Jim Driscoll's Blog

New way to attend JavaOne talks

One nice thing about being around this stuff forever: You end up on all the interesting committees. I'm on the JavaOne Program Committee, which helps decide the content at JavaOne.

Every year at JavaOne, we have the same problems - some talks are in big rooms, and you can hear the crickets chirp. Others are in medium (600 capacity) rooms, and there are people sitting on the floor - or worse, turned away at the door. We do the best we can at guessing, but it's impossible to get it right 100% of the time. And hastily thrown together overflow rooms are only a patch on the problem.

So, JavaOne is going to be run a bit differently this year. Instead of queueing up outside the talk you wish to attend (or camping in the room prior to the previous talk ending), there's going to be reserved seating.

No, really... reserved seating. You'll be required to register ahead of time for the talks you wish to attend, and everyone who's registered will get to go in first. Only after all the registered people go in will the unregistered people be allowed to go into the room. This also means that the rooms will be cleared after each talk.

I have to say, when I first heard this, I was a little skeptical - the Esplanade, in particular, tends to get really crowded with human traffic at every talk change. Visions of mobs of unhappy people milling about in that narrow corridor quickly sprang to mind. I wasn't the only one who was skeptical, either. But the conference organizers assured us that they'd done the same thing for other conferences as well (many as big as JavaOne), and they had answers for every objection that we threw at them. In the end, we were convinced - moving around 10k+ people is a difficult job, but it's what they do for a living, and they're pretty good at it.

Doing things this way has some obvious benefits - for one thing, it'll allow us to change our plans for how we'll be giving talks BEFORE everyone shows up at the door. It also has some interesting benefits that may not be readily apparent at first glance - such as allowing us to really plan a better JavaOne next year, since we'll finally have a truely accurate attendance number for each talk.

So, I'm optimistic about the process. But it will only work if everyone goes out and registers for the talks they want to attend. So, if you've already registered for JavaOne, please don't skip this step.

As always, I'd love to hear what you think of this idea. Let me know below.

Comments

• One question you have to ask yourself is if the organizers have the same main goal as you (Sun) do. Maybe they want a good and clear organization of the show and you want the attendees to see as many good sessions during the show as possible?

  Then there is always the conspiracy theories where the less conspicuous is the one where you are forced by the show organizers to do it this way and you just have to explain that to us in a way that will make us not freak out. ;)

  I usually (I've attended JavaOne twice before) change my schedule every now and then during the show. Maybe I talk to someone about something and want to attend a session about that, or I meet/see a speaker and want to gosub his session instead. Or, I meet another developer from Sweden that is attending my second choice for a time slot and we go there for some pre/post chat. With the mandatory Session Builder that is very hard since you need to be at the session earlier and you have to get to a web-computer to see if the show you want to change to is full or not.

  In any case, IMHO you are approaching this the wrong way. Instead of making the Session Builder mandatory for attendance (just the mandatory part will piss some off i guess) I would create a nice and useful application that people could use to plan and re-plan their visit. Something that would be of value to them. This JWS-startable application would be a way for conference attendees to communicate (as in chat?) and get updates to the schedule live. They could have more than one session planned for a time slot so they can decide later what to attend.

  If just 20% of the people used this application you would have good enough statistical data for correct room allocation. That data mangled with the data for earlier years (such as the fact that this or that kind of session have this or that ratio between pre-planned and actual attendance) should make it even very good.

  Cheers,
  Mikael Grev

  Posted by: mgrev on May 05, 2006 at 12:32 PM

• I guess it works a lot like college, except I'm not sure that if for a given time period, if one session is full and another one isn't filling, that Sun couldn't be agressive in rotating through locations for the given time-- swapping larger audencies into larger rooms and opening more room for attendance.

  Posted by: jhook on May 05, 2006 at 02:20 PM

• Actually, 20% is NOT enough for analysis, unless it's randomly distributed, and then we need a lot less. Unfortunately, all the previous reporting wasn't randomly distributed - it tended to strongly undercount some people, and overcount others.

  RE: Conspiracy theories - what possible reason would I have to lie? Wouldn't just NOT blogging be the wiser course if I didn't agree?

