A Special report by Talin

Open Source Development

August 1998

read book now




BOOK Reviews

 Talin, onetime dream master of the Dreamer's Guild, is an old friend and has long been a source of ideas and sanity checker for both fiction and non-fiction. This is his report on a recent meeting I didn't get to. As part of his report he mentioned the Yorktown Affair (described below). That has resulted in a great deal of discussion which I have attached following this report. First Talin's report:
read book now





I had a lot of fun hanging out at the Open Source Developer Day, which was put on O’Reilly and Associates.

The first session was a talk by open-source advocate Eric Raymond, author of "The Cathedral and the Bazaar" and "Homesteading the Noosphere". Most of the talk was a recap of news recent news that most of you have already heard (those of you who regularly read will have heard most of it anyway.)

For example, I’m sure that many of you have heard of the sad story of the USS Yorktown. For those that haven’t: The Yorktown has been completely networked using Windows NT as part of the Navy’s "SmartShip" program, to try and reduce the manpower requirements for large warships. However, because of an operator incorrectly entering a zero in an entry field, the system crashed and somehow corrupted the central database—as a result, the ship was dead in the water for two hours, and had to be towed back to port.

However, there’s a follow-up story which I hadn’t heard: Apparently, the Navy also holds an annual contest for autonomous submersible vehicles. The vehicle has to navigate a maze, and then drop a marker at the deepest point in the maze, all without operator intervention. The MIT entry was a six foot long torpedo-shaped thingy, running Linux of course. However, about halfway through the test they realized that there was a bug in their program and that a crash had occurred. No problem—as the Navy divers were towing the submersible back to the starting gate, the MIT programmer used wireless telnet to open a secure shell to the vehicle, open up an emacs window, fixed the bug, and recompiled the program; So that by the time that they had gotten back to the starting point, it was ready to go again, and this time completed the test successfully. I’d always thought of embedded systems as the ultimate in "closed source"—it never occurred to me that something like a torpedo could have enough extra computing horsepower to carry a text editor and a compiler inside of it.

The next session was about running open source projects, and included Larry Wall (creator of Perl), Brian Behlendorf (one of the group that does Apache), and Jeremy Allison (one of the Samba guys). A number of interesting points were made. For one thing, a lot of people have described open-source development as "anarchistic", "socialistic", "communistic", or even "democratic", when it is in fact none of these things. Successful open-source projects are almost always benevolent dictatorships. In the case of Perl, people take turns being the dictator; In the case of Apache, it’s a dictatorship by a small oligarchy which votes on important issues. However, these are not tyrannies, because people always have the freedom to quit and go elsewhere—there’s no coercion. It’s true that a successful OS project will accept changes and patches from anyone, but only the "dictator" can decide which patches actually go in the "official" release, although it’s best to be pretty loose about things if you want to maintain a high participation level. They also talked about some other issues, such as the mechanics of release vs. development versions, what to do about disruptive individuals, burnout, etc.

The next session was on business models. Robert F. Young, president of Red Hat said an interesting thing: "We realized that the harder we tried to give stuff away free, the better we did financially." (In fact I picked up a free copy of Red Hat Linux 5.1 CD, sans manual) There are a number of overlapping reasons for this, which I don’t claim to understand yet.

John Ousterhout, creator of TCL, talked about some of the weaknesses in the open source model. He’s starting a new business, Scriptics, which is a hybrid of the open source and proprietary approach—the main product is open source and free, but there will be value-added extensions which will be proprietary. Some people were rather uncomfortable with this, including Richard Stallman who made some rather tactless remarks about it, including calling Ousterhout’s new company a "parasite organization" (but that’s what people expect from Stallman—the purist’s purist.)

A representative from IBM talked about their work on Apache, and how they are going to make it the center of their new web strategy. They plan on contributing to the Apache community, they don’t want to co-opt or control the Apache development process, but rather contribute to it like anyone else, except that they’ve got a lot of people working on it. (This makes a lot of sense to me—IBM is one of the great hand-holding organizations of our time, and their job is to solve the customer’s problem; Whether they do this with open-source or closed source is irrelevant to them. The do plan to build some proprietary extensions on top of Apache however, these extensions will most likely be in Java.)

After that was a session on legal issues. Pamela Samuelson, who’s one of the top IP lawyers in the country, said that there’s some questions as to whether open-source licenses are actually enforceable; Certainly the advertising clause in the BSD license isn’t, since advertising is a separate work. She said that the article 2B of the upcoming revision to the Uniform Commercial Code, which makes shrink-wrap licenses enforceable, will also have the effect of making open-source licenses easier to enforce as well;

However, she felt (as did most of the audience) that article 2B was generally a bad thing. (I agree in general; I’m seeing a bunch of new laws coming at me that criminalize a whole bunch of behaviors which many people consider both acceptable and societally beneficial. Can you say "war on drugs"?)

Stallman, who was also on the legal issues panel, replied that the Gnu license _was_ enforceable, since it was not a contract, either explicit or implicit, but merely a set of permissions on what would have been illegal under copyright law anyway—that is copying, modification, and distribution. It does not place any addition stipulations, it only removes ones that were the default in law. (Other licenses, unfortunately, are rather ornate, and these extra provisions make them problematic.)

Stallman was, in general, his usual outrageous self, making some really odd statements ("Well then perhaps you shouldn’t be in the programming field. Maybe you should consider being a waiter...there’s no shame in being a waiter.") But Stallman’s point is consistent, and always has been. A lot of people mistakenly things that because he’s a top-notch technologist he therefore has the kind of technological optimism that many technologists have; But Stallman doesn’t care about advancing technology, or how popular his software is, or how much money he makes; he only cares whether or not he’s advancing the cause of "freedom" as he defines it. A couple of times, the audience couldn’t help but chuckle at some of his more bizarre and naive statements, but still you kind of have to respect someone who appears to live out their principles with every breath.

After the conference, there was an "Open Source Town Meeting", which had a lot of random questions, most of which I don’t remember. Afterwards I had a conversation with Tim O’Reilly, and had I been willing to take on the work I could have gotten a book contract right there, but I decided that it was more work than I really wanted to get into, what with my day job and all :-)

Talin ( -- Systems Engineer, PostLinear Entertainment.


"The only mind-altering substance I use is breakfast."


Notes by Pournelle:


Richard M. Stallman, known as MIT as RMS, was responsible for much of the software used in the ARPANET. I first met him on line in about 1977. He was just finishing emacs, which he wrote in TECO. Emacs was the first full screen editor in use on big systems; I was using Electric Pencil on my CP/M system at the time, and I very nearly converted to emacs for all my work. It was feature heavy, and made use of the special equipment such as "meta" keys (something like the ALT key you find on standard keyboards today) which were on most MIT systems. Alex once quipped that control-meta-cokebottle had special meaning in emacs. We didn't have either alt or meta keys on our systems here, and that's one reason I stayed with Electric Pencil and later Tony Pietsch's WRITE for my books.

Even at MIT Stallman was militant that 'software ought to be free' and was considered a troublesome genius by the masters of a couple of the MIT nets I had accounts with because he wouldn't install certain updates due to conflicts about philosophy. I forget the details. He is of course best known for gnu, the first of the freeware UNIX systems.

The USS Yorktown Incident has generated a lively discussion, which will now have it's own page.


birdline.gif (1428 bytes)