Advogato's Number: Is Linux business viable?
Posted 14 Jul 2000 at 07:14 UTC by advogato
Advogato has been off his antidepressants for a few days now, and
feels like ranting viciously about the so-called Linux business. What
do we, as the "free software community", expect from businesses
operating in Linux space? If part of the answer includes being a
funding source for development of critically important technology,
they may simply not be up to the job. Don't worry, I'll be on my new
meds in a day or two.
While this cat was doing his daily scrounge of the garbage bins near a
well known Linux distribution company, he came across a crumpled up
piece of paper bearing some curious inscriptions. We'll present these
and also offer some analysis and interpretation to see if we can make
a bit of sense from them.
The first rule of the Linux Business Club is: you do not talk about
Linux Business Club.
This rule is an interesting contrast to the usual prattle about the
superiority of the free software process. Even though free software is
presumably based on academic principles of open inquiry and full
information, once you get into the business, you'll find yourself in a
tangle of NDA's and other confidential information.
The second rule of the Linux Business Club is: you do not talk
about the Linux Business Club.
In fact, there's a bunch of stuff this cat would like to use as
examples for this essay, but is bound not to disclose. So I'll try to
limit myself to Open
Source examples. The rest will all come out in the wash
The third rule of the Linux Business Club is: it's damned hard to
make money from pure free software, so instead try to create some kind
At fashionable parties these days, if somebody says they're doing an
open source company, the natural question is, "what's your business
model?" Of course, now that RedHat has a multi-billion dollar market
cap, one of the most popular is to be a Linux distribution. Obviously,
the market can only bear so many of those.
So, the next layer down, there is great creativity going into
exploring business models. Despite this diversity, most of them have
one thing in common: the business is not based on pure free
software. Many development shops pursue a dual-licensing
strategy. There's a free software release, and also a proprietary
release. In some cases, the free software release is disadvantaged in
some way (in the case of GhostScript, the GNU version lags about a
year behind the non-DFSG main release). In other cases, the free
version is available on Unix, but for other platforms you have to buy
a license for the non-free version (as in Qt). Finally, in some cases, the
version is every bit as good as the proprietary-license version, but
licensees pay for the privilege of being able to integrate the code
without having to license the whole thing under a free license (the
model used by Sleepycat for
the Berkeley DB, and artofcode
for Libart). An additional model is to release a bunch of software for
free, but hope to sell services of some kind (consulting, support,
Internet services, etc.).
I don't consider any of these models to be evil, but I would like to
point out that business interests may not always be aligned with the
interests of users (and developers) of the free software. A personal
pet peeve of mine is the fact that every major rev of Berkeley DB uses
a file format incompatible with the previous rev. This causes quite a
lot of hassle for people trying to integrate the code. It seems likely
to me that if it was developed as "pure" free software, there would be
compatibility code so that it would Just Work with existing
databases. It may be only coincidence, but incompatible file formats
are a compelling reason to upgrade. If you don't believe me,
Similarly, if you're deriving your revenue from services, you've got a
business incentive to make sure the software needs those
services. Sendmail, for example, has a well-deserved reputation for
being a pain in the ass to configure. Is it possible, then, that Sendmail, Inc. has a better market
for their proprietary products and services than if sendmail was
simplified and cleaned up?
The fourth rule of the Linux Business Club is: nobody has any money,
although everybody likes to pretend they do.
And now for an issue close to my heart. Even though free software has
made some amazing accomplishments, there are still large areas where
it falls flat compared to proprietary offerings. One of these areas is
tools for graphic arts, one of the serious loves of this cat (my
father did book publishing, among other things, and I've always loved
printing). In one of my previous lives, I spent four years developing
a complete color separation system for newspapers, including a GUI,
drivers for scanners, transmission and reception of photos over analog
phone lines, the actual image processing for color separation and so
on, screening, and drivers for color printers (back then, the color
printer of choice cost over $5000, weighed 80 lbs, and had 300 dpi
resolution) and imagesetters for color separation film output. Since
I've also been doing free software six years, I'm more or less
uniquely qualified to work on graphic arts applications for
Linux. So I've been on a bit of a personal crusade to see if there's
any business out there who might be interested in funding this
work. So far, the answer seems to be "no." In short, there doesn't
seem to be a "business model" for any company to benifit substantially
from the existence of free graphic arts tools.
