My name is Lucas, and I'm with Spindletop, an LLC I began in Cambridge, MA. Spindletop (www.spindletp.com) is essentially non-profit. I designed it to run with extremely low overhead and low margins (ix86 hardware has a very small margin to begin with). I do my own artwork, designs, and coding. I also designed the Blackbird. Most (if not all?) of the things I do are available under the GPL.
What has interested me for a long time is creating a decent, reasonably priced, quality-oriented (e.g., non-distributor-built- whitebox) Linux hardware platform for Linux end-users, catering to the people who know free software best and really like it: you and me... I see this as most relevant to improving open source software acceptance, since most people begin hacking around with free software at home.
Such ideas right now are ignored by larger corporations right now because they are unprofitable. You won't find a Linux box in CompUSA anytime soon. Margins on server boxes are 400%. Margins on your average PCs are about 25% - 30% at best, cost at worst. At this time, I use a standard and fair margin of 15%. Meaning, if Spindletop buys something at around $34 each, it sells for $5.10 extra, in turn, going directly into expenses. I spoke with a friend who works at Compaq as to how they could support PCs; she confided that Compaq is always on the brink of dumping its entire PC line, since it only represents less than 5% of the entire revenue Compaq takes in.
Since Linux and *BSD users are ignored, there's not a whole lot of information about the user end of the Linux market. This is one of the reasons I'm posting this to you, my peers - I'm curious. On one hand, being ignored is good; there *are* merits to being left alone by the larger companies. Poorly built hardware is not dumped onto the market as with Windows machines - there still seems a certain element of quality to Linux solutions that exist. And, using old equipment is great -- we don't have to figure out which elements don't work, as most of their idiosyncracies have been documented.
Yet on the other hand, there is a real need to support Linux users who want a full solution that uses *new* technology as opposed to older stuff.
My thoughts are that it is not going to be Red Hat or any other VC- backed company moving in this direction. They can't afford to run lean and they are struggling just to keep it moving on the enterprise end; it has to start from the ground up.
The basis of my idea is to treat hardware design like an open source software project. Use a legal entity (e.g., a "cooperative") with low overhead to "slide" certain, agreed-upon goods to end-users using relatively low margins. The entity would take care of the accounting work and handle negotiating for the cooperative.
Everyone involved must be an end-user in some sense; it's not a cooperative otherwise.
Members would find an open standard (open design) of a few *current* configurations that work well together and have a high degree of price/performance. There will probably be a few factions who want to use particular hardware -- these can all be listed.
IMHO, it is important to use name-brand, well supported parts. The problem with today's current listings of "what appears to be compatible" with free software is that some of it is outdated, it's confusing for people to narrow down choices.
If implemented properly, people can use the list like a shopping list, ordering pieces from vendors.
Why focus on only a few sets of hardware? If the group is able to focus on a particular set of hardware, the "cooperative" will be able to purchase them in small bulk batches and sell them at significantly reduced margins to members. Members can upgrade their boxen as they see fit. If too many sets of hardware are supported, the cooperative cannot function -- it would become a standard retail operation because the cooperative would have to offset losses in products that aren't selling well.
My idea is that hardware can be purchased assembled or raw (OEM, plainboxed), depending on what the user needs. It would not be in retail packaging.
Hardware should be, for example:
a.) Non-proprietary. If it doesn't work 100% with free software, it shouldn't be included.
b.) Non-dirty. As an example, this means using more expensive, better- built, "clean" motherboards without built-in sound and video.
c.) Upgradeable (to a certain extent). The user should be able to update it (i.e., not have to throw the box away in 1 year to buy another one) in a non-proprietary way. I use the YY-0221 cube case to silkscreen my logos because it is flexible enough, easy to work with, and.. well... it looks cool as hell in real life (cool cases are always important ;-).
My idea is that people would not have to "pay" to be a member, rather they would contribute into the product by voting and participating in message groups. They would have a stake in it simply because, if they want to upgrade their boxen, they want to make certain that good hardware is used. I'm thinking along the lines of Debian right now: stability should be a key element.
