What is Open Source? No, really.

by Steve Mallett

What is Open Source?

I've spent time this week with every range of attendee at OSCON. From those who've been in the open source
community for years to those who have heard of it and were interested enough that they came to OSCON to learn more.

No one has the same idea of what 'open source' means & talking to everyone has left me wondering if my ideas are too concrete. Too rigid.

To some of the experienced folks here software isn't open source unless it meets the open source definition as written by Bruce Perens. There are nine tests
of a software's license to determine if it is open source, and affording the user the freedoms the definition addresses.

Now, many folks here who've been involved for a short-time have one absolutely critical concern. Access to the source code. That has several implications; the most important one being permitted to change it to their needs. These are the two "Under God" elements of open source.


Beyond that many folk are not that concerned if they are simply the users. I'll
address this now just as an end-user. Doc Searls yesterday at lunch was addressing some of his problems with the politics of what open source is. Linux Journal (LJ) is talking about adopting a blogging element to their website, but it has to be 'open source'. But, as Doc put it
, "What's that? It's political, right?"

For him, personally it's cool if you can get the source code and can mess with it. He was implying that that wasn't going to be good enough for the guys back at LJ. Either they wouldn't technically accept that or the reader wouldn't.

This differing view was also raised during the talk "Introduction to "http://movabletype.org">Movable Type." The software itself is extraordinary. But, alas its license doesn't meet the formal open source definition. This
was raised by some folks who are software developers. The presenter, Mena Trott
, one half of the husband and wife development team, mentioned aloud that they weren't exactly sure what they were doing at an "Open Source Convention". The audience didn't think their presence was odd at all. You can have the code and mess with it. They simply ask you don't redistribute your changes freely because
it makes it difficult for them to provide support to folks who may be using code
not supplied by them. That's not unreasonable to many people. Many folks also didn't have much of a problem with there being different licensing for commercial vs. individual use. This is only slightly different from the license of the
widely beloved and deployed qmail so don't get too excited.

Now this all begs the question. Just because there is a formal definition does it mean anything if people don't recognize its full doctrine in real-world use? I'm left questioning this myself. And as an end-user I wondering what my
definition is. It has changed. Again.

Keep in mind that the current PHP license doesn't meet the formal definition either. Would any of you choose not to use PHP because of that? Linux
Journal didn't let it stop them from making their website with PHP.

And your take on the subject is...?


2 Comments

garym
2002-07-26 23:38:52
Blame it on ESR
I'm surprised a site like ORA could post this and not realize the problems with the premises.


1) Doc has a day job. His job does not include modifying code that other people might use, so yes, it's appropriate that he can use it, except, as a writer, is he really using it for "non commercial purposes"? Can he use it, for example, to write about his magazine or his books? He's making money off it, so isn't that a breech of contract unless he's paid for the commercial license? Just asking.


2) "Open source" is so watered a term that its practically useless. It does not even mean you can mess with the code, it only really means you can look at it. Looking at code can be helpful in debugging, but it's most often, as with MT, wrapped up in an ego thing of "this is mine, not yours" which fails to pay homage to Babbage and Lovelace and all the others, and therein is the point.


3) Ok, I use MT and I really, really like it, but I can't use it for our company because of the above two points. The company I work for is a commercial entity which sells new services based on combinations and modifications of other people's code. They turned to free software because it was the simple solution: If we have to hack something to sell it, there's no lawyers involved, and because there's no license involved, we can also take a chance on something without having to assess a business plan to justify and estimate the license first. We've had to overlook a lot of otherwise promising software because of this policy, but that's ok.


The notion that MT should have to support a forked version is bullshit. Excuse my French, but that's the truth of it. You tell me one documented case where an OS/X user called BSD for support, or even where a WebSphere customer called Apache.


If I take MT, as I'd like to, and replace it's geeky text-edit screens with people-friendly screens that accept WikiText, and if I then sell it as MagicType or whatever, branded under our company logo, is what I have done "intellectual theft" or is it "building on the works of others"?


If you want my personal opinion for which I am only occasionally responsible, I don't think MT is locking the source because they believe only they have the smarts or because they want quality control. I think its because blogs are sexy in 2002 and it's personally valuable to them to be known as the exclusive authors of such a sexy blog kit. It gets them noticed, it gets them press. If there were 40 excellently forked clones of MT, they'd forever have a place in the geek pantheon, but the media wouldn't consider them darlings. Sorry guys, but that's my take on it; I can't think of any logical alternative explanation to lock it down so arbitrarily.


I don't fault anyone for their choice of license, I only want to explain what works for me. Everyone is free to place whatever restrictions or freedoms on their code just as I'm free to use or not use their works as I see fit, and to release my own work however which way suits me.
For my professional purposes, I don't want to become tied to vendor, and I don't council clients to become tied to one either. If our clients wanted a blog, I'd be wary of recommending MT because that crew might disband, or they might change their direction in ways we don't like (which is what MetaDot did to us), or we may simply migrate away from their ideals. Using MT limits their options and reduces their freedom, and binds them to MT. No one should be bound by legal red tape to follow the plan of some other company's agenda because life is just too short. After all, it is only software, and ALL software is built on the shoulders of others.


What MT is doing might work as a business/project plan, but I can't think of any successful cases off hand. Cloistered "vendor controlled" development works for GCC because no one really needs to fork it; the purpose of a compiler is pretty black and white. Linux, on the other hand, is forked all over the place with its PDAs, Mainframes and PowerPCs yet it doesn't hurt anyone, and we all go back to the people who sold us our Linux to ask questions; we learn pretty darn quick that Linus is not going to support our PowerPC questions.


As for whether LJ should use quasi-proprietary software or not, well the EPROM BIOS on their server is probably 100% proprietary,as are their routers, so any argument to the contrary is also bullshit; after all, the inverse argument, that Internet infrastructure already uses free software, is how I got many corporations to accept using GPL software ("if you can't use GPL, you have to log off the internet Right Now" was the line I gave Bell Canada ;) It's OK to voice a preference but yes, being as blindly dogmatic one way is no better than a dogma in the other: If you have need to modify the source AND redistribute, then you need the license that lets you do that within your business model. If you don't redistribute ANYWAY, then you can license a for-profit MT, and if you are not a business AND you won't redistribute your hacks, you can use it as is and for free.


Doc is clearly in the second or third category. Being part of a larger association of websites, LJ may indeed have need to hack MT to fit corporate structure and then share that with sister corps under the same structure, hacks that would make no sense incorporated into the mainstream MT, so it could well be that MT won't cut it for them. MT is up front about the rules, it's their perogative to make those rules, and I may not like them or even understand them, but that's what they did and I can accept that. On both sides, it may be purely political, but it may also be totally practical.

spaceman
2002-07-28 12:24:25
Blame it on ESR
[snip]
I'm surprised a site like ORA could post this and not realize the problems with the premises.
[/snip]


After your first paragraph, above, you seem to come to the same conclusion I did. There is no universal meaning and everyone has their own acceptable definition and comfort level.