Getting our acts together

Getting our acts together Four months ago at the end of my article Wither BSD? I reported that my daughter's school uses FreeBSD. Some time later, after I had submitted the following article, I heard from Duncan Sayers, who wrote:

You will be pleased to know (if you aren't already aware) that the freebsd network you referred to at the end of your article at Strathalbyn school/community library is part of a larger network that spans the state from Ceduna to Mt. Gambier, and most places in between. I was part of the project that installed freebsd servers and public access clients to around 80 regional libraries throughout the state. The servers have a 20MB flash drive in them, from which the server can rebuilt itself should it detect a problem with the hard drive that may have been caused by a dodgy power outage (flakey rural power supply being one of the biggest problems that had to be overcome). The public access clients are also running a minimal freebsd setup, loading netscape from the server when they reboot. If the setup on the clients get messed up, either accidentally, or deliberately by the public, the librarians only need to reset the client and it will refresh itself.
At ADC (the company I work for), we use freebsd almost exclusively. Time and time again, we are demonstrating how you can make freebsd jump through hoops to acheive solutions that are simply impossible to do with the surprisingly more popular alternatives.
FreeBSD is alive and well in the state of South Australia. Nay-sayers be damned!

It's nice to know that BSD is getting the attention it deserves.

Article length

In the last article I asked for feedback about the length of the articles. Of course people weren't in complete agreement, but the majority of people appeared to think ``write what you have to say, then stop'': in other words, the length wasn't as important as the content. I'll do that: it might make some articles longer than others, but nobody wants obviously padded texts.

I also heard from ERICHUF@aol.com, who writes:

I think money is the least of your problems and the lack of leadership is the most significant. With proper leadership you would attract money and support.

That's an interesting observation. What of our leadership? As I've observed before, the BSDs are anarchies. Later in this article, I'll look at the implications.

Craig Mundie and internal squabbles

In early May, Craig Mundie, a Vice-President of Microsoft, published a ``prepared text'' of some remarks he made at the New York University Stern School of Business. They were controversial, to say the least, and a large number of people got quite upset by them. On the other hand, the text also has a certain amusement value, not only because of the content, but also because of the bugs in the markup. He writes:

Phase 3 is what is being worked on now. It?s all about connecting the currently separate complex systems of information and transactions and bringing that power to the individual in a readily accessible format on a variety of devices.

Note the text it?s. This is the way Netscape renders the invalid characters in the original web page; this one is 0x92, an invalid character in the ISO 8859-1 character set in which the page claims to be written. Microsoft documents are riddled with these characters.

You may also have difficulties accessing Microsoft's main web site; in most cases, I get a blank screen and an error message indicating that a JavaScript error took place, and that I should enter some text into some non-existent box. I'll leave it to you to decide what that kind of breakage says about proprietary software.

The gist of this rather verbose document is, who could have doubted it, ``Proprietary good. Open Source bad.'' It takes the undoubted mistakes made by many startup Open Source companies and claims that they are fundamental flaws in the model. The examples he gives are so tied up in the Microsoft view that I have difficulty understanding them. I certainly don't see anything in there with which I could agree.

Naturally, the Free Software movement replied quickly with, for the most part, well-founded responses to the document. Nevertheless, I think they're missing the point. Look at the target audience: The New York University Stern School of Business. Of course they're interested in Business. Of course they're not interested in Free Software. I suspect most of the attendees were not overly interested in the article, because it represented a confirmation of the values they (are learning to) hold dear. A Free Software advocate could have taken the stage after Mundie and torn the speech into clearly defined, self-evident shreds, and nobody would have paid any attention.

One of these responses, formulated by Bruce Perens, was entitled Free Software Leaders Stand Together. As the name indicates, it's signed by a number of important names in the Free Software industry. If you look more carefully, though, you'll note that not a single BSD leader has signed it. Why? Don't we care?

As I've said, I don't necessarily agree with the necessity to counter Mundie's speech. On the other hand, I think it would have been reasonable for BSD community leaders to sign the statement as well. The real reason that nobody did was that they weren't asked. Lots of accusations went round the FreeBSD-advocacy mailing list, claiming that Perens and others deliberately excluded FreeBSD. Some were paranoid, some were just plain nasty. None addressed the fact that a number of leaders in the Linux space were also missing. Bruce's own comment was:

They weren't approached. Mea culpa. It was a big enough group of headstrong people to manage as it was, though. I wonder what the BSD reaction in general would have been? Although I assume they use gcc, some of them have been rather GPL-hostile. Hopefully they'd be able to stand together with everyone else.

Given the tone of some of the follow-up messages on the FreeBSD-advocacy mailing list, I really don't think I can blame him. In the last few years, we've made a lot of progress in approaching the Linux and GPL communities, but it seems that we haven't quite got there yet. Sure, people on the ``other side'' will do things that we don't agree with; we all make mistakes. There's no reason to accuse Bruce Perens of malice when he has already admitted that he made a mistake.

This is probably one issue that free software needs to deal with: by nature of the projects, too many things fall between the cracks. We need to defend ourselves against the accusation of being disorganized. It's a difficult defence, because much of the time we are disorganized. In the next section, let's look at the problem areas.

Herding cats

As I mentioned above, some people think we need better leadership. I certainly can't deny that; the question is not whether we need it, but whether we can get it. There are a number of issues running a free software project:

There are no bosses. In a commercial environment, your boss can tell you what to do, and you do it. Try to tell that to somebody in a free software project, and you're likely to get an abrupt negative reaction. Quite possibly you'll lose a developer. Now it's true that people in commercial environments also frequently change jobs rather than do what they consider is unreasonable, but the threshold is much higher; in general, you expect to be told what to do. This is one of the problems that the core teams have to contend with: there's no effective way to discipline a project member who misbehaves. You have the choice of accepting the behaviour or running a significant risk of losing the developer to a different project. You run this risk in commercial companies as well, but the threshold is significantly higher.

People come and go. In a well run commercial project, some people will stay for a long time, maybe decades. By contrast, the oldest BSD projects are only just 8 or 9 years old, depending on when you count from. In that time, most of the founders have moved on to do other things.

The core team members have other duties apart from running the project. Indeed, their duties to the project usually rank a long way further down the list than would be the case for professional manager's duty to his company. My own experience in the FreeBSD core team has shown that very clearly, and as I write this, messages pour in complaining about lack of responsiveness from the FreeBSD core team

This lack of direction makes it difficult to coordinate long-term projects. We're seeing problems in this area with the FreeBSD SMPng project.

What do we do about the problem? I wish I knew. Sometimes it seems to me that we have reached the limits of what we can do with the pure open source model.

One thing's clear: the problems we're seeing here are not the result of malice, any more than Bruce Perens' omission of the BSD people was malice.

Solving the problem

So what do we do about it? The answer is probably that we'll have to compromise. In an earlier article I pointed out that having source code availability is nothing new. In years to come we might look on the period from, say 1975 to 1995, as a time where people mistakenly thought that hiding source code was going to be good for the industry. Before that, most source code was available, at least to users of the hardware on which it ran. Now more and more source code is becoming available. That's the ``Open Source'' side of things; it doesn't say much about how the projects are being managed.

Let's look back at those companies who want to make money out of free software. They may be the key to the survival of the idea; pure volunteer projects are probably self-limiting.

Does this sound like heresy? You tell me. I don't think that free software and volunteer projects are going to die out any time soon, but I can't see a solution to the problems they represent.