I think a community that is decentralized is stronger than one that's centralized.

There are a number of things that are all mixed together under the open source label.

1. Licensing / availability of source code
2. Charging for software or services around software -- Get Robert Lefkowitz to explain the Islamic notion of Sharia law, and you'll get over the notion that "it's free"
3. Community based/distributed/decentralized software development
4. Whether or not people working on software get paid to do it or not -- note that this is separate question from whether we charge money for the software.

Earlier today, Dave Winer posted on the topic of getting paid to work on software. I obviously agree that this is important, since I'm getting paid to work at OSAF. But we are going to make the source code available, we are not going to charge users for it, annd we want to do community based/distributed/decentralized software development. We've moved some of the sliders. Will it work? We don't know for sure, but it's an experiment worth doing.

Dave also pointed to Jim Fawcette's editorial on open source. As I've posted before, I agree that it's time to stop copying -- actually its easy to say, because we've almost run out of stuff to copy. So now we get to find out whether or not an alternate model of building software can produce something innovative. But that aside, I have a different explanation for what's been happening. Everyone whines about how Microsoft can innovate faster because they control the OS/platform, have access to all those hidden system calls, etc. It's called standing on the shoulders of those who have gone before. So over the last n years of copying, the open source community almost has a decent platform that it totally controls.

Why does this matter? Go look at Tim O'Reilly's wishlist for 2004 and look at his wish for Dashboard. Nat and friends were able to implement Dashboard and instrument all the important Dashboard apps in a few weeks before OSCon. True, Nat had been working on some of the code sporadically since February, so if you want to, you can call it 6 months. It doesn't matter. What matters is that you can make these kinds of changes much faster when you have the source to the whole system.

The better question to ask is why haven't any other people from the teeming open source hordes stepped up to finish the job that Nat and the other Dashboard hackers started. After all, it's open source, right? I don't have a good explanation for that, but if someone is looking for a cool, innovative open source project to get involved with, they should go help Nat finish it. And when you're done, then the community can go on to do any of a number of interesting experiments on top of the Dashboard code base -- try doing that with a shrink wrapped no source code product. Much as I like Mac OS X, Safari and Mail.app won't be speaking in Dashboard packets in the next 6 months.

It all started innocently enough. Last night Julie told me that she had gotten on the aggregator train. I was impressed and curious as to which aggregator she had chosen. I was less happy when I found out it was Bloglines. I have nothing against Bloglines (I've never seriously used it), but I know that Julie's goal in using an aggregator was to spend less time blogging, and I don't think that a web based aggregator can compare to a client side aggregator for efficiency. So I suggested that she try a different one. I've played with RSS Bandit and there were some recent laudatory posts about the latest versions, so this morning I downloaded a copy (after doing Windows update for Win2K on the Thinkpad, rebooting, installing a newer version of the .NET framework, and rebooting...) and installed it. Things seemed fine, at least until I started adding feeds. The first two feeds I added were hers and mine. RSS Bandit choked on both. Now we have a internal/external network setup, complete with split DNS and a whole bunch of other stuff. I figured that might be a problem, and started tweaking. The deeper I got, the more I realized it wasn't going to work. I foresaw many pleas for technical support followed by frustration -- I mean, *I* was frustrated. So I dumped that and went for Plan B, as it were. FeedDemon. I've felt bad that I used FeedDemon so long as a beta tester and then Nick never got any money because I switched to the Mac. But he's gonna get paid now. Julie took a quick drive through the opening wizard and seemed like she could deal.

After dinner, we did a quick tour through FeedDemon. Our two feeds added fine, along with some others. Everything was going great. Then Julie asked, so how do I blog from the aggregator. I remembered something about w.bloggar, and I've played with it from time to time but always dumped it because pyblosxom only has blogger API support, which means it doesn't read my categories, and that just won't do. In any case, I downloaded a copy of w.bloggar, and got it configured for Movable Type. Posts were showing up in the editor, and all was good. So then we tried to "Blog this item" from FeedDemon. There were some weird refresh problems, but sure enough there was the info all ready for the posting. Life was good, I was making big points. Then I tried to get rid of the test data, so I hit delete. Turns out that I didn't delete the test post, I deleted a post that Julie had been drafting, that she had invested about half an hour in. Now, when you are a parent of three young children, a half hour of your "no kids" time is a lot. I went to the Movable Type admin pages, hoping for some kind of undo or such. No such luck. So, due totally to Pilot Error, I killed a precious half hour. I was apologizing up and down... Fortunately Julie just looked at me and said "It's only a post". The shoe was surely on the other foot, and I was thankful to be on the receiving end of grace.

Mistakes were a theme today. At work I've been working on unit tests for the Chandler repository, trying to learn my way around the system, and the APIs. It's been a good experience, I've learned a lot and turned up some bugs. So on Friday I asked Andi to review the test code that I have written in order to make sure that I'm using the API as designed. Today he started IM'ing me his feedback -- a nervous thing for me because I've been wondering if I've really been using the API's "as designed". I asked him to post his diffs and commentary to the Chandler dev list, so that others could benefit from my experience. Of course, another way to look at it would be "so that others could benefit from my mistakes" - but I try not to look at it that way. In my mind, this is an essential part of the worldview necessary for open source / community based software development. We don't hide our problems and mistakes, or try to pretend that they don't exist. We recognize that they are there, and we illuminate them so that we can deal with them. That way others can learn from our experiences and we (I hope) make it easier for people to get involved.

It's nice when the things that work at home also work at work...