Dreamsongs Wiki

Diagnoses

• There is a perception that OOPSLA is dominated by some companies and a set of insiders who are closed to new ideas that don't meet their approval.

• Some believe that objects are passé or even have failed to live up to their exaggerated promise. Maybe that debate on "Objects Have Failed" was too close to home.

• Companies don't see much value in sending employees to such an expensive conference which provides little value-objects are old and there are plenty of journals and local experts.

-rpg

My first OOPSLA was 1993, Washington. I went because I was designing an OO system with Expert System integration and I was reading conflicting articles on OO design techniques. As you all know, this was in the middle of the "methodology wars". I continued going because I returned to work energized, with good ideas, partial solutions, and a feeling that things were in flux.

If one can generalize from my own experience, perhaps OOPSLA needs an injection of controversy. But not by saying objects have failed, since that is either too close to home (reuse never really happened) or patently wrong (Java is everywhere), depending on your point of view. What I mean is that perhaps to attract attendees any conference needs to address the trendy controversy of the day.

--gh

I see two different kinds of decline

• InsiderDecline -- people who were insiders, the names with the quality, publishing or giving tues at OOPSLA, doing work in the field, who move away. This can be negative (they give up and go into family therapy) or positive (they set up AOSDXPAgile Galaxy/NoFluffJustUs and go there instead). For these people, comming to OOPSLA must be about possibilities networking, about the technical programme, about workshops, about invitations to speak about their pet topics, etc.

• OutsiderDecline -- instead of 3000 punters we get only 1000. This can be the result of Insiders moving away, but I don't think it is - I doubt AOSD is pulling hundreds of punters from OOPSLA, nor the PLoPs (tens of academics, yes!). Part of this must be OOPSLA being a victim of its own success, Objects have won, and people now want major-brand objects rather than the generic OOPSLA experience.

To address InsiderDecline, we could set up a new little ''buzzy'' conference for 50-100 insiders. It's easy, and people here have already done it more than once.

How could we address OutsiderDecline? Do we want to? (if only to pay for the buzzy insider core at meets up at every OOPSLA). What would this look like? --JamesNoble

I believe OOPSLA has declined for two primary reasons:

• OOPSLA has morphed itself into a training conference: the tutorial program is bigger than the technical program; more money comes from tutorials than the rest of the registration; more attention goes to the tutorials; etc. Unfortunately, OOPSLA committees are not professional training conference organizers and thus being a training conference only worked when the demand for generic object oriented training was high and the supply was low. That time is past.
• the OOPSLA program committees have been overly dominated by a few people (although not by evil intent; let me explain). By trying to be a generalist conference, the program committee needs to have a very wide spectrum of expertise. In order to do this, there are only a couple people on the committee in any given speciality. Thus when a paper is evaluated, it is essentially evaluated by just a couple people rather than a larger number of people. This is not good for accepting innovation.

Bjorn, your second statement about a small number of reviewers in a given area being bad for accepting innovation is interesting, but a little mysterious. Is this a "Wisdom of Crowds" thing, where diversity and independence of opinion can be more accepting or smarter? Or is it that a small group of experts in an area are either too protective of their area or else prefer only papers that bolster their own research ideas? -rpg

On the diagnosis note. The OOPSLA success in the 90s must be correlated with the dot com bubble. OOPSLA peaked at about the same time as that bubble's peak, and declined with it too. This is not coincidental. Companies (thought they) had plenty of cash to send people to conferences, especially software conferences with lots of tutorials and with a sneak preview of tomorrow's hot software technologies. OOPSLA was square in that role.

So, the decline in numbers isn't necessarily a bad thing or a consequence of the uncoolness of objects. (But I would say that objects were hyped almost as much as the internet.)

The question is: given this, how can OOPSLA shrink gracefully? [Agreed. OOPSLA so far has shrunk ungracefully. -rpg]

CristaLopes

Many interesting things that happened in the past 10 years didn't come from pure academia. I always regarded academics to be behind in the area of (OO-) software development, because they lacked the context of real software creation (some exceptions prove the rule IMHO). However, I do not have yet an idea how to change OOPSLA regarding this. --PeterSommerlad

What is your point? Nothing you said was controversial or new. Do you think that OOPSLA is driven too much by academics? Do you think we should try to give academics more context? -RalphJohnson

