Eric Armstrong


Date: Sun, 23 Apr 2000 20:05:56 -0700

From:   Eric Armstrong


Subject:   License Model: Preliminary Suggestion

Paul Fernhout wrote in a letter on April 22, 2000

"Open Source" (TM) Licenses cannot discriminate among classes of users. The system ... would not be considered "open source" or "Open Source"(TM).

I don't understand that. Why not? The reality is that unless a revenue stream comes from somewhere, there is no way to "fill in the picture" with little things like user documentation, customer support, do interface usability studies, and do a dozen other little things that produce a truly usable system.

Paul Fernhout wrote in a letter on April 22, 2000

Violating these guidelines would likely lose the participation of open source developers. I have been participating in this colloquium on the basis of the promised "open source" nature of the eventual OHS/DKR. I such a license was chosen as is described below, I would not be too happy about that.

That's important feedback to keep in mind. Its something we have to make a conscious decision on, if it turns out we go that way. I'd love to open source it to everyone, but I have yet to see a business model that makes sense, other than one where your *real* business is built on top of the software.

Paul Fernhout wrote in a letter on April 22, 2000

One can make money from open source software if that is one's goal. You just can't do it by selling the right to use the source or resultant binaries of the core distribution.

The goal is to provide the wherewithal to sustain future development, and the freedom to pursue important projects when they become apparent.

There's only so much you can make from support contracts.

If we're going to enable GM to make or save an extra 10 billion one year, as a side effect of getting the software we need to save our collective skins, it's hard to see how it would be wrong to get a slice of that, so we can do a better and faster job of (hopefully) saving our skins. On the other hand, for some reason the open source standard seems to have completely ruled out that option, even though it appears to me, on the surface at least, to be completely reasonable. Do you have any insight into how that decision was reached? Do you know of any good sources that give a rationale for it?

Paul Fernhout wrote in a letter on April 22, 2000

If Sun had delivered Java with an open source VM code base on day one, there would never have been this hord of over 100 slightly incompatible reimplimented JVMs all over the place -- making reliable Java code delivery to an arbitrary end user the nightmare it is today. That is why Java is considered by many to be dead on the browser for end users, and is now being used mainly in servlets. I use this as a cautionary tale -- pick the wrong license and much effort and good intentions may go for nothing and the wheel gets reinvented (badly) anyway.

One wonders what the result would have been had the results been freely available to the "evil empire", without the financial resources to carry on a legal battle -- or to fund the small army of developers who have developed the GUI libraries, multimedia libraries, and other stuff for the platform. It would have been nice if it had been truly open source -- or would it. Would the result have been as useful for servlets, or any more standard for browsers?

Paul Fernhout wrote in a letter on April 22, 2000

Frankly, I don't think making money from selling stuff to support this effort should be a *primary* goal. If money is an issue, there are foundations and governments with billions of dollars spent annually on efforts less worthwhile then what is proposed here.

The primary goal is building it, supporting it, and being able to improve it and/or take on other projects, as needs dictate. The only question is... "What is the best way to do that?". Open source has advantages. Making money has advantages. What works best?

When it comes to foundations and governments, I would point out that Doug has been pursuing this course for 30 some years. Where is the army of developers he should have at his fingertips to take on new tasks? Where is the collection of funding agencies lining up to back his vision? The point of a "vision" is that is not a universally accepted view. If you're going to make your vision a reality, it helps to have the financial resources to pursue it -- both individually and as an organization. Will open source get us there? Possibly. What about 6 years from now, when an equally important but slightly different opportunity comes along? Will we still be doing it in our spare time? Or will we have the freedom to devote all of our energies to it?

Right now, frankly, I'm concerned about my job performance. I am investing a huge amount of time in this proposition, because I find it compellingly necessary. But the intellectual energy that goes into this project is *not* going into my "bread and butter" job. Typically, I run at 200% or 300% (I write programs to augment my productivity, so I can do a week's worth of work by setting up a task and pressing a button). But recently, my "on the job" productivity has been a fraction rather than a multiple. And its a hell of a strain -- I come in evenings and weekends hoping to make up for lost time, and then find myself with another irresistible intellectual challenge. It's too important not to do, and I am both too motivated and too undisciplined to walk away from the challenge.

But how exactly do I keep working at this and keep a full time job? I've figured out that I ought to just stop *wanting* a life. But that doesn't really work, does it?

Paul Fernhout wrote in a letter on April 22, 2000

It would be better for individuals in my opinion interested in making money as part of this effort to either: a) be funded by grants individually or through a non-profit like the Bootstrap Alliance

No problem. Where are they?

Paul Fernhout wrote in a letter on April 22, 2000

be funded by companies as employees (or contractors) and use and improve the DKR as part of their job to increase the companies efficiency (at the home company or on loan a Bootstrap alliance participant)

Great idea. Know any companies willing to sign up?

Paul Fernhout wrote in a letter on April 22, 2000

provide services as a RedHat/DigitalCreations/VALinux style company (installation, training, hosting, security analysis, customization) to companies to increase the companies efficiency using DKR technology, or (less desirably)

Now that the vacuous nature of *that* business plan has been exposed (80% drop in capitalization in the last week), who do we sell this idea to? It's fine for an existing company. You can make enough to keep going. But I have yet to see the business model that makes a compelling enough case to support venture capital risk.

The good news is that there are signs that *some* venture capital thinking might be changing. There is a move towards "venture philanthropy". Maybe something along those lines is possible. But the business needs to create a revenue stream to support growth and development. That is a necessity.

Paul Fernhout wrote in a letter on April 22, 2000

provide proprietary add ons to the core distribution.

Look at the Zope business model for a good example of the possibilities.

That is a really solid possibility. I keep asking myself, what kinds of add-ons make sense in a DKR? What would be a small enough core to be useful, and yet allow for Adonis and upgrades?

I think its a given that you have to give away the browser, or use existing browsers, if possible. So the add-on has to be on the server. The avenue I saw is that corporations may want a server of their own, so that they can keep their knowledge private. Educational institutions and government, on the other hand, may well provide publically-available servers. At the "informal meeting", it suddenly occurred to me that, if successful, the DKR would be the best library you ever saw. So we may see some future Carnage setting up Darkness points in cities around the country.

At bottom, then, what you are selling the corporation is *privacy*. But that is not a major add-on. You could keep the code proprietary, of course, but it's hard to see what would prevent it from being added. Maybe a DKR should by its nature be publicly available, and you pay if you want to make it private... But I still want to restrict the license terms to companies that are making a profit -- so startups can use it, among others. Or they can use any of the publicly available servers.

Paul Fernhout wrote in a letter on April 22, 2000

I think broadly put, the best license choices are...

The Open Source proposition is still very much on the table. All it takes is a business model that can work...


Eric Armstrong