Saturday, April 26, 2008
The conference is free of charge. It will include scheduled talks, Lightning Talks, and unconference-style Open Spaces.
To get more information or to volunteer, see http://pyohio.org. See http://wiki.python.org/moin/PyOhio/CallForProposals for information on proposing a talk.
Wednesday, April 23, 2008
I hope there are a few people still reading! There was a lot to talk about in an 8-day, 1000-plus-attendee, volunteer-run conference like PyCon 2008.
Organizing a conference of this scale is a lot of work, much more than any one person could do. I didn't mention everyone who helped out... but you know who you are. To all the speakers, sprinters, attendees, sponsors, organizers, and volunteers:
I had a blast chairing this year's PyCon, and I hope that next year's conference is even better.
To me, PyCon is no longer primarily about the talks, or the projects, or the code. (I hardly got to see any talks this year, my projects are on hold, and I haven't written much non-work code lately…)
Don't get me wrong -- talks, projects, and code are important, especially to new attendees. These aspects used to be the most important parts of PyCon to me too. The talks and projects are what allow us to have a conference at all; they're the framework everything else is built on.
But to me, now, PyCon is about the people, and the community.
PyCon's motto is "Connecting the Python Community", and that's what it's really all about. It's about renewing friendships and making new friends. It's about exchanging ideas face to face. It's about sharing meals with your peers. It's about putting faces to names that you've interacted with over the sterile internet, adding human warmth to cold code (beyond the obligatory Monty Python references, that is!).
When I communicate with people I've met at PyCon, via email or IM or IRC or blogs, I hear their voices in their words, and I can picture their faces in my mind. This has great value to me.
I believe that many repeat attendees share this view, especially those who are deeply committed to the Python community. I hope that new attendees gain a sense of this as well. We have something special in our community and in our conference. As PyCon and Python grow, we will have to work to keep it.
(So, ONE LAST TIME:) As a growing volunteer-driven community conference, PyCon needs lots of help -- your help. No matter how much or little time you have to give, we can use your help!
I hope you'll join us, and I hope to see you at PyCon next year!
PyCon 2009 will take place in Chicago again. This year's venue was near capacity; we're currently exploring options that will allow PyCon 2009 to have more elbow room and even grow. Stay tuned for more.
This report has discussed mistakes that were made this year, which we will not repeat next year. If you have ideas or opinions that haven't been covered, please reply in the comments, email me (goodger at python dot org), or better yet, join the mailing list!
I will chair PyCon again in 2009. But while I intend to stay involved beyond that, I do not intend to chair PyCon 2010. Changes of leadership bring fresh ideas and perspectives, which is a good thing. We'll be looking for a "chairperson-in-training"; let me know if you're interested.
Very shortly we'll be deciding the venue for PyCon 2010: the deadline for bids is May 1, 2008. An earlier blog entry describes the process, which is much simpler than in previous years.
We won't be asking local groups to negotiate with venues as they did in the past. Instead, we'll accept bids from motivated local groups promoting their regions, and we'll all work with the meeting planners to find suitable venues. This leverages strengths and avoids weaknesses: few Pythonistas are experts in conference logistics and venue negotiations, which are time-consuming and potentially subject to costly mistakes.
Monday, April 21, 2008
PyCon has shown impressive growth. As PyCon continues to grow, we'll continue to outgrow venues. At some point growth will plateau, and eventually we may even see declines, but for the near future growth seems to be on the menu. (I'm not expecting PyCon 2009 to grow another 77%; but who knows?) Keeping the community/volunteer nature of PyCon as it grows will be a challenge.
We have to decide which direction to take. I see several options:
Cap attendance. We had about 1040 people at PyCon 2008, and the venue has a hard limit of about 1200 people.
Hopefully we won't need to take this drastic a measure. We'll do everything we can to avoid it.
Raise rates to limit attendance. This would mostly affect the attendance of hobbyists, independent consultants, developers at startups & small businesses, and students. One way to counter that would be to offer discounts or financial aid, but that entails a lot of additional work (I know, having handled financial aid last year, and advised Ted Pollari this year), and many would be reluctant to apply for aid.
-1 from me. Raising the rates would prevent many students and hobbyists from attending, which would turn PyCon into a corporate-only conference and remove all community spirit.
Continue to expand. This implies larger venues, which are often more expensive. Hotels only get so big, after which conference/convention centers come into play. We may have to charge more in order to make use of larger venues. Sponsorship may cover it, but we don't know yet.
In order to keep the conference affordable, we'll probably have to continue expanding sponsorship. As we saw this year, we have to be careful to keep the community spirit with expanded sponsorship. I think it's possible, and I think sponsors will be fine with the compromises they may have to make compared to larger corporate conferences. One advantage that an affordable community conference like PyCon has over corporate conferences is the investment that attendees make (self-funding, taking time off work, wanting to attend). That's very valuable to sponsors.
Split PyCon by sector, e.g.:
- "PyCon: Developers" for core developers, contributors, and advanced users.
- "PyCon: Learners" for new and inexperienced users.
- "EduPyCon" for the education sector; etc.
- Project-specific conferences.
Split PyCon by region: into two or more smaller annual regional conferences.
I don't like either of the splitting options. Forcing a split would segregate the community, preventing cross-pollination. Key attendees, like Guido, would not be able to attend all of the smaller conferences. And smaller conferences may not be able to attract the quality of speakers and sponsors that a single conference does.