In many ways, this is a classic tragedy of
the commons situation. If good, free tools were available, users
would benefit directly by not having to pay for proprietary tools -
somewhere in the range of $1
billion / year for Adobe products alone.
Yes, I'm whining about not getting enough money for the work I do in
free software. As I say, I'll be on new meds soon.
The fifth rule of the Linux Business Club is: nobody wants to give
money, but they do want to hire you.
Now why would this be the case? Cue the X-files soundtrack
and we'll speculate.
Presumably, paying for work as an employee and pyaing based on the
project would be about the same to a company. After all, the work gets
done either way, right? However, paying for consulting is essentially
a liability - an expense that has to be carefully justified within the
business goals of the company. Having a talented employee, though, is
an asset. One of the main differences is that employees have loyalty
to the business goals of the company, even if the company undergoes a
strategic shift or an acquisition.
In fact, for a person who feels a primary loyalty towards free
software, one of the real hazards of working for a free software
company is that it may morph into a classical proprietary software
company. Anyone remember Scriptics, the company founded to support the
free scripting language Tcl/Tk? Well, they're Ajuba now, provider
of XML-based Business to Business integration Solutions. Same
CEO. Different business. Oh yes, there's still a Tcl page.
Much of the actual value of a Linux company is the fact that, if the
whole Linux thing doesn't work out, the company could use its
resources (ie employees) to profitably develop nonfree software. Not
so with a consultant, especially an uppity one who believes
passionately in free software.
Ok, you can stop playing that X-Files mp3 now. In fairness, this point
is not by any means unique to free software. While I'm at it, I'd also
like to give a shout out to Eazel
for being a dream consulting client.
The sixth rule of the Linux Business Club is: IPO early, IPO often.
For the most part Linux companies aren't making a profit. There are
people getting rich from free software, but so far the real money has
come from investors, not from actually making money as a business.
Thus, I find myself left with some serious questions. In spite of all
the highfalutin' glowing economic analyses of free software, at its
core there's still a great big tragedy of the commons. The classical
capitalist model doesn't deal well with these kind of tragedies. I'm
not yet convinced that the hype-frenzied Linux IPO market
fundamentally changes the base facts.
It may be, then, that there is no "Linux Business" that is capable of
funding much-needed development. In the end, this might not be a bad
thing. The true heart of free software may be beating among the people
who do it for the love of it, and for whom the ability to make a
decent living at it is a nice perk, rather than money-dazzled VC's who
hope to strike it rich betting on the possibility that Linux is going to
the Next Big Thing.
Thanks to Jim Meier and Manish Singh for interesting discussions about
the rules of the LBC. Also thanks to Adrian Likins for inventing the
whole market niche of being a jaded bastard, and to Larry Ewing for
futile efforts to talk sense into me.
silver platter, posted 14 Jul 2000 at 08:01 UTC by graydon »
why would an industry offer random cash rewards to employees for doing
what they want to do, rather than what their employers want?
forget it. insofar as you may need money (which is a whole other issue)
you must do what people are willing to pay you to do. free or nonfree,
that is how one gets by in our economy. charge by the hour, and decide
what jobs to take by trading off your ethics against your desire to be
filthy rich, just like everyone else.
free software is not about business. it's stupid to pretend it's
about business. it's about emancipation from having to let someone else
make all the decisions about your software. same way literacy and
emancipation from dependence on others for reading and reasoning. once
emancipated, you can write your own software, and draw on the volumes of
historic software publicly available.
hurrah. this doesn't solve any economic issues whatsoever for
you, aside from giving you a new skill set.
emancipation is valuable to companies and humans alike. if you are
working for someone who doesn't yet see the intrinsic worth of it, well,
you can always shop around for another employer. there's no guarantee
the emancipated will be any wealthier than an obedient slave though.