  Posted by: driscoll on May 05, 2006 at 02:41 PM

• Oh, and here's a blurb from the show registration email that may explain a little more concisely how it works:
  *IMPORTANT: YOU MUST ARRIVE BEFORE THE SESSION BEGINS TO SECURE YOUR SEAT.
  Space may be limited, and empty seats will be given away three minutes prior to session start. Schedule Builder allows you to enroll in technical sessions and Hands-on-Labs only; attendees will be admitted to BOF and general sessions on a first-come, first served basis. Session room information will be on your schedule - remember to print and take your schedule with you.

  Posted by: driscoll on May 05, 2006 at 02:42 PM

• Jim,
  When you are doing your stats are you correcting these with for instance the type of personality that are likely to attend the type of session in question? I thing that would increase the quality of the predictions a lot.

  Say for instance you have a business session. Those are mostly populated by business people and they are less likely to use a Scheduling Tool and that has to be compensated for by using data from prior shows (this might be correct but only data from earlier shows can confirm this).

  You can even do this in many dimensions, not only the profile of the attending person but the time of day, hot vs mature technology, speaker nationality, how famous the speaker is and so on.

  Since you have data from ten years you can even fine tune the model and compensate for trends!

  Cheers,
  Mikael Grev

  Posted by: mgrev on May 05, 2006 at 03:38 PM

• When will conference rooms for all sessions be published??

  Posted by: alexismp on May 07, 2006 at 05:06 PM

• Conference rooms for sessions are published - look at the link in my main article.

  Posted by: driscoll on May 07, 2006 at 07:33 PM

• I really don't like the idea of clearing the rooms between sessions. In some of the more congested areas people won't just have to leave the room but will have to walk some distance. Presumably they will also have to go to the back of whatever line has formed in the mean time.

  Given that this is probably irreversible for this JavaOne, please consider not clearing the room if there should be a lot of space in the following session. Perhaps if the reservations for the following session are <60% of the room capacity or some similar number.

  Posted by: desperado on May 08, 2006 at 05:25 AM

• ...... less than 60% of the room capacity.

  Posted by: desperado on May 08, 2006 at 11:18 AM

• Well, as I said, they swear they've got this sussed. So we'll have to wait and see how smoothly it does or doesn't work.

  
  One nice side effect will be to boot all the people who cluster around the previous speaker, preventing the next session from starting on time. That's always been an issue.

  
  But yeah, it might be a real drag, esp for Esplanade 310, at the end of the long hall. And since most of the big Web talks are there, I'll be watching to see how it works out during the conference.

  I expect, though, that it's going to work out better than you think - simply because the people executing this have more practice than you expect. But we'll see.

  Jim

  Posted by: driscoll on May 08, 2006 at 11:35 AM

• Hi Jim,

  Last JavaOne there were webcasts so people who couln't make it to the JavaOne could watch some parts online. What about this year JavaOne, will there be some webcasts again, and if so, we could we find them?

  Thanks

  Posted by: carmello on May 08, 2006 at 01:43 PM

• IIRC, the webcasts were (and still are) available, but I can't seem to find them on the SDN site. AFAIK, that's planned to happen again, for SDN members. I'll check.

  Posted by: driscoll on May 08, 2006 at 02:02 PM

• >Well, as I said, they swear they've got this sussed.
  >So we'll have to wait and see how smoothly it does or doesn't work.

  I'm sure they can make it work for their definition of "work" but I suspect their definition of "work" and mine are rather different.

  Your description makes getting in to each session sound like the lines at airport security. I do enough waiting in line for that already. If getting in to each session is an airport security type experience then I'm just going to turn up late for each session and take the risk of being turned away.

  Posted by: desperado on May 10, 2006 at 08:09 AM

• Well, my understanding is that they're just swiping cards - which if you remember last year, is what they did then. They'll swipe, and if you are "on the list" then you're in. Otherwise, they'll ask you to wait - presumably at the back of the wait line.

  Should work. And it'll be nice to know that I can get into the talk that I want - or that I should plan to go elsewhere. Incidently - the Intro to AJAX talk is 100% booked already.

  Posted by: driscoll on May 10, 2006 at 11:50 AM

This work is licensed under a Creative Commons License.