I think regional and sector-specific conferences are inevitable, and good for the community. But these should begin at the local/group/project level. The PSF shouldn't start up such conferences, but we should definitely play a supporting role.
Helping user groups seems like a good route to encourage larger local events.
Help existing conferences to feature Python tracks. OSCON already has a Python track, but the PSF isn't directly involved with it (although individual PSF members are); we could participate more in the planning. Conferences such as LinuxWorld or Usenix could add such tracks; perhaps we could encourage them to do so.
We could bridge between the Python community and regional conferences such as Penguicon, offering help finding speakers, maybe funding travel, and sponsoring the conference.
The big advantages to this approach are that we'd only be concerned about content, and we'd be free to leave the logistics of food, venue selection, etc. up to someone else. The disadvantages are that we'd now be vulnerable to bad decisions by the conference's management (but the expenditure per conference would be much lower for us), and the PSF also wouldn't be getting a cut of any profits.
The current status of the PSF precludes much of the kind of support described above, as we're all volunteers with day jobs. I hope this status changes before long.
One area that was lacking this year, and that we really need to work on as we grow, was navigation. We had maps in the program guide, but not nearly enough signs in the venue itself. As PyCon grows it will be more and more spread out, so we need signage both to help people find where they want to go, and to remind people what is available (specifically: open space, expo hall, lounge, & registration/information).
This article incorporates text and ideas from Andrew Kuchling. Thanks Andrew!
PyCon is organized and run by volunteers. How?
The PyCon organizers are a self-selecting group. You can't make people volunteer, or make them do things they don't want to do once they have volunteered. All you can do is ask for help, and encourage people to take full ownership of their area. As the "leader" of the PyCon organizers, I was less a manager and more a cheerleader and a watcher of the "big picture". The people who helped to organize and run PyCon were doing what they chose to do, what they wanted to do. That alone provides plenty of motivation.
Boiled down to the essence, the PyCon organizers all wanted a great conference, and they did whatever needed to be done to achieve that.
However, we all have day jobs and lives, and some things fell through the cracks. PyCon 2008 was very ambitious, and many people put in enormous efforts. We didn't really have backups or deputies for most areas, but that's something we should consider for the future. Usually people supported each other and stepped in when necessary. If nobody did step in it was my job as chairman to recognize such tasks and situations, and find someone to delegate to whenever possible. As a last resort, I dealt with issues myself, but I was already maxed out and taking on extra duties mean they sometimes didn't get done. In a couple of (mostly minor) cases, that's what happened.
As much as possible I tried not to make unilateral decisions. We always discussed issues on the mailing list and in our meetings. The only times I had to exercise authority was when we couldn't come to a clear consensus (which was rare), and when the matter was urgent and there was no time for discussion. If we had planned better, the latter case wouldn't have occurred nearly as often as it did. But I think that overall, as volunteers, we did a very good job.
The PyCon organizers communicate electronically, by email, wikis, IM, and IRC. We're now discussing ideas for next year's conference. To join us, please subscribe to the pycon-organizers mailing list (all are welcome!), and attend our regular meetings on IRC when you can.
Wednesday, April 16, 2008
The website software and the registration system saw a complete overhaul this year, mostly implemented by Doug Napoleone. It was absolutely necessary, as we were unable to collect the data we needed previously. However, the registration system and the rest of the Django-based PyCon-Tech project is very ambitious and there's a lot left to do, and some rough edges.
Brantley Harris did an initial site design. And then, when we didn't have the volunteer capacity for some much-needed further work, we bought some discounted development from Imaginary Landscape (also a PyCon sponsor).
PyCon-Tech is an open-source project based on Django, and other conferences are looking into using it. Hopefully they'll contribute to it too! Of course, there's a PyCon-Tech mailing list where you can help out!
PyCon is a community conference run by volunteers, and is underwritten by the Python Software Foundation (a non-profit public charity). The PSF wants PyCon to be an accessible, affordable conference. To that end, every year PyCon is budgeted to operate at a (small) loss. But (almost) every year we have had more attendees and more sponsorship than the year before, and PyCon has always come out in the black. We're still receiving invoices and paying bills for PyCon 2008, and won't know the final results for a while yet.
Any proceeds go into the PSF's general funds and are used for the good of the community. For example, in recent years, the PSF has sponsored other conferences and events: EuroPython, PyCon UK, PyCon Italia, Jornada Python en Santa Fe (Argentina), and more.
PyCon introduced a new, higher "corporate" registration rate (the old "regular" rate was renamed to "hobbyist"), and I was pleasantly surprised by how many attendees chose the higher rate. This enabled us to be even more generous than we had planned with financial aid, among other efforts. Even at the higher rate, PyCon is a bargain.
Speakers Must Also Pay?
Some have asked why PyCon doesn't give all speakers free registration.
The consensus so far has been that in order to keep PyCon affordable for all, everyone should support the conference by paying their own way. There have been a few exceptions to this rule. Free registration was offered to invited keynote speakers and organizers. Not many took the offer, although those paying their own way were encouraged to take free registration.
When PyCon had 300-400 attendees, we couldn't afford to give free registration to 60 or more speakers. They represented a significant fraction of the attendance and PyCon's revenue. Now, with 1000+ attendees, free registration for speakers would be much easier to accept. But should it be automatic? I think we should continue to require speakers to ask individually. However, the wording in the CFP should be changed to make it clear that for speakers, waiving registration is easy.
