Eric Raymond In The Cab Going to The Bazaar
Dec 19, 1999, 02:52 (12 Talkback[s])
(Other stories by Paul Ferris)
Re-Imagining Linux Platforms to Meet the Needs of Cloud Service Providers
By Paul Ferris
Editor, Linux Today
The last morning of The Bazaar, I was able to catch up with Eric
Raymond in a cab on the way to the convention hall.
Linux Today: Eric, this is the first
Linux Today: I understand that you wrote
this piece of email or something, called the something something
bazaar, or something...
Eric: The Cathedral and the Bazaar.
Linux Today: Oh really?!? Then maybe your
paper has something to do with why the Bazaar is called "The
Eric: Well, I think so, yes....
Linux Today: Eric, This has to feel
Eric: It's part of a larger pattern of things
that feels good, which means that we're winning!
Linux Today: Yes! We're well on our way to
world domination. A lot of people just want to talk about Linux
being Linux without any long term goals. But I don't see that being
anything very negative. I just want to hear your thoughts on World
Eric: I think that the most important thing is
not Linux in itself dominating the world, but Open Source
infrastructure dominating the world. Solving the software quality
problem. It's because Linux right now is our most plausible means
to that end that I back it.
Linux Today: So, for example, the
BSDs... [BSD used in the context of this interview refers to
several alternative operating systems that are also Free Software or under various Open Source licensing schemes.
There are three popular variants, FreeBSD, NetBSD, and OpenBSD. LT ed.]
Eric: I would actually like to see them succeed
more than they are currently because mono-cultures are vulnerable.
If you get a mono-culture that has just one genetic line then the
first plague can wipe it out. I would like to see more diversity in
the open source community.
Linux Today: For example, what's going on
with Outlook right now?
Eric: I don't know about that...
Linux Today: That's how Melissa is
traveling, a current virus plaguing the non- open source and free
software communities. Because the mail program will run executables
automatically, because of poor user choice in some instances or in
others due to extremely poorly thought out security measures in the
mail client, this leaves people open to a one, two, three punch
The user gets a piece of mail that contains the virus
embedded in an executable or maybe it's a piece of Java script
that's taking advantage of a well-known hole. The virus infects the
current machine, reads information off of the client computer, and
then passes itself on to other computer users.
These virtues would be less of a problem if it were not for
the fact that everyone is practically using the exact same email
clients with exactly the same or very similar security
For example, you can get a spreadsheet that will load a
piece of ODBC code that will do some viral damage. The viruses
wouldn't transfer if it were not for the fact that everyone is
using the same piece of proprietary, bug-ridden, crash-prone
software. Not that I'm opinionated or anything... So, anyways, you
don't want to see Linux end up at 99% like this and then it being a
similar thing right, only I would have to add that Linux would be
far more secure than that, but it would still be vulnerable for
similar reasons -- am I right?
Eric: It would be more secure, because it
wouldn't depend upon "Security Through Obscurity" which any
cryptographer [People that ensure that programming code used
for encrypted transmissions and stored documents is un-crackable LT
ed.] will tell you is a very bad idea. Still I would feel
easier in my mind if there was some diversity.
Linux Today: I think that there is going to
be some of that anyway, even if something like Linux took over in
terms of like 99%, not everybody would be running the same mail
client -- That's something, um, a horrible thing called "choice".
God Forbid we have choice! You know? It's been a great thing for
America, right, Eric, that everyone has no choice? I mean, that is
Eric: Hey, ask a Soviet central planner...
Linux Today: Would you agree that BSD users
are necessary and a key to Linux's success?
Eric: I don't think they're a key to Linux's
success, but I think that something like the BSDs may eventually be
important for making the whole Open Source ecology succeed.
Interestingly, I just learned from some BSD people I was having
dinner with the other night that there is now code in the BSD
kernel that was written by Linus [Linus Torvalds -- Creator and
maintainer of the Linux kernel LT ed.].
I'm not sure that he sent it himself, but the patch came in with
a notation that said "Linus did this..."
Linux Today: Wow, that's pretty wild. Would
you agree that there's "a bit" of competition among the Linux and
Eric: Well, of course there is! But I think
that it's less virulent than it used to be because the Linux crew
has proved its point now so they don't have to be really
territorial about the whole business.
Linux Today: Yes, but would you ever like
imagine, you know like, somebody working for Microsoft, sending someone from
Apple, they would first not have
the opportunity, but even if they could... You know what I'm
Eric: Light snow in hell the day that ever
Linux Today: So we're looking at impossible
phenomena, it's in a lot of ways Darwinism... survival of the
fittest and cross-pollination in that particular respect ... I
guess not survival of the fittest... as you get more
diverse pieces of stuff, things grow and they get better, that's
what I'm saying...
Linux Today: As opposed to, for example, if
you keep breeding human beings with each other, they will end up
with third arms growing out of their back kind of situation like --
Or, for example the way that Microsoft products seem so inbred to
me in terms of quality. They don't ever have like a check where
somebody says like "Hey, this is a better product, so everyone's
gonna buy that until you fix yours."