Business & Linux, posted 14 Jul 2000 at 15:10 UTC by lucas »
I don't think ppl should be necessarily hostile toward business. Many
times, business is good for Linux, as it pushes money into the
community of like-minded people. I am against the idea that throwing
loads of money on a Linux business will make it successful, however.
This was the VC philosophy that I think you're referring to.
People getting rich from Linux happened a year ago, even six months
ago... but that era is over. IPOs? Again, symbol of a bygone era.
Tech IPOs are flat, as is their hype. As you well remarked, no one
believes them anymore without a business model that actually works and
has been proven to work.
It's just now that the business-unsavvy free software/open source
community has finally figured out the logistics of what went on, and
feels cheated with good reason. It was the VCs that made the serious
money, along with a few hackers who gained notariety and/or good
positions. Respected Linux figures like ESR and Red Hat are equated to
profiteers, milking their positions for all it is worth.
Some really business-unsavvy people on Slashdot are addicted the word
IPO... They really want to get bad business entangled into Linux and
free software... "Linus should make Linux a company and IPO it! That
would make LOTS of money for Linux." etc. These people are really
clueless, because you become subject to the whims of tech-unsavvy
speculators and are grouped with other tech stocks, such as Apple and
Microsoft. It would also make Linux Inc. into a Red Hat, where they
would have to satisfy shareholder expectations by generating
The pendulum will swing to the other side for Red Hat, because there
is a very slim chance they will survive the money they received and
keep themselves above water in their repayments of cash and loss of
equity to the VCs. When Red Hat begins to show they have problems, I
suspect that most people in the community will secretly fret.
In my very first (and rather long-winded) diary entry on
advogato, I addressed something approaching this issue (scroll down to
near the bottom).
I propose that there's only room for a limited number of big,
publically held companies in the free software market. There's no way
that there will ever be enough Red Hats and VAs to employ us all - and
even if they do, advogato rightly points out that that might not be so
great for linux anyway.
So I make an analogy to the automobile industry and think - okay,
there's only a few big auto manufacturers but there are a lot of
small dealerships and auto-repair places that employ a dozen or so
people. With free software, you can actually imagine something similar
happening for software - if you mangle your computer's configuration,
instead of reinstalling and losing everything you've done, you take it
to your local software-repair shop. Or if you're a small business who is
finding that Linux would be a great solution for you if only it
supported Xxx, you pay your local software dealership to get one of
their hackers to add the feature. Or - imagine a dealership selling
software with a warranty!
Obviously this completely falls down for proprietary software where
the dealership or repair shop simply can't fix anything. But with free
software they certainly can. Selling a year's warranty also negates the
argument that selling services creates an incentive to produce bad
software - because you make the most profit if the user never actually
needs the warranty.
I can imagine that each of these dealerships might employ a dozen or
so hackers, and would never have any intention of going public - so
there would never be the issue of having to please shareholders rather
than sticking to their principles. Since they would be local, there
would be competition in any given area but the incentive would be to
cooperate with the vast majority of other dealerships, because they'd be
too far away geographically to compete with you. Hopefully, you would
employ enough hackers that for any given feature your customer wanted,
you would have someone whose "pet area" would cover it - if none of your
staff could do it, you could actually "subcontract" the task to another
dealership who did employ someone.
Of course the hard part of this model is "how do we get there from
here?". Currently, the demand would probably not be high enough to make
any such business profitable - most consumers wouldn't even grasp why
they would want to use such a company. And you wouldn't get any VC
funding because there's no prospect of a million-dollar IPO.
Additionally, the model has a lot of network effects - especially the
What do other people think? Could this model work? And if so, how
would it get started?
oops, posted 14 Jul 2000 at 17:43 UTC by sab39 »
I meant this
diary entry. I really should use preview...
Most of the time, companies prefer employees to contractors. However, I
think in the case of free software companies, this is not so they have
the option to make all those employees work on non-free software. For
instance, if my current employer moved off of their current free
software strategy, I would quit and so would many other key engineers.