This year for the first time, PyCon is working with professional meeting planners. Working with Conference Technology Enhancements (CTE) has helped a lot, and I wouldn't want to go back to the old way. Without CTE's help, either PyCon would not have been able to take place in anything near its present form, or I would have lost my job and gone insane. Possibly all of the above.
CTE handled the negotiations with the venue: room rates, catering, A/V, and more. Volunteers did this in the past (Jeff Rush in Dallas, Steve Holden in DC), and perhaps they enjoyed it (although I doubt it), but I wanted to have as little to do with negotiation as possible. Also, CTE continues to offer us a wealth of experience and ideas. For example, CTE originated the idea of significantly increasing the sponsorship levels while increasing our investment (in the Expo Hall, etc.), in order to properly value the conference. I couldn't imagine a conference of this size and complexity without the kind of help CTE brings.
There were some communication issues. While the PyCon organizers communicate by email, IM, and IRC, I had to get used to calling CTE on the phone in order to get a quick response. It was just a matter of getting used to each other's way of working.
We're learning, and I fully expect that this year's experience will improve our working relationship. I would not hesitate to continue our relationship with CTE.
This year as in years past (especially, I think, last year), the grass-roots blogging and word of mouth served to get the PyCon word out. This year we also tried something different: professional PR.
Last August Van Lindberg got in touch with Page One Public Relations, a PR firm specializing in open source companies. Page One offered their services to PyCon at a discount, and we took them up on the offer as an experiment, to promote PyCon and Python. Catherine Devlin and Van coordinated efforts with Daniel Schneider of Page One PR.
Daniel wrote and distributed four press releases, but much more importantly he arranged for multiple telephone, email, and in-person interviews with reporters, resulting in at least 25 articles, podcasts, professional blog posts, and event listings (listed below). I personally took part in several telephone interviews (5 or 6), some for podcasts, others for press. We don't know how many articles may have originated with the press releases.
When I asked Daniel about measuring the results, he told me that "You can measure the tangible results of PR by comparing the number of press clippings from last year vs. this year." I don't think we kept track of this last year (if it was even possible!), but I'm sure that we had a lot more coverage in the professional press this year than ever before.
Here's Daniel's report:
PyCon 2008: PR Report
October 2007 - March 2008
- PyCon 2008 received great coverage, both in terms of quantity and quality, for a small project. As a programming language, Python appealed to developer publications, such as Dr. Dobb’s Journal, as well as mainstream IT outlets like InfoWorld and eWeek.
- Most of the media opportunities occurred due to opportunistic pitching around PyCon itself, but also other cases arose to insert PyCon.
- PyCon 2008 chair David Goodger served as the primary spokesperson but select others, depending on topic, contributed as well.
- On top of written articles, PR for PyCon utilized audio components too, as there were [at least] four podcasts recorded and posted.
- InfoWorld, Paul Krill (11/21): “Python upgrades readied for 2008”
- .NET Developer’s Journal (a Sys-Con publication), Maureen O’Gara (11/29): “Python Creator Guido van Rossum to Present the Next-Generation Python 3000”
- Dr. Dobb’s Journal podcast, Mike Riley (1/28): “PyCon 2008 Conference”
- Dr. Dobb’s Journal, Jon Erickson (2/4): “Python in the News” in the Editor’s Eye piece
- Developer.com, Chris McAvoy (2/14): “Creating Excel Files with Python and Django” (contributed article)
- Dr. Dobb’s Journal (homepage—2/28): Podcast posted with picture of PyCon 2008 chair David Goodger and couple notes about the conference, with dates
- eWeek, Darryl Taft (3/3): “Sun Hires Python Experts”
- BZ Media Event Calendar (3/3): PyCon listed as upcoming event on upcoming conference schedule.
- LinuxOnline event listing (3/3): PyCon listed as upcoming event on upcoming conference schedule.
- Internetnews.com blog, Andy Patrizio (3/6): “Sun hires two more open source gurus.”
- E-Commerce Times/LinuxInsider, Erika Morphy (3/7): “Negroponte Goes Fishing for an OLPC Chief”
- O’Reilly FYI blog, Kathryn Barrett (3/12): PyCon Keynoter Van Lindberg shares insights on Intellectual Property
- Dr. Dobb’s Journal, Jon Erickson (3/13): “Does Geography Really Matter Anymore?” in the Editor’s Eye piece
- Dr. Dobb’s Journal blog, Mike Riley (3/15): “PyCon 2008: Day 1, Part 1”
- Dr. Dobb’s Journal blog, Mike Riley (3/15): “PyCon 2008: Day 1, Part 2: Guido van Rossum's Py3 Keynote”
- Dr. Dobb’s Journal blog, Mike Riley (3/15): “PyCon 2008: Day 1, Part 3”
- Dr. Dobb’s Journal blog, Mike Riley (3/15): “PyCon 2008: Day 2 Summary”
- Dr. Dobb’s Journal podcast, Mike Riley (3/16): “Silverlight and IronPython”
- Dr. Dobb’s Journal podcast, Jon Erickson/Mike Riley (3/16): “Johannes Woolard On Crunchy”
- Dr. Dobb’s Journal podcast, Jon Erickson/Mike Riley (3/16): “Guido van Rossum On Python 3 -- and More”
- Dr. Dobb’s Journal blog, Mike Riley (3/16): “PyCon 2008: Day 3 Summary”
- Developer.com, Chris McAvoy (3/17): “Building a Photo Gallery with Python and WSGI” (contributed article)
- Chicago Public Radio (Vocalo.org), Steve Walsh (3/18): Interview with PyCon 2008 chair David Goodger for general interest radio show [the interview starts at 1:31:30, or 1:30:25 on the on-line widget]
- InfoQ, Robert Bazinet (3/28): “Microsoft shows Django running on IronPython”
- InternetNews.com, Sean Michael Kerner (4/4): “Python Fans Take Aim at the Enterprise”
—Daniel Schneider, Page One PR
Sunday, April 13, 2008
The budget for PyCon 2008's financial aid was several times that of 2007, and we were able to help more people come to PyCon than ever before. Over 40 people received financial aid of some form. This is a trend I'd like to see continue.
