Executive Summary: Let the market rule. PC members get "dollars" to "buy" submissions. Every paper that has enough "dollars" on its head gets in.
Implementation: Give everyone on the PC N weighted minutes, which correspond to
the minutes that he would attend paper sessions. Then have them spendthese minutes on the submissions that they want to see. Every paper with,
say, at least MIN minutes worth of votes gets in. Period. And the authorgets that many minutes to present the paper. We should probably also impose an upper limit on a paper (30 minutes?).
Experience:
I proposed this idea to Scheme 2003 and they used it (though with discrete quantas). The attendance went from the 35 average to 79 (I know because I paid for the lunches).I also used the idea for NEPLS, a PL workshop for the northeast region. We told authors to request a certain slot size and then allocated minutes depending on how much we liked it. Often we granted what they requested, sometimes more sometimes less. Attendance was again close to the max of what NEPLS has seen in 10 or so instances. Surprisingly, people seemed to like the N x 10 min "idea" sessions a lot. -- Matthias
I've been reading a little about the Iowa, which are used to predict (set the price for) elections. They use real money and they buy and sell contracts over a period of time, which is a real market. The idea of spending minutes to vote for papers seems like a first-order approximation to a real market, where the "real value" of the submissions might be determined over a period of time based on a number of trades. Any thoughts on using a market approach rather than a retail one? -rpg
Your question doesn't make sense to me, so I respond to the paragraph.
I believe that you're confusing two markets here: the market of ideasand the market of papers. Publishing papers is one way of contributing
to the market of ideas, but by no means the only one. I bet that youagree with this considering what you say/write elsewhere. There is, however,
a market of papers and the consumption of papers. The latter happens intwo ways: listening at the conference because you think you might be inspired
by what the speaker has to say and you might read his stuff later on; and readingthe paper later because you have an idea for a paper that should be compared
to someone else or that re-uses someone else's idea. The former is common,the latter is rare in our world.
My proposal is about fixing the first tier of the publication market. (1) I
believe that ACM's policy of turning conferences into the major vehicle forpublication has killed (or is in the danger of killing) the academic discipline
of computing. (2) I also believe that conference have become boring becauseof this policy. So I am proposing to ignore it, to destroy conferences as they
exist, and to replace them with lively meetings of people who wish to exchangeideas.
-- Mathias
My question has to do with the process of selecting papers. I gathered you are proposing allocating PC members some number of minutes, and they then purchase papers with them. The total for each paper is used to select papers. I was asking whether it would be better to set up something like the mechanism IEM, instead. This mechanism would perhaps allow a few rounds of iteration so that the "true value" of papers to the PC can be converged on. So, same basic idea, but different mechanism. -rpg
In that case, I am with you. Several rounds of "buying" is clearly what you want.
I am not insisting on the exact mechanism, but I really like the idea of openingup and just "buying ideas" at a conference. -- Matthias
First, as Mathias says, it narrows the scope of the conferences. Since academics are the ones who produce the bulk of the papers, and their papers are usually optimized towards small increments, that is what we get. Most ACM (and other CS) conferences suffer from this. Then it enables an emerging behavior from the CS community, which consists of the following: I work in X, which has attracted some attention within an existing conference; me and my buddies decide to spin out a conference on X; we make the technical papers component have only a 15% acceptance rate; bang!- our conference is automatically considered 'good', and well-accepted academically.
I can imagine a very different conference on programming where lots of people would have their work exposed, independent on whether their papers are academically solid or not, as long as there is some interesting idea in them. But, being an academic myself (at least for now), I can also see that a conference like that would possibly not attract academics, and that would be bad in the long term.
So a possible solution for this would be to have a journal associated with the conference, published once or twice a year. Articles for that journal would be drawn from the pool of papers submitted to the conference. The PC would have two major responsibilities: 1) to choose, say, 50% or more of the submissions to the conference, using one of the market-based methods suggested above; 2) to choose, say, 20% of those and to review them thouroughly for publication in the journal.
Crista, I share your concern of course that some academic (those wedded to
"highly selective, refereed conferences" and bragging about them) will stayaway. My two experiments show, however, that the vast majority of people,
including academics, seem to enjoy an open forum where people just have achance to present ideas.
Naturally, our long-term goal (speaking as an academic) has to be the
destruction of the conference system altogether and its reconstructionas a place to meet and exchange ideas.
There are no refereed conferences and publishing them in unrefereed
archival journals doesn't change it. Let's build a real discipline.-- Matthias
Crista's idea is similar to what we do with !PLoP. Each !PLoP is a set of writers' workshops. There is a mimeographed set of proceedings which are used at the workshops (I use this old term to emphasize that this is not a publication). Then after the workshops (which serve to improve the material), some of the papers are put through a review and referee process, and then some of the papers (patterns and pattern languages) are published in a book.
I think, then, that Crista's idea supports a real discipline: conferences are for exchanging ideas, including incremental ones. Some of those ideas, once reviewed and shepherded through a journal-like process, can appear in journals. Is this not right? -rpg
I agree with one thing implcit here: to move passive outsiders to insiders, people have to be able to contribute to the conference, where contribute means more than paying USD1000 to sit at the feet of Ivan Grady Ralph Dick Christa...
In effect, the main OOPSLA tech proceedings are a journal. Just an odd one that takes on 30 papers and publishes once a year. SIGPLAN and SIGGRAPH officialy make their proceedings a journal, OOPSLA as an archival conference has the same status. But: there are ways to invite more people to (relatively) less prestigious publications at OOPSLA, without undermining the JournalNature of the main tech proceedings (with I think is WhatsNotBroken i.e. WorthKeeping).
For example, CristaLopes