for only $75K!).

That's why we get overcrowding, combined with sluggish
advances in capability -- e.g., MySpace v. Facebook v. Google etc.
Gazillions of dollars to see who can write the most dazzling
javascript library (and a lot of ill-considered hot-air about
"social networking" and "web 2.0").

Historically, the software VC industry goes through little
bursts in growth from time to time  -- periods of time when
overcrowding is less of a problem because there are so many
new product opportunities.   During these periods, the
fog crowd shines and presents good deal after good deal.
Microcomputers, the Internet, etc..    Are these growth periods
random events?   No:

Every single one of those bubbles -- just about every one
you can name in the past 30 years (or more) -- all of them
correlate in history with disruptive advances in what we
software engineers  call "systems software" -- software
of use mainly to other programmers rather than software
that non-programmers are likely to directly use.

People know that, to an extent.   "Everyone knows", for
example, that it was the invention of the LAMP stack
(a systems programming breakthrough -- of no direct use
to non-programmers) that largely gave us the modern web
as we know it.    But, the fog crowd and the VCs don't seem
to know what to do with this.   They stand before it slack-jawed
and say "Gosh, advances like LAMP just happen randomly
but we're standing here to catch them when it happens."

Given a state of systems software in which it costs 200K
to bootstrap a photoblog, and another 200K to bootstrap
a video blog, etc.  -- given that the fog crowd will systematically
muster each of those 200K investments.   A systems programming
researcher, looking at that agenda, would instead spend 200K
to simplify the problem -- so that all of those variations on blogging
would become 20K plays, not 200K plays.   A systems programming
researcher (if he's any good) is a good source of ideas for what kinds
of simplifications might be possible.

So, why doesn't the systems programming research just spend
$220K and win the first start-up for himself?   Because $200K
of that would go into *research* not into building up the management
or hosting infrastructure and, after that sunk cost, it only takes
$20K to buy *1* lottery ticket to win from the research but only
one in ten lottery tickets pay off.   So, a systems programming researcher
who wants to hoard the start-up opportunities he creates has now
doubled his costs to something like a minimum half million to seriously
explore any one idea (cost averaged).   (Hello, Danny Hillis!)  Look
at me, though.   I'm making (and, in other forums) defending a
claim that my "human scale" approach is, currently, beginning to
uncover a new and very promising breakthrough in systems programming
researcher.   I'm sunk in by a good $200K easily (more, really).
I'm just at this interesting stage where, basically, the people I talk
with over the next couple of years are going to have a seriously
advantageous position.   I can't pick exactly which of the easy plays
I'm creating are going to win (or else I'd work harder at making one
myself) but I can help others do analysis in placing their bets
(and can fold their interests together with mine in terms of how
I spend hacking hours).