Almost everyone who asked was granted aid. Only a few were turned down, either because their application was incomplete (after multiple requests for clarification) or because they requested an excessively large amount.
As I stated in my opening remarks, I believe I was the recipient of the first PyCon financial aid back in 2004, PyCon's second year. I hope that this year's aid recipients were able to get a fraction as much out of PyCon as I have, and I hope they give back to the community in turn.
Ted Pollari was in charge of financial aid this year. Thanks Ted! Ted could use some help next year...
We had some great conference swag this year. When the Python xkcd strip came out, the Python blogosphere exploded. It didn't take long before the idea arose -- from several people simultaneously -- to use it for the conference shirt. Ted Pollari was particularly enthusiastic and tracked down the artist to secure his permission. (And Randall Munroe has done it again, this time in a more philosophical way.)
We decided to upgrade the conference bag this year -- it's really sharp. In addition to being attractive, it's durable, and a useful size too. I can't wait to see one in the wild, and I fully expect to. Use your bags! It's great advertizing for Python.
The good people at Python Magazine worked with Van Lindberg to produce PyCon's first program guide. And is it ever slick! 60 full-color pages, a real collector's item. A special PyCon edition of the March 2008 Python Magazine was also part of the swag.
And stickers! We had thousands of Python logo stickers made up, with both white and black backgrounds. They instantly adorned most of the laptops in the place. PyCon sponsor Tummy.com also produced stickers -- ovals resembling country stickers for automobiles -- a welcome addition.
(Does anybody have pictures of the stickers? How about your laptops with stickers on? Add them to Flickr, tag "pycon2008"!)
Thanks to Don Spaulding for taking care of the swag production details.
We'll have to come up with designs and swag ideas for next year too. If you're artistically inclined, how about helping out?
Businesses are a big part of the Python community, and sponsorship is a major way for companies to show their support and enable us to keep PyCon affordable. Sponsorship allows companies to interact with individuals, raise awareness, recruit, and even sell some product.
We increased the sponsorship rates significantly this year, to bring PyCon more in line with other conferences of our size, and the sponsors responded generously and in great number. Sponsorship was at an all-time high: three times the budgeted amount. Among other effects, the exceptional response from sponsors enabled us to offer catered lunches on all four sprint days, and allowed us to record all of the scheduled talks. The results are beginning to be seen in videos published on the web.
Interesting aside: the lanyard sponsorship was a very popular option; several sponsors asked about it. Next year we'll have to add a significant fee on top of the simple production costs.
There was a mix-up in the timing of arrivals of conference bag inserts versus Expo Hall materials: not everthing was there when we needed it for the initial bag stuffing on Wednesday evening. We ended up re-stuffing the remaining bags on Thursday, and we made supplemental packets (including the conference T-shirts!) for those who had already picked up their bags. While we made every attempt to get all materials to all attendees, inevitably some did not return for their supplemental packets. My apologies to the sponsors who were affected. Next year we will ensure that all swag bag inserts will be on hand for the beginning of the stuffing session.
We filled one ballroom (the "Expo Hall") with booths for sponsors & vendors, with two other vendors in the Atrium/lounge together with the registration desk. This was new this year -- the most PyCon had ever had before was a few tables in the hallway.
The sponsors had a variety of giveaways which were popular. The usual suspects -- literature, CD/DVD-ROMs, and T-shirts -- were well represented, but there was much more. At the Symbian Press (Forum Nokia) booth, the authors of the Mobile Python book were giving away copies, and I got them to autograph mine. Wingware was giving away little balsa wood glider kits -- very apt. Rackspace had fortune cookies and elastic yo-yo-like things resembling multicolored jellyfish, which were very addictive. Lucasfilm had Star Wars Lego pens as well as decks of playing cards with different movie characters on each card -- very cool. IronPort Systems had little tins of mints -- tasty! It was good to see White Oak's squeeze-brains again (I still have a few from when PyCon was in DC; one is stuck to my monitor at work to remind me to use my own brain), and even their magnets were cool -- once we made sure people would keep them away from their laptop hard drives!
Note to sponsors: if you come up with great ideas for swag, it will be used! The more interesting and/or useful, the better. (For some reason, geeks tend to like toys and T-shirts. Or maybe it's just me. ;-)
From speaking with many of the sponsors, the Expo Hall was a great success. And next year, we hope to have more space so we can have more and larger booths.
Van Lindberg did a fantastic job coordinating sponsorship this year. You too can help!