I think the reason is that employees have a much bigger vested interest
in the success of the company as a whole, whereas contractors only have
a vested interest in fulfilling the terms of their contract. This really
is a big difference. At Eazel I don't just do my assigned work - I find
work, I help other employees, I butt my nose into completely different
aspects of the company. In addition, the presumption of greater
long-term loyalty means the company can expect all the domain-specific
knowledge I'm acquiring to make me an even greater asset to them in the
future. And finally, I have a large interest in making sure the company
is successful even after I am no longer an employee. I write my code not
jsut to get the job done, but to make sure it can live on and be
understood and modified by others later. I'm not saying that contractors
don't do these things - in particular, I think Raph does good,
maintainable work, from what I've seen of his work for Eazel. But the
incentive structure does not encourage it nearly as much.
Personally, I like the fact that companies (especially startups) have a
heavy preference for hiring employees in many cases. I like knowing that
the people I'm working with plan to stick around, are focusing their
full energies on the company, and are looking longer-term than just
their current project.
First off, there is no reason to think that this is the fault of
medication, or lack thereof. There is a ring of truth to it. Advogato,
you are right, not unmedicated.
There are two general points to this, and I think you mixed them up.
The first is whether the "Linux Business" model works, or if it will
collapse into the original models. The second is a discussion of much
needed work and how it relates to those Linux Businesses.
I was chatting with Advogato last night. I mentioned (again,
I'm sometimes too pushy :-) that the big thing GIMP lacks is the
professional colormatched printing capabilities that Adobe has a
stranglehold on. Of course, that would be a large, involved project,
and to much of an undertaking (maybe) for someone to do for free. As
Advogato certainly has some of the qualifications for it, he would make
a good choice, but someone, somewhere has to foot the bill for his time.
This is an example of needed work. Another good one is a quality word
processor. Certainly some things have been tried on each, but with out
dedicated funding and manpower, neither has been done. And guess what?
When people speak of Linux's limitations, these two are biggies.
It's not about a random cash reward to the developer for doing what they
want. It is rather about finding the needed projects, the developers
who can do them, and the businesses to pay them.
I think the statement:
It may be, then, that there is no "Linux Business" that is
funding much-needed development.
could be very accurate. Things still need time to play out, of course,
but most of the businesses, like all businesses, have to chose their
target, and aim for it exclusively. They cannot be blamed for that.
But it also means that important tasks that aren't central to their
target must vanish from their radar.
I'd like to see Linux as the OS of choice for graphic designers, but it
won't be until they can draw ala Illustrator, and print, ala anything
Adobe. I'd like to see Linux as the OS of choice for regular users, but
it won't be until they can quickly create a WYSIWYG document just as
easily as they can with Word or AmiPro. Heck, even with the old MS
Obviously, many folks (myself included) don't find mass
acceptance of Linux to be central to their enjoyment of it. But many of
these needed tasks are helpful to the power user and the novice alike.
And why not get the mass acceptance while we're at it?
As long as the "Linux Business" is couched in terms of "Pick something,
IPO, get rich, who knows what happens next?", it won't last. I know
that this doesn't cover every Linux business, but it does cover
a lot of them (and every other .com out there). At some point, real
value has to be returned. It has to be done in such a way that the
particular business is obviously the one to benefit.
My thanks go to Advogato, who has put so much effort into the free
software community and free software itself.
I'm sort of biased about this, as one of the main developers, but the
claim there is nothing approaching the quality of MS Works (a truly
awful product) is simply untrue.
- AbiWord is the only truly cross-platform word processor.
AbiWord runs on AIX, BeOS and QNX, not to mention Windows 95 and 2000,
Linux and NetBSD. I challenge any modern word processor to say the
- AbiWord is the lightest word processor around. The current binary
is about 4 MB, about 1.5 MB of which is help files. Comparison to
virtually any other WP in this area is just a joke.
- AbiWord is usable right now. As a college student, I write a
lot of papers. Last year, every one was written in AbiWord.
- AbiWord has the best MS Word importer around. It's so good, the author was recently hired by StarDivision.
- AbiWord has an active development community. Come join in the fun.
- Just like people here have claimed is impossible, AbiWord is
sponsored by a corporation, called
SourceGear. They wrote the original code, and released it for
everyone, and still maintain an active presence.
AbiWord doesn't have everything yet, but we are making rapid progress.
Don't dismiss us yet.