Eric: They sure seem to have the software
equivalent of genetic anomalies...
Linux Today: The effects are the same. You
know it's like some of the mail systems you can use instead. Samba
doesn't really have a heavy competitor but it's still an awesome
product because at least a lot of people can work on it at the same
time. There's a lot of Fear, Uncertainty and Doubt being yammered
by people about what's going to happen when people want paid. Could
you answer some of that?
Eric: We're solving that problem in a variety
of ways. One of the ways is with task markets like CoSource and
Linux Today: Explain what these
Eric: These set up clearing houses where if you
need a particular piece of code made, you can go there and post a
job request and then open source programmers will come along and if
they are interested in that particular job request they will pick
up on it and people can cooperate to put up money in order to have
it written in Open Source.
Another way that we're attacking the problem is as Linux
companies get more successful you see them actually hiring major
developers and paying them to work on useful things that they're
The third way we're solving the problem is that as these
companies IPO there's a tradition developing of putting
contributers to distributions on their friends and family lists so
they make, you know, 10,000, 20,000, in some cases more from the
Linux Today: So, it's a myth thinking that
there aren't people actually making a living at this at the
Eric: Lots of people are and more people all
the time. It used to be that I was one of maybe a dozen people in
the world who could do Open Source development full-time. Those
days are long gone, there are hundreds, maybe thousands of us
An argument that I've made in my papers is that over time, the
kind of gift culture that hackers have is more efficient for
creative work than the scarcity based exchange economy. So, what's
happening is that the market is seeking that efficiency.
Linux Today: You have a paper in the works?
Or is that still classified Eric?
Eric: I've got another book I'm working on.
It's called "The Art of Unix programming". It's a book about the
philosophy of Unix considered as design and vise versa, how to
think like a Unix guru.
Linux Today: What about the paper on the
Eric: That's not a paper exactly, that'll just
be one of my press broadsides. I've added a section to my talk
recently on "The Seven Bullets Microsoft Has To Dodge To Survive
The Next Eighteen Months". They're in much more trouble than
anybody knows, and the DOJ lawsuit and Linux are actually the
least of their problems.
The biggest problem is that Microsoft is going the way that all
monopolies go, they're pricing themselves out of their own market.
Something that we can see very clearly happening with Windows CE at
the low end.
They won't price themselves out of the high end for a while yet.
There they are getting beat on quality by Linux.
You have to look at the market from the point of view of a PC
integrator. You're paying a fixed Microsoft tax, it varies,
depending upon how much Microsoft likes you and how much you
co-operate with their party line. Typically it's maybe 80 to 100
dollars per machine. That's something that looks reasonable when
the price point of your total systems is around $2500, but when
your price point is down around $300, -- it's too much. And this
means that as the price of hardware drops, the PC integrators are
coming under increasing pressure not to pay the Microsoft tax.
The real issue is that this argument about margin means that
there is a price below which you cannot make any money cooperating
with Microsoft. Right now that price point is down in appliance
land. Ok? Which is why Windows CE is doing really badly.
Microsoft technology partners pulled out of their Windows CE
alliance about 10 days ago. I think it was Nokia, GTE and Sylvania
-- I'm not sure I've got the names right.
Over time as the price of hardware drops the functional point at
which PC integrators can't make money dealing with Microsoft is
going to rise. When that price point rises past the price point of
the average consumer desktop PC it's game over for Microsoft.
And the problem is that they can't lower their prices because
their revenues have to go up every quarter. If their revenues don't
go up, their stock price doesn't go up every quarter. If the stock
price doesn't go up every quarter, very bad things happen
to them. Their employees start cashing out, their talent leaves, it
turns out that they make more of their money playing option games
with their own stock than they do selling software, 36%, all of
that income would go away.
So, if their stock price ever tips over, it's going to
nose-dive. They can't let that happen, therefore, the knock-on
effect of that is that they can't lower their prices, and that
means that they are gonna price themselves out of a market, which
is what always happens to monopolies.
Linux Today: In regards to the stock: if
you envision a pyramid, with the base of the pyramid being your
current sales, and starting at zero back when Microsoft sold their
first product, that's zero, the top. That pyramid has to grow 30%
at the base, every year, in order for them to justify their
inflated stock price that they have right now. It doesn't take a
genius to take a look at that pyramid and say that at some point
that base is going to include the GNP of the world or it's going to
Eric: Or it's going to stop and it's not gonna
take very long to stop either because they have 91% of the total
market now. They just can't get much more revenue expansion out of
new customers, they're aren't that many new customers out
Linux Today: And certainly not with a
quality product like Linux, for example, involved.
Eric: Well, Linux is really an opportunist,
it's exploiting a situation where the high prices for software are
not sustainable for PC integrators anymore.
Linux Today: So it would be a really good
thing for them to choose it.
Eric: Right, and a harbinger of this trend is
that Dell is now shipping Linux on all of it's power-edge servers.
It's just not cost-effective for them to do the NT Server thing
anymore. This is the handwriting on the wall.
Linux Today: Eric, it's been really good
talking to you! Look forward to seeing you again. Thanks for your
Eric: See you on the net!