After the lightning talks session on the afternoon of Sunday March 16 (by all accounts the best one, natch), we had an "Intro to Sprinting" talk & panel led by Brett Cannon. The talk was one of the few I actually saw (much of), and it should become an instant classic once it hits the web. This was followed by sprint tutorials, where each sprinting project went off and did introductions of various lengths.
The sprint tutorials seem to have been quite successful, and were a good way to segue between the conference and the sprints (a transition which felt a bit jarring in the past). Many new sprinters learned what they needed to be productive during the sprints themselves, and many people who couldn't stay this year may have been convinced to extend their stays next year.
There were over 20 projects sprinting. On the first sprint day (Monday March 17), there were over 250 people participating. There were 263 people at lunch, and I know of several who ate elsewhere; but some people who ate with us may not have been sprinting. I believe this was more than double last year's participation, and it continued that way through Thursday (tapering off, but still way above the 2007 numbers).
A list of accomplishments has been compiled (although woefully incomplete I'm sure). (Sprint leaders: sing your own praises by adding to the list!)
We are grateful to Canonical, Sun Microsystems, and Google for sponsoring the sprints. Thanks to them we had four days of hot catered lunches: Chicago-style comfort food. This was a welcome addition to PyCon at our Chicago venue, not least because unlike Dallas in 2006 & 2007, we weren't a quick walk away from dozens of restaurants.
Thanks to Facundo Batista for coordinating the sprints. (Sign up here to help!)
Open space was under-utilized at the beginning of PyCon. It was slow on Friday, and it only really kicked into high gear on Saturday evening. I don't think it was used much at all during the day, opposite talks. The open space rooms should have been used for more follow-up sessions after talks. Next year perhaps we'll offer the scheduling of open space as a service to speakers; a debate is going on now.
We had a split upstairs/downstairs venue this year, with talks in the ballrooms upstairs, and open space rooms downstairs. There wasn't an ideal place to put the open space schedule board. Upstairs it would have caused terrible bottlenecks, but downstairs (where we did put it) it wasn't "in your face" enough. In the future, no matter what the physical layout of PyCon, we need signs directing people to the different areas.
It seems that another reason for the slow start may have been because many attendees didn't know exactly what open space was, or how they could use it, or that they should use it. Next year we have to do a better job educating attendees on the features of PyCon. We may open the conference with a "PyCon how-to" session. Ideas are welcome!
There were ideas about having video monitors around the hotel displaying dynamic schedules including open space, so people could see what was happening (or about to happen) at any time. The projectors in the ballrooms could switch to these displays between talks. This didn't happen because we ran out of time. Perhaps next year... interested in helping?
One very pleasant surprise was Steve Holden's "Teach Me Twisted" learner-led session. By the time I remembered about it, it was almost over. I popped into the packed room to find a smiling, laughing, rapt audience; everbody seemed to be having a wonderful time. By all accounts was a great success, and learning was had by all. I wish we had recorded it. This is definitely something to try again.
Thanks to Peter Kropf for organizing open space!
Thursday, April 10, 2008
(Here's the part many of you have been waiting for! ;-)
We had more time for lightning talks this year than ever before: 5 hours total. I'm looking forward to seeing many of them when they show up on the web (the disadvantage of chairing PyCon is that I didn't actually get to see much).
Last year we invited sponsors to give lightning talks during a special lunchtime "sponsor lightning talks" session. This year's sponsor package repeated the offer of sponsor lightning talks. We didn't have any sessions at lunch time though, because that scheduling decision had mixed reviews.
Many sponsors took us up on the offer this year -- too many perhaps. It seems that 2007 was a sweet spot, with few enough sponsors that sponsor lightning talks were feasible. 2008 saw too many sponsor lightning talks, and they crowded out the attendee talks. It seems that the nature of the talks was not communicated as well as it could have been, leading to some confusion. There was a lot of negative feedback regarding the sponsor lightning talks this year, and nobody wants that, especially not the sponsors!
We've come to the consensus that in 2009 there will be no sponsor lightning talks, at least not during a plenary session. Sponsors will still be able to sign up for lightning talk slots, but they won't have priority or be able to sign up in advance. Perhaps we'll hold sponsor lightning talks during lunch again, or in a "sponsor/business/management track" (scheduled opposite regular talks), or in some other form.
Perhaps there's no longer a need for sponsor lightning talks, since we had a very successful expo hall. The expo hall should only get bigger in 2009. That may provide enough opportunity for attendees and sponsors to interact. A few sponsors also took advantage of the breakout rooms to schedule receptions -- yet another avenue.
We also need some guidelines on what makes a great lightning talk: informative, entertaining, energetic, interesting, fun, quick. Too many of the sponsor lightning talks fell into a pattern which ended up being boring. And 5 minutes is a maximum, not a requirement.
Jacob Kaplan-Moss and Mike Orr coordinated the lightning talks this year. Along with the speakers, they should be given credit for the successes, but they should not be blamed for the problems. The problems were a result of collective decisions, circumstance, and the explosive growth of attendance and sponsorship. Thanks Jacob and Mike -- great job! (Correction: Richard Jones lent moral support only ;-)
We experimented with the keynote format this year. Rather than daily 90-minute talks, we had 60-minute and 15-minute keynotes. The format seemed to work well.