Free software has proven that it works not by financial gains. Just
like science. And the solution of money issues should be like science
too: government should provide support and funding. The Free software
community has become a critical institution of the Internet era;
important information infrastructure (like gcc and Apache) are
developed by hackers (and originally for hackers but benefitting
everyone else). National Science Foundation supports scientists; the
government should legally recognize the special status of free software
and set up an agency to support (not manage) it. Companies support
science too; thus this does not mean commercial entities should be
excluded from free software development. But government support with
ensures that free software has the resource to advance regardless of
the happenings in the marketplace and share holders' meetings.
(Sorry about the seventeenth-century chapbook style title, but I
yelling is in order. And this isn't a simply reply.)
The commercial software industry exists today because existing copyright
product of the 19th century, era of the printing press) covers software
provides intellectual property rights that can be marketed (regardless
sanity of asserting an absolute right over copying when the marginal
copying software tends towards zero).
Some far-sighted software businesses prefer not to rely on a wholly
and ultimately unenforceable "right" -- they make their profit centre
or services. Such companies include most open source vendors -- Red Hat,
for example. Other businesses admit the impossibility of enforcing
but make money through passing the contribution plate -- shareware.
Some businesses claim to believe copyright is enforceable ... but
this area, you'll note Microsoft making support services a profit
Ultimately I'm not sure the software industry can survive at all,
is not an industrial product like automobiles or concrete bricks. (I
wrote a column
exploring the paradox of intellectual property law; you can find it
and it basically expands on the theme of this reply at length.) Software
is a service, which once created can be replicated indefinitely. This
has no precedent in human history -- it's a fundamental change in
the rules that all our existing economic theories are predicated on.
The question isn't "is there a business model for free software" -- it's
"is business possible in a post-scarcity economy". And this has a huge
long-term implication attached to it ...
nanotechnology (pace Eric Drexler) is possible, and if a
replicator (as specified by Drexler) is a practical technology, then all
assemblies of bulk atoms will tend towards the state of software today
-- that is, as long as you've got the raw feedstock and a minimum of
energy, the cost of replicating any manufactured item will tend towards
zero (or, according to some estimates, around US $100 per ton for the
raw matter and energy).
This isn't an entirely arbitrary assertion. Nanotech designs for
assemblers basically recapitulate the intracellular machinery of DNA
transcription, mRNA, and ribosomes (albeit with some very exotic
chemical engineering thrown in). This is essentially a
software-controlled process. Designs for proteins are transcribed into
mRNA and then control how proteins are assembled by ribosomes. We can
expect the same sort of software-mediated process to impact on physical
reality if a genuine nanotechnological replicator is possible. And
according to Drexler's median term projections we're less than twenty
years away from this. With such a tool, "if you can design it you can
build it" applies to any stable configuration of atoms. A bit like the
way "if you can write the source code, you can compile it" applies to
sequences of machine instructions. A ribosome is a compiler,
after all -- it translates DNA into polypeptides -- and a molecular
assembler will be a compiler, too.
If we can't figure out how to make money off GNU EMACS without annoying
Richard Stallman, how in hell are we going to handle the entire
manufacturing base of our civilization vanishing down the same
Some time ago I made an attempt to tackle the problems described here.
See here for details. I'd be interested in any comments.
Of course, the term "free software" has been the souce of so much confusion. I, like many others, first thought the Free Software
Foundation was all about making software free of price. The term open source fits open source software well, but not free software. Liberal
Source does not sound right to me, sounds like you will get a "liberal helping of source code with that software". If we're going to utilize
unambiguous term, I think it should convey the qualities of free software that make it, well, survive. Free software can't ever really die, it is
not locked up by IP laws and licensces, it is freely available to all and always will be. It is analogous to the books in your local library.
Everyone has access to them(the programs), everyone can use them in their research. Free software belongs to the community, to
humanity really. It is software for anyone who can run it, for anyone who can change it. Maybe we should start calling it "community
software", and we can have the "community software license". We have Public access T.V. maybe we could call it Public Access
Software? Most of the Public TV I've seen stinks so maybe that wouldn't be such a great idea, unless of course most free software stinks!