Editor's Note: A few hours after the news broke regarding
VA Linux's decision to leave the hardware business in favor of
concentrating on its SourceForge OnSite software and its OSDN web
properties, we wrote Eric Raymond with a few questions. Mr.
Raymond, in addition to being a "noted open source evangelist,"
also sits on the VA Linux Board of Directors, so we were able to
get his take not only on the immediate picture from the ground but
the broader picture of just what VA Linux's decision means for
Linux and open source software in general. We also asked a question
or two about yesterday's other big story: Microsoft's decision to
support FreeBSD in its .NET initiative.
Mr. Raymond couldn't answer a few of our questions without
making forward-looking statements... something he's constrained
from doing as an officer of VA Linux.
LT: What does this mean about Linux-oriented
hardware specialists? Penguin Computing is still in place while a
couple of smaller and more specialized operations have either not
made announcements recently enough to indicate that they're really
still in business or simply dropped out altogether. Will the
industry ever need a Linux hardware specialist? Does VA Linux's
inability to succeed in this area represent uptake of Linux by the
real giants (Compaq, IBM, and Dell) on a scale that precludes the
need for Linux "pure-plays"?
ESR: The world used to need Linux hardware
specialists, but times have changed. In 1998 stuff like
server-ready SMP motherboards was still exotic hardware and Linux
was still novel; having expertise in either was unusual and being
able to sell the combination was an attractive business, especially
with an Internet buildout in full roar.
But a funny thing happened on the way to the bank. Linux
succeeded too well, and attracted too much money. VA helped that
happen; it built damn fine hardware, it subsidized a lot of
open-source projects, sponsored a lot of shows, and did marketing
into places the Linux community never reached before. I'm proud to
be associated with the company that did those things. We're glad we
helped make Linux a thumping success.
But the result of that success was that established companies
with a lot more financial mass started hunting the same customers
VA was after -- offering Linux on their boxes and cultivating the
community in the same way VA did. VA's hardware business got
increasingly squeezed between the likes of IBM and Dell on one
hand, and lots of tiny screwdriver shops on the other.
Even with that problem, I think VA's original business plan
might have gotten it to profitability on schedule. While the
technology boom was still booming, there was enough demand to go
around for everyone -- and the dot-coms, mostly run by bright young
people who liked Linux, loved VA.
In fact, the dot-coms very nearly loved VA to death. When the
bubble popped, a big chunk of VA's customer list evaporated. The
company tried to refocus on selling hardware to larger customers,
and probably would have succeeded given time. But lots of time is a
luxury you don't get when you're spending VC money -- and
especially not when the whole economy has gone into something very
VA's customer base outside the dot-coms is pretty loyal, too --
it's a company people feel good about doing business with. Given
another nine months of economic boom, I think the company might
well have grown enough to make the hardware business work even in
spite of margin erosion and the competition.
Instead, VA's hardware business got double-whammied, once by the
side effects of Linux's success and then by the macroeconomic bust.
It's really sad. We had the people, we had the drive, we had a lot
of good karma. We just didn't get the time.
LT: On a broader level, what does this mean for
Linux in general? A couple of years ago, it seemed as if the
question many were asking wasn't so much whether Linux and open
source software could get a piece of the pie: people seemed more
convinced that there was a whole new pie to be carved up. Now
there's been some culling and attrition, and companies are either
specializing in order to cater to the needs of larger companies
(like SuSE working so closely with IBM on high-end computing) or
they're diversifying (Red Hat making big moves in embedded
computing while introducing new services offerings and a database
product). In other words, it seems as if we've returned to a sense
that there's really only one pie after all, and that the Linux
market has a constrained piece of that pie. How much of the
contraction we're seeing in Linux companies is a result of broader
economic forces, and how much is a cooling of enthusiasm over
ESR: To quote some politician or other, "It's
the economy, stupid." :-)
Don't take that personally, because you're asking an intelligent
question that doesn't have a simple answer. Yes, I think there has
been some cooling of the initial enthusiasm for Linux -- Wall
Street's trendoids are notoriously fickle that way, always looking
for the next buzzword.
But I think that has only been a minor factor in the recent
troubles at VA and elsewhere. All the market surveys and
projections still show Linux as a technology doing extremely well
against its competition. The trouble is that capital spending is in
collapse, hardware is commoditizing, and it's getting increasingly
hard to find decent profit margins anywhere you turn. It's tough
all over, and Linux companies aren't the only or the worst
In a macroeconomic rainstorm like this, the best you can do is
hunker down, minimize the burn rate, work on getting the
close-to-profitable bits of the business really profitable, and
hope the sun comes out soon. That's what VA is doing. We want to be
there for our stockholders and stakeholders, with our skills still
sharp and our commitment to open source still powerful, when the
next expansion begins.
LT: On a related note, aren't the same forces
that caused VA Linux to find itself in trouble on the hardware side
similarly arrayed against it on the web content side? While no
one's going to buy into a "ZD-Dot" or a "Fresh|Net," aren't the
same risks of widespread acceptance present -- the risk that a
larger, better established player will just say "we're going to
cover Linux and open source computing in general as another piece
of the industry puzzle" and simply keep their existing reader base
intact by saying, effectively, "we know you're interested in this
'new' technology, and we're covering it with n reporters?"
ESR: I'm not sure I ought to answer this one.
Backing off from VA's business position a bit, I'll just observe
that it's a lot harder to corral Web audiences than it is to boost
hardware market share. Financial mass doesn't help in the same way;
you can't buy audience share by lowering the margin you'll accept.
Once a web community like eBay or Slashdot has reached a certain
critical size, it's hard to lure people away from it to a competing
service that's technically equivalent because most of the value to
them is in *that particular community's* traditions and its trust
ESR: I'm sure Microsoft's stamp of approval for
BSD is very sincere, and I'm also sure it's an attempt to frame
Linux out of the picture. The Windows TCP/IP stack is built around
BSD code; from Microsoft's point of view, the BSD crowd are a bunch
of suckers begging to be exploited again and again. I'm certain
Microsoft would love for the entire open-source world to turn into
an acquiescent source of free R&D for its monopoly.
That may sound like I'm lining up with those who think the GPL
is the open-source world's only defense against being co-opted by
greedy monopolists. But actually I think that both Microsoft and
the GPL zealots underestimate how powerful and how subversive
open-source development truly is. It's a profoundly disruptive
technology, much better matched to a world of commodity hardware
and cheap communications than the proprietary closed-source
Oddly enough the people who seem to grok what this means most
fully aren't techies but economists, people like Clayton
Christensen (author of "The Innovator's Dilemma") and Dan Shapiro
and Hal Varian (authors of "Information Rules"). What they get is
that the whole proprietary-software industry is probably a
transient phenomenon supported by high hardware profit margins.
It's had thirty good years, but the combination of open-source
development and Moore's law probably won't allow it many more. If
this is true, the BSD way would be the doom of Microsoft just as
surely as Linux; it just might take a little longer and involve
less overt confrontation.
As for looking at .NET? Sure, we should be looking at it -- the
same way the Samba guys look at SMB file shares. C# is a weak
language design, a sort of pale imitation Java. But that just makes
it easier to co-opt.