The choice of keynote speakers and talks also seemed to work well. It's impossible to satisfy everyone, and foolish to try. I believe keynotes should not be music to everyone's ears, but should be challenging as well as informative. I think this year's mix of technical and general talks was successful. What do you think?
We also had short keynotes from our diamond sponsors (White Oak Technologies and Google). While I heard some grumbles of discontent, most feedback was very positive.
Should we change the keynote format further, keep this year's format, or return to the old format? Should we continue having diamond sponsor keynotes in 2009, or drop them, or modify them in some way? We'll be discussing these issues and more over the coming months.
I'd like to thank Noah Kantrowitz for wrangling details from keynote speakers.
Wednesday, April 09, 2008
Carl Karsten was a real hero of PyCon 2008. His seemingly boundless energy and wealth of ideas catapulted the technical aspects of PyCon far beyond what we'd had before. He worked with the hotel's A/V staff to record every talk, and is currently leading the effort to organize, edit, and post the videos to the web.
Tip to organizers of other conferences: get a clone of Carl. You can't have Carl though, he's ours.
I first noticed Joe Baker going from room to room on tutorial Thursday, changing tapes in the video cameras. Joe appeared as if from nowhere (from my perspective), and it was great having him there.
Cosmin Stejerean, who ran the Python Lab in open space, also helped out the A/V effort a lot. At one point there was a cot in the A/V room -- Joe and Cosmin were rotating between sleeping and swapping tapes in the cameras for digital video playback to computer files. That's dedication!
When we realized that we wouldn't be able to offer as many tables as we wanted (because attendance was so high), we decided to provide power to every seat. Carl, Sheila Miguez, and Peter Kropf went off to Fry's and bought 8 carts full of 48" power strips (they had to borrow the valet luggage carts to bring them all in). In a subsequent visit to Fry's to pick up some digital video cameras to get the video from tapes to computers, the store greeter recognized Carl and Sheila and told us, "You can't have any more power strips!"
We were going to have the hotel A/V crew tape down all the power strips, but it ended up being a bigger job than they thought. So Carl rounded up some people and they got them all taped down.
And guess who took all the power strips out of the hotel, and put them in storage? You guessed it: Carl (undoubtedly with some assistance from Sheila). The next time you see Carl and Sheila, or Peter, Joe, or Cosmin, be sure to buy them a round!
The PyCon Conference Days took place from Friday, March 14 through Sunday, March 16. More than 60 speakers gave freely of their time, and presented some great talks. On behalf of the PyCon organizers: thank you all!
We received over 140 proposals for talks, more than twice as many as we could accept, and unfortunately we had to turn down many good proposals. The PyCon Program Committee (all volunteers, led by Ivan Krstić) did a great job in the long selection process, but it's hard to balance the needs of beginners and gurus alike. It's inevitable that there were some complaints.
The only way to improve in the future is to have more participation, especially from under-represented sectors: core Python, advanced topics, PyPy, etc. There has been talk of switching to a topic-based track program (web track, advanced/technical track, introductory track, etc.). This could work, but it needs the participation of many experts. If you care about Python's North American flagship conference, join the pycon-pc (Program Committee) mailing list!
There was an issue with the printed schedule: the talk levels weren't included. Unfortunately, nobody caught this in time. We won't repeat that mistake. The levels were visible on the online schedule though.
There were some issues with talks and speakers during the conference itself. We had ideas about a speaker ready room and info packets for speakers and session chairs, but there wasn't the time or manpower to implement all the ideas we had. In future, this needs to be organized earlier and publicized better.
We had 4 tracks this year, instead of 3 in past years. How did that work out? Should PyCon remain at 4 tracks? Return to 3 tracks? Or increase to 5 or more tracks? Given the size of conference PyCon has become, we now have the freedom to choose. Also, the cost of A/V becomes more reasonable when it's spread out over more attendees, and this opens up the possibility of recording and subsequently (or simultaneously?) releasing everything.
Some people would like to see more, shorter talks. Many others believe that 30 minutes is not enough time for an in-depth talk. But if we increase to 45 or 60 minutes, we'll have far fewer talks. We can mitigate this with more parallel tracks, but then people have to make hard choices, and if any one talk is overwhelmingly popular the (inevitably smaller) room could be filled.
There is a lively discussion going on right now on the pycon-organizers mailing list. Join us!
Tuesday, April 08, 2008
This is a response to Titus Brown's article, "PyCon '08: The Brain Dump". My response was too long to post as a comment.
Titus' text is in block quotes (indented):
I wholeheartedly support the adoption of an advanced-technical-only track. As it was this year the talks I was interested in (mostly very technical) were embedded in the middle of a bunch of other talks that were not technical. I wasn't up to picking them out of the mix.
Speaking of "good talks", I think the whole review system is effed up. What's with the anonymous authorship of proposals?
There's an anonymous review phase, during which proposals are rated based on the information we were supplied. This is to level the playing field and allow new speakers a chance. Then there's the decision phase, during which the proposals are NOT anonymous, and the experience of the authors is taken into account. We have debated the question of anonymity before, and it may change in the future. Those who participate get to choose the process. (hint, hint)
Yes, it's a flawed system, but it's the best we came up with. I invite you (or anyone with a strong opinion) to join in and help make it better!
During dinner with Leapfrog people, a talk scheduling proposal emerged: rather than trying to group talks in some logical coherent way, why not try to minimize scheduling conflicts and auditorium changes by asking people what talks they want to go to?