The person who runs the cooperative is more of a "monitor" who makes sure that operations are functioning well. He also assembles boxen for people who want to pay a certain percentage for assembly. When jobs become available that the "monitor" can't do, he keeps costs down by bringing in users from the community as consultants... or trading services for hardware.
a. Collective bargaining strength. We could even bargain with Linux Service Providers (LSPs) for things like discounted technical support.
b. Support of a platform that is used and agreed upon by many people; support is more widespread and more business will be lost by vendors who do not support drivers for free software.
c. The Community is supported by using workers (e.g., consultants) from within it.
d. Significantly lower cost of hardware purchases, particularly for individuals. Corporations who want to implement end-user-based Linux solutions (e.g., non-server) would also benefit.
e. Ability to help people on an individual basis to cover their hardware needs.
a. Users must know how to install equipment; it is not for everyone.
b. Support would be limited. There would be no "distribution-base" support.
c. User apathy would hurt the system; the users (and especially the "committee" should be interested and familiar with products out on the market.
d. The model doesn't scale well. On the other hand, the amount of Linux end-users in comparison to Windows users is very small. Equity- based venture capital (VC) would be a definite no-no, as it would break the system.
I'd like to get some feedback about this idea -- think about it, hack with it. Is it viable? What are your needs as a user? What services can make it more useful?
I'd also like to hear from people who would be interested in becoming involved.
What a thoughtful and well-organized post. I also read your recent diary entry where you reflected on your reservations about posting this article... Not to worry. It doesn't matter what the 'dot-com'ers and 'high rollers' think. Like you say, happiness and self- respect before 'anything for a buck'. You already understand the foundation. Your challenge is to figure out how to do big things without being big.
IOW, you're looking at 'Small is Good' and its corollary 'Simpler is Better' as organizing principles for your business model. And that means 'elastic networking' among trusted, competent entrepreneurial free agents and 'dejobbed' small businesses. Here are some resources that might help you flesh out your vision/plan:
An on-line article by former MIT wunderkind turned Founding
Leader of the world-class program in eCommerce at NC State, Michael
Rappa -- Business
Models on the Web. This is free on-line and gives a good
overview of the range of business models being explored on the
Internet. Knowing this taxonomy will help you understand and analyse
other businesses and it will help you envision and plan your own.
Think of Rappa's taxonomy as the 'front-end, public interfaces' of what your business model might be. Now think about the 'back-end'. We know that 'Small is Good' business models are going to be collaborative, networking types of organizations. So dig deeper into this by reading Tapscott and his co-authors latest book, Digital Capital: Harnessing the Power of Business Webs.
Of course a lot of this book is about business webs built among
Industrial era corporate organizations struggling to 're-invent'
themselves to compete in the web-based New Economy. Don't let that
distract you. The point is to understand the matrix of types of
Now, bring it down to a narrower class of business webs...
the 'Small is Good' kind of organization. I have been thinking a lot
about this class of business model in the context of rural and urban
economic redevelopment. The fourth installment of The Nanocorp
Primer is called "Shamrocks and Nanocorps:
Bridging the Digital Divide with 'Small is Good' Business
Webs". Lucas, you might read this with an eye for mining
its ideas to apply to your 'community-based hardware house'
As you pursue these ideas, I suspect you will meet and begin collaborating with like-minded members here at the Advogato community. And that is no accident. Trust metrics are an important part of the software infrastructure of any OSS platform that will support these 'Small is Good' business webs.
For just one example of the cool people I've met here this week,
I'll point you to an interesting Londoner, Pelle Braendgaard. Pelle
is bearing it all with his NeuDist.org
project. This project is all about creating a secure infrastructure
for the economic 'transaction space' which is, in effect, the
distributed accounting and payment distribution subsystem in
such 'Small is Good' business webs.
So, Lucas, you see. You have nothing to worry about. This is a place for bouncing around creative ideas to see what happens. Getting 'shot at' occasionally is part of the fun! And you know what they say about what doesn't kill you...
Whatever you do... keep having fun,
I think the idea of a Linux hardware cooperative is a good one, but what would it take to get the ball rolling with something like this? Could it be competitive with mail-order options in pricing, or not until critical mass?
>I think the idea of a Linux hardware cooperative is a good one, but
>what would it take to get the ball rolling with something like this?