Comdex has been cancelled for this year. Comdex is a technology trade show where many companies launched new prodicts. It was held in Las Vegas in November. The decline in attendance is interesting and seems to mirror what we've seen at OOPSLA.

| 1998 | 1999 | 2000 | 2001 | 2002 | 2003 |

This parallels the dotcom breakdown, and it is reasonable to argue that dotcom was based on Web Java. This implies perhaps that companies are not as interested in big-dollar, web Java-related conferences. -rpg

That last point of rgp seems important to me. We all know that OOPSLA is not a big-bucks conference, but maybe the people from industry etc. do not. We have to make sure that although OOPSLA is Java and sometimes web related, it is beyond the Internet bubble hype, and worth going to. --tla

Here are some possible reasons for decline:

• Dilution: The rise of AOSD, Middleware, etc conferences on the research side, and !JavaOne and MS conferences on the practitioner side and Agile/XP
• somewhere in the middle draw away possible attendees. Additionally, papers and presentations on OO topics are now common in all sorts of other technical conferences including PLDI, POPL, ICSE. Given all this, it's not so clear what OOPSLA is "about" anymore.
• Normal-science mode: By the usual paper review standards, the quality of accepted tech papers over the past five years has been much higher than in OOPSLA's first five years. But this is mostly because the papers tend to be solid but narrow incremental reports on now-long-established areas. So, not only are these less interesting, the conference is now less likely to draw people who are excitedly pursuing work in an area that they just barely understand and want to talk to others about. If you are a researcher without an accepted paper, this is just about the only reason you would want to attend a conference. There used to be a lot of such people.
• Normal-practice mode: Similar. Fewer people need basic exposure to OO stuff. And OOPSLA hasn't had a clear program for the average second+-time practitioner. Some commercial conferences that are doing well (JAOO is best example) largely feature invited one-hour presentations designed to teach the audience something useful. Other than tutorials, OOPSLA doesn't have such a thing. Adding such a track is the only suggestion I have at the moment that I think might actually do some good.

-DougLea

Doug's last point really struck a chord with me. Early on, I would come away from OOPSLA with the feeling that I had learned something, but much less so in recent years. And that decreases the incentive to attend. I think the JAOO presentation model would be a great idea.

--NeilHarrison

And, perhaps relevant to the discussion, JAOO attendance increased significantly last year from the previous year. Indeed, it was the largest JAOO conference since it began seven years before. It is a fun conference with a focus on short sessions, panels and long tutorials.

This year they will be experimenting with running a number of more focused symposia on more leading edge (wrt the mainstream) topics and a stronger track bias to the short sessions and panels.

--KevlinHenney

The explanation that seems best to me is "dilution", but that term doesn't really capture how it happened: FailedBySuccess.

results in many other small conferences and a dimished Oopsla. It's especially bad because Oospla always seemed to be the only real software design conference; not algorithms, not languages, not engineering, not process, but something more general. Agreeing on OO made for a helpful truce on an important issue, but the win was an emphasis on software design.

--RobertBiddle

(Maybe we should get pragmatic and consider WhyPeopleComeToOopsla.)

• ResearchDominatesTechnicalProgramme
• ConsultantsDominatePanels

I see two reasons for the decline as well.

1. Overall tutorial quality: When I first started attending OOPSLA 1
2. years ago, I was able to take some quality tutorials from the real movers and shakers in the industry. The three amigos, the Gang of Four, Craig Larman etc. It was exciting to meet these people, and hear what they had to say. A lot of these people are gone now, and the tutorials are given by the likes of me. We gave a great tutorial, but was it work $8
4. ? And who would spend that kind of money to see some unknowns speak.
5. Cost, cost, and cost. I originally attended as one of Linda Rising's minions. Our company sent so many people on year, we represented over 1% of the total attendance. But companies stopped footing the bill in the late 199
6. s, and I cannot afford to spend 2 - 4% of my yearly income to fully attend the conference. I go now because most of my expenses are paid for giving a tutorial, but even then, I don't pay for anything extra, and I don't take any tutorials. I'd like too, but they cost too much. Bottom line, is can we make it cheaper? Some suggestions:
   1. Choose cities with inexpensive flights, & cheap transportation.
   2. Choose convention sights with a good mix of nice hotels, and inexpensive, nice ones.
   3. Packages for individuals or companies allowing them to reduce costs.
   4. Free half day tutoral for every two half day sessions bought.
   5. Allow presenters to attend tutorials for free.

   --PaulBramble