We had that capability (sort-of), but too late to be useful. The scheduling app was only avaialble a couple of weeks before the conference. Even if we'd had it months before, I wonder if people would have used it early enough for the data to be useful.
It wouldn't be too difficult to make an app that lists the talks and asks people's preferences, but we simply didn't have time. Even if we did, scheduling is a hard problem.
The conference support for tutorials was kind of minimal: normally we don't need anything more than a projector and a mike, but (for whatever reason) the conference organizers alternated between treating us really impersonally (sending mass mailings that ignored previous information we'd sent them) or really curtly ("No. That's your problem.")
I know that there were some reminder mailings that went out to all instructors, even though the content may not have applied to all.
I think I found one specific instance you refer to: my March 7 reply to you request for info on the network. I replied, in part, "that's up to you". Who do you suggest should have taken care of it? We're all volunteers with day jobs; please cut us a little more slack.
If there were cases beyond this, could you provide specific examples? Feel free to write to me privately (goodger at python dot org). If there's a problem, I'd like to fix it.
Next year I may also ask for the tutorials to cover up to my own expenses (registration, hotel room and flights) from student fees, rather than having them simply give me $500 & free reg. I feel like I'm paying out for the privilege of giving each tutorial, and that's a bit frustrating.
PyCon paid $1000 plus free registrations for each tutorial. Your tutorial, with two instructors, worked out to $1440. Should PyCon have paid double because you had two instructors? Last year PyCon paid $50 per tutorial attendee, up to a maximum of $1500; no free registration though. That was a bit of a pain to administer though.
PyCon was budgeted to lose money -- includng the tutorials. (That PyCon probably ended up making a bit of money, for the PSF, was unexpected and thanks to the hard work of volunteers.)
OSCON tutorial instructors get reimbursed for airfare ($300 west coast, $500 east cost, $700 international), plus they get one night at the hotel, free registration, and a $500 honorarium. (I don't know how OSCON handles multiple instructors per tutorial.) One big difference with PyCon is that OSCON's registration & tutorial fees are many times PyCon's. OSCON is a much larger, more established and commercial conference, and I think they have more money to work with than PyCon does, but their tutorial compensation is about the same as PyCon's.
Probably they'll say "no", which will then leave me/us with the option of cancelling the tutorial or just sticking with it...
I can't say "yes", but I'm not saying "no". Let's talk.
we could also move the tutorial to a "sprint day" and encourage people to stick around for real "free consulting with grig and titus". I think we'd have more fun that way, and I'm damn sure we'd be more useful!
That would be great! We had a pre-sprint intro talk & tutorials this year, which seemed to work out well.
Oh, I almost forgot -- I'm now a member of the Python Software Foundation (unless they retract it for criticizing both PyCon and OLPC in a single post)!
On the contrary, this kind of feedback is valuable and wanted! The PSF needs active, passionate members to move forward. But even more than words, the PSF, PyCon, and the Python community needs action. Titus, through your stewardship of the PSF's participation in the GHOP contest you've shown that you're a man of words and action. Thanks!
I guess this means I'll have to run PSF/GHOP again, yeargh.
PyCon 2008 had 28 tutorials in three sessions (morning, afternoon, and evening) on Thursday March 13. Over 420 people attended, more than the total attendance of PyCon 2006. Almost all of the tutorials were full. More than half of the attendees took 3 tutorials, resulting in a very full day for many people. I'm sure more people would have taken more tutorials if they hadn't filled up.
We were using the ten largest breakout rooms, in the lower level of the hotel. They were of different sizes. The five largest rooms could handle up to 70 - 90 people, according to the hotel's capacity chart. We limited registration to 50 in these rooms. The other rooms could handle anywhere from 18 to 36 people. We limited registration in these rooms too, cutting about 20% off the maximum, to allow for elbow room.
It's a good thing we set these limits, because the hotel capacity charts leave out one crucial detail: the maximum capacities listed are for a room crammed full of tables, with no space for the instructor! I suppose if they're running some test like the SATs, they could actually fit that many people. But we needed room for the instructors.
During registration, Greg Lindstrom (the tutorial coordinator) and I watched the tutorial numbers carefully, assigning tutorials to rooms as soon as they (or other) tutorials reached certain numbers. This worked well enough, although it was a bit nerve-wracking. It would be nice to automate the process next year -- but a logic error could have serious consequences.
On Tutorial Thursday the wireless network didn't work. I wrote about it in part 3, and Sean Reifschneider's report goes into detail. The result was that students weren't able to get access to the internet, which was a requirement for many tutorials. Next year, we will have wired ports available in the classrooms, so even if the wireless fails we'll have a backup plan. That was lacking this year.
Once the lack of wireless was worked around, the tutorials themselves seemed to go well. We had enough people taking tutorials that our overhead was more than covered, and we were able to offer dinner as well as the expected lunch.
By the evening session I could see weariness set in. 9 hours of learning (12.5 if you include breaks) is a long time. Next year, maybe we'll try two days of tutorials, either both before the conference, or one day on either side (one coinciding with the first day of sprints). What do you think? This is another topic we're discussing on the pycon-organizers mailing list.
