In review: linux.conf.au 2006
Probably the thing I most love about linux.conf.au is that every year is an experiment. In 2001 we changed its name, location and time, in 2002 we tried moving it outside Sydney and Melbourne and extended the programme to cover Monday and Tuesday, in 2003 we moved it to the furthest capital we could find and trebled the events on the Monday and Tuesday, in 2004 we doubled the events on Monday and Tuesday again, worked with AUUG to add a “government” co-conference, and in 2005 we moved it out of its standard late-January timeslot into April, and added a quiz show and lightning talks. And obviously handing the conference over to a new organising team (usually with no prior experience) and a new city each year is a risk too.
LCA 2006 had quite a few interesting experimental features, not all of which were entirely deliberate.
The most obvious change was moving the conference from .au to .nz. There were a few rationales: that .nz isn’t significantly more difficult to get to than Perth, so shouldn’t be considered specially, that introducing New Zealanders to the conference would broaden its appeal in future and give Linux in NZ a kickstart that it wouldn’t otherwise get, and that the NZ team was the the most enthusiastic group interested in running the conference at the time.
As far as being as easy to get to, that wasn’t quite the case: a few people had difficulties getting passports, and while many folks could get cheap Freedom Air fares direct to Dunedin, others weren’t so lucky. So that theory turns out to have been only a qualified success — it came close, but it definitely was a bit more difficult.
Having the conference in NZ definitely did introduce a bunch of kiwis to lca, though, and looks like it might have generated a few stronger ties. There was a BOF on NZ OSS, the New Zealand Open Source Society, where a bunch of folks from various NZ Linux user groups hashed out some ideas for re-establishing that group as a body more like Linux Australia in NZ that can be there to support projects that ought to be bigger than just a single LUG — I think the plans that came out of the meeting were to do things like arrange for interesting speakers to tour NZ LUGs every now and then, and to help coordinate this year’s Software Freedom Day activities in NZ. It’s hard to say for sure how well that’s going to hold up until we see if we have a bunch of attendees and speakers from NZ at next year’s conference, but at least at the moment, that hope looks like it’s turned out perfectly.
An additional aspect is that NZ is a lovely place to visit, and holding a conference there gives attendees the opportunity to take some leave either before or after the conference to tour around some very picturesque scenery. From the delegates list, it looks like quite a few people have done that, myself included.
All in all, I think the New Zealand experiment turned out pretty well; while it’s not something that should happen that often, I can certainly imagine a bid from Auckland, Wellington or Christchurch being seriously considered in a few years.
Billions of streams
The main portion of the conference has traditionally had a few lectures at once, normally three or four. This year had six. That means every single attendee, even if they attended everything they could, missed out on something like 80% of the conference. On Monday and Tuesday, there were eight miniconfs. There was a lot of technical content this time round.
There are two risks there: one is that in trying to fill that many slots you might end up accepting some low quality papers; the other is that you’ll have more people who see abstracts for some talks they really want to see, only to find they clash. It’s hard to say that those things didn’t happen — because it’s hard to prove a negative — but from what I can tell, it actually turned out pretty well.
That level of parallelism has a couple of upsides of course: notably that the more good talks there are at any one time, the more chance there is that any given attendee will be interested in one of them; and that by having the attendees split amongst more talks, you can hold them in smaller rooms making it easier to find a suitable venues.
Another aspect of the scheduling was mixing the tutorials in with the talks. I’ve no idea whether this worked well or badly or didn’t make a difference, though.
A side aspect is that Dunedin is actually a fairly small city; it’s population is only 122,000 it’s 10% of the size of Brisbane, Perth and Adelaide, and a third of the size of Canberra. While it does have the benefit of being a university town, it’s still quite notable that even though lca’s grown quite a lot, we still have a fair degree of flexibility in where it can be held. So Cairns, with a population of 140,000 ought to be able to hold an lca pretty easily, though Broome, with a population of a mere 14,000, might still be pushing it.
One of the “dropped” balls of this years conference was the media aspect; I might be mistaken, but I don’t think there were any press releases related to the 2006 conference put out in 2005 at all.
The first one was, ttbomk, a press release about Pascal Klein winning the “Community Member of the Year” award. (Pascal then went on to be part of a winning team in one of the competitions, switched from KDE to Gnome, switched from SuSE to Ubuntu, volunteered to help the folks running the next lca and got a lift home on a private jet. Not really a bad week’s work)
There were a couple more press notes afterwards, including a LinuxWorld Australia piece giving a general rundown of the conference, and a ComputerWorld New Zealand piece doing much the same. And that’s more or less it for the advance press, as far as I know. Unlike past years, we didn’t have slashdot as a media sponsor, so didn’t have a banner ad or some gratuitous stories about registrations opening or the call for papers.
There are three possible negative consequences to this: fewer speakers, fewer attendees, and missing opportunities to get messages out to people who don’t attend. Obviously the number of speakers wasn’t an issue. The number of attendees might have been an issue: this year didn’t sell out — but on the other hand, the last figures I heard were around 450 attendees out of a maximum possible of around 500. Which compares pretty well to the previous couple of years (which sold out at around 500), and may have been caused by previous issues — namely holding it outside Australia, or holding it in a small city, with a consequently smaller number of local Linux hackers.
Getting the message out doesn’t really require press in advance of the conference though — it can be done while the conference is happening or after. And there was a reasonable amount of press after the conference, including some newspaper stories, a short tv piece on the daily news, a number of articles about Samba 4, a series on lca2006, blog style, the Annodex Launch, the Linux Australia elections, and Van Jacobson’s talk. One of the nice things about these articles is that, excepting the Samba release, they were written by journalists who thought the topic was interesting to their readers, rather than being built up from a press release. I like it when news works that way.
I call this a “dropped” ball, because while it was (afaik) something the lca team had intended to work more actively on, it wasn’t forgotten so much as treated as a lower priority than making the conference work, and when making the conference work turned out to take up all the time they had available, it didn’t happen. And it looks like it’s worked out fairly well — which is good in that while future lca’s will hopefully do more press stuff, we can also be confident in saying that if the press stuff doesn’t get done well, it’s not likely to be a huge crisis.
A couple of things weren’t risks. Having a conference wiki editable by delegates was a great idea. HP and IBM continue to be excellent major sponsors, and definitely seem to grok how to make that work well. Offloading networking stuff to the Waikato guys appears to have worked exceptionally well, and the stuff they’ve written to manage the network will probably provide a good basis for future conference networking setups. Changing the t-shirt auction to a raffle was a neat tweak, given the amount bidding was getting up to in previous years. Which meant that at the dinner Rusty could instead auction off a copy of the John Lions book, autographed by various Unix luminaries, with money raised going to founding a John Lions chair at UNSW. Which ended up raising $10,000, matched by Linux Australia for $20,000, with additional contributions from delegates, all of which will eventually be matched by USENIX for a total of over $40,000.
Of course, underlying all the experimentation are the traditions, all the things that weren’t changed. From what I can tell, the primary formula still seems to be get interesting speakers, get a bunch of attendees who like helping each other out, and treat them all as well as you possibly can. And Mike, Drew, Kelvin, Gobby, Nick, and Alex seem to have done that to a tee.
(All of which is pretty fortunate given the bad puns that could’ve arisen from the way Kiwi’s pronounce ’06, or the airport code for Dunedin…)