>Could it be competitive with mail-order options in pricing, or not
>until critical mass?
It could be competitive from day one; absolutely. I think the key is really to keep small, keep flexible... and keep going. Use the help of others, even if it is a little bit. There are always going to be people screaming exploitation, even if the price is $1 over wholesale (e.g., $1 over what it costs).
My question is: will people volunteer for the common, abstract good as they do with Debian or will they see this as a Red Hat? Will people fight virulently for a Collective or against it?
People don't like to see profiteering, but on the other hand, selling retail goods cannot be done like free software. This is where I think we can do it differently - by being responsible, conscientious, and even creative with our effort, a Collective could place us at the bargaining table. We've shown, with free software, this is possible. Is this viable to accomplish in the other world of hardware/retailing?
My general inclination is to say yes, but there is a lot more work to be done.
Once the courts are finished tromping microsoft, Dell, Gateway, and all the OEMs will start pushing Linux boxes seriously. After that happens, most of the end-user difficulties will start getting better.
Instead of starting with the end users, let's consider first the other small companies much like yours, which integrate components into private-labeled machines. There are hundreds (thousands?) of such companies in the US, and many of them are able to assemble computers that are price/performance bargains compared to nationwide name brands. I've bought a number of these machines over the last 4 years, and one thing that has always annoyed me was how little these companies knew about Linux. (And how much nicer it would be to have handy local companies who do know about Linux.)
Thinking about this, I came up with an idea which is similar to yours. I call this The Linux Systems Guild. The idea is to publish a set of specs that describe how to put Linux systems together. The designs would be open, publicly debated, with feedback tracked. Anyone who wanted to build a system could tap into this knowledgebase. LSG itself could be managed as a brand name and certification mark, which should bring business to LSG Craftsmen.
In addition to common system designs, it would be interesting to try to write up a "Guide to Ethical Pricing" for LSG businesses. Such a document could help both businesses and customers understand and agree on prices. In particular, it could go some ways towards explaining who pays for free, and how, and why. (I'm not sure that the "extremely low overhead and extremely low margins" that you run your business on will be popular with other businesses, but they set a baseline for viable competition and optimal customer value, and that's a line worth exploring.)
My LSG thoughts were the basis for the home networking and automation business model that I've proposed. In this business, I see the umbrella organization as a for-profit company, partly because it's a much bigger proposition, hence it needs more capital. However, it wouldn't need to be structured with any VC if there was a sufficient network of expertise ready to seed it. Certainly in the LSG case the expertise for such a network does already exist.
When I think of end users, I tend to think of people who just use computers, and not necessarily even Linux. Those who would build their own box (the DIY users) are a very small subset of end users, especially if you pass the savings of using standard, open design and research, and joint purchasing on to the customer. How to service a broad spectrum of potential Linux users is an interesting question. I gave a fair amount of thought to this problem whilst at my former employer (thought which I'm quite certain you will not see implemented). To make a long story short, the key is that one be able to spread the costs out over the largest possible user group. (The costs include the ability to digest a lot of widely scattered information, and to build tools which plug into the digested information. This may also be simplified by standardizing procedures, including how projects are developed and how distributions are assembled. In the long run it should be as easy to build a personal distribution as to assemble the hardware for a computer.) This could be born out by a large company, but it could also be done by a broad-based community effort (probably a lot cheaper). In fact, this is the great charm of lowball business models: nobody can compete with them, and if everyone's welcome to join, nobody needs to.
My immediate thought on the original article was 'Nice, pity I won't be able to participate'.
Why not? Because I'm in Australia. My assumption (possibly incorrect) was that the collective was likely to be sited in North America someplace. Works fine for those in North America. Shipping costs mean it won't work outside North America. Power supply variations mean alternative cases would be required where voltages differ. Etc.
On the other hand, an organisation which sold hardware kits AND posted the contents of the kits (and any necessary drivers) would be immensely, incredibly helpful even to those of us who live on remote continents. If we could recommend to friends sets of hardware that we knew would definately work, it would save us a lot of bother. (Not to mention being helpful for our own machines!)
New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.
Keep up with the latest Advogato features by reading the Advogato status blog.
If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!