Quite frankly, we’re dying off because the companies who use our software do not give back. The irony of the situation is that, in order to improve my motivation to do open source, I have to charge for it.
On the basis of what metrics does he come up with "dying off"?
The only graphs I have that show a reduction in open source contribution coincide with the financial crisis.
That is to say, when lots of developers lost their jobs with companies, contributions went DOWN.
I think that suggests the opposite case is true. There's plenty of people improving open source on behalf of companies, it's just that the companies don't feel the need to make a song and dance about it.
Re:I don’t believe it
Aristotle on 2009-07-15T04:06:33
Are you commenting on the quote or the entire article?
[long ramble here, but I have to go watch the Tour de France now.]
I had this discussion with a consulting customer a couple of days ago. They were considering making some of their work open source, but decided against it because it would be more work with no benefit to them since they didn't see any significant patches coming back to them. They concluded that the local benefit wasn't worth the local work. They would essentially give something away for less than free because the work they'd need to do distribute it and manage it in the public sphere when they users wouldn't ever even acknowledge its use, contribute fixes, or participate in a community.
That's a different argument than the total, global benefit of open-sourcing a project. The overall benefit of open source across the entire universe is a net positive. I believe that, and even if I didn't, I'd concede that fact. However, it's also like saying that every farmer giving away his crops for free is a net global benefit. Or the hard, unpaid work of a small group of people like p5p makes the universe a better place even if it doesn't actually benefit the individuals in p5p.
So, there's a big Dark-something out there just like we acknowledge the DarkPAN. You can't look at something like open-source contribution rates and say anything. How much should that rate be? I would think the rate should be virtually exponential (well, to some carrying capacity after which it might become asymptotic), so even an increase in rate might not be as large as it should be. I heard this argument about Twitter too. What should the participation rate be? There are many popular feeds, but most people are virtually unfollowed and stop significant push participation after a week (I didn't verify this).
I don't think any of us can say why open source contributions have gone down. I would have expected underemployed people to contribute more, but my economic model is just messed up. However, in talking to a lot of "regular" people about this in the last year, I can note a definite awareness of open source, but it's not the sexy thing it used to be (and note that venues like OSCON are re-defining themselves as there is less and less new to say about the idea open source). People are already past the idea that Microsoft software is necessary (or even necessary), but open source also isn't ready to replace it. Regular people just don't go for Linux on the desktop, for instance.
Clay Shirky has a talk about this where he shows a Zipf distribution of open source contributions. He talks about the long-tail where 80% of the people have only every contributed one patch, and he counts that as a benefit. That's the many-eyes theory, and I can't really argue against the overall benfit to the project. However, when I saw that data, I had the opposite reaction: most people don't participate beyond a single patch, so engagement is mostly really low. My own bias, of course, is that I send a lot of patches to projects.
Adam is struggling with the same idea. In his Top Fail 100, someone responded that he can't fix what people don't tell him about. I've discovered the same thing in my code. People code around bugs rather than even mentioning them to me.
Some people talk about the "social contract" of open source, but there is no contract because there is no exchange. There might be a "social expectation", but no contract in an legal sense of that idea. Maybe we should have a new license where you can only keep using it if you supply a patch.
Re:Local versus universal benefit
Aristotle on 2009-07-18T01:21:15
Ah yes… the long now and the big here…
Thanks for your musings, I’ll have to chew on all that for a while.
Re:Local versus universal benefit
chromatic on 2009-08-15T21:17:26
Maybe we should have a new license where you can only keep using it if you supply a patch.
I don't hate that idea, but I prefer requesting some kind of community contribution: filing a bug, making a feature request, sending a thank you note to the developers, answering a question on a mailing list, contributing money or hardware or resources, or any of countless other activities.
Such a license would be difficult to enforce (and a very bad idea for other reasons), but I like the idea of suggesting that organizations consider themselves part of a community which develops and maintains software.
Re:Local versus universal benefit
brian_d_foy on 2009-08-15T21:58:00
I wasn't serious about that license. I have considered, however, making my CPAN code "postcardware". If you use it you have to send me a postcard.
:)