I taught two tutorials myself, wxPython I & II (afternoon & evening). I underestimated the amount of free time I would have to prepare in the weeks & months leading up to PyCon, so I was not really ready. The first tutorial was about 3/4 prepared (although there was no handout). For the last part of the first tutorial and for the entire second tutorial I did code walkthroughs and Q&A. The second tutorial became a "master class". I hope it was worthwhile to attendees, and I apologize for being underprepared.
Next year, I vow not to present any tutorials or talks beyond my duties as chairman. If I start to indicate a desire to present something, please direct me to this post.
Our original plan was to have plated, hot lunches. When attendance went through the roof, we had to abandon that plan. The problem wasn't the hotel's ability to produce the food, it was their ability to serve it, and provide seating. There just wasn't space for everybody to sit at a table, so a "sit anywhere" approach had to be substituted.
Next year we hope to solve that issue.
Our meeting planners worked with the hotel's chef and catering staff to customize the "boxed lunches" we had, with soup and other hot side dishes. With the space constraints (that we imposed upon the hotel, with PyCon growing to over 1000 attendees!), they did the best they could.
There were some complaints that the food & drinks were being removed too quickly. I spoke with the hotel about it, and that seemed to solve it. At least, I didn't hear any complaints after that.
Whatever the critics say, I know I'll be back in Chicago next year for sure. I just want better network connectivity (why is it so hard to ensure decent wireless connectivity at PyCon year after year? it's a mystery) and better food.
Yes, the wireless network this year was less than stellar. It was nearly unusable on tutorial Thursday and only approached usability Friday afternoon. The equipment deployed precluded wired connections in many rooms. For a detailed analysis and recommendations, see Sean Reifschneider's report.
Two years ago, PyCon's first year in Dallas, we trusted the hotel when they said they could handle our wireless networking needs. That was a mistake: they couldn't. It took a lot of effort, by Sean Reifschneider and others, to make their wireless configuration usable. Last year, Sean designed a wireless architecture for us and we deployed our own equipment. It worked beautifully, but it ended up being a working conference for Sean.
This year, we wanted to let Sean enjoy the conference. As Tummy.com, he's also a sponsor of PyCon, and a busy guy. Last fall when we were getting the hotel's buy-in on our networking needs, someone found a company in the Chicago area who said they could do the job. It would be more expensive than us (Sean) doing it ourselves, but would allow us all to enjoy the conference. Obviously, that didn't work out. We trusted our wireless supplier, but they really weren't up to the job.
As Sean states in his report, "Nobody really understands our networking needs like we do."
Next year we will do the networking ourselves. We will have wired connections in all tutorial, open space, & sprint rooms, as well as to all talk podiums. Our wireless will conform to Sean Reifschneider's plan.
On a brighter note, our bandwidth provider, BOB Broadband, was great. We had zero problems there.
From 2006 to 2007 we saw an increase of about 43% in attendance, from 410 to 586 attendees. I was expecting 800-900 people for 2008, but I feared we would see more... and my fears were realized with over 1000 attendees, an increase of 77% over PyCon 2007. We had over 420 people registered for tutorials on the Thursday before the talks (March 13), more than the total conference attendance in 2006 (the first year with a PyCon tutorial day). Attendance at the development sprint was at an all-time high: at least 250, possibly many more.
Registration started late this year, as did early-bird registration. Early-bird registrations accounted for about 75% of the total, and about half of the early-bird registrations were done in the last week (300 over 4 days; over 100 in a single day!). Next year we must open registration earlier and close early-bird registration much earlier, so we have some real numbers to work with early enough to be useful (e.g. for swag orders & catering plans).
The registration system itself had some rough edges, which we've noted and will be working on. It's quite something that there were only rough edges though, since PyCon-Tech was mostly written by one person in his (not-so) spare time. Doug Napoleone could sure use some help on the PyCon-Tech project!
In a classic snafu the credit card processing service used by PyCon & the PSF up and quit the day before the end of early-bird registration. It wasn't our fault or specific to us (many other online businesses were affected), but that didn't stop complaints. Andrew Kuchling and Doug Napoleone did a great job behind the scenes solving problems and soothing people's anxieties.
The phenomenal increase in attendance caused us some growing pains. We had to revise the seating plan in the ballrooms to fit more people (we'd planned to have more tables, but we couldn't do that and fit over 1000 people). The catering plan had to be reworked. The wireless network was affected. I'll address these issues in parts 3 & 4.
PyCon 2008 has come and gone. Overall, it was a great success. We tried some new things -- some worked, others didn't. There were some problems, which I will address in later posts, and which we'll correct in the future. I think that the PyCon organizers -- volunteers all -- put on a great event.
I've been collecting notes since before PyCon for a grand report. I was adding to it, polishing it, and it was growing to the point where it would be too long for anyone to read the whole thing. Also the longer it takes to post the report, the less relevant it may be. So with "publish early and often" and "just publish!" in mind, over the next week or so I'll post what I have, in small chunks, and see how that goes. We won't know how many parts there will be until there's a post titled "Conclusion".
This way, each post can start a separate, focused discussion, here and on the pycon-organizers mailing list. Please join and help out!
Friday, April 04, 2008
Kevin Dangoor, Michael Foord, and I participated in interviews with Sean Michael Kerner of InternetNews.com. Here's the result: "Python Fans Take Aim at the Enterprise".
Apart from a couple mistakes (it's the Python *Software* Foundation, and I'm a director, not the director), it's a good article.
I still have my original (extensive) answers to his questions, which we should be able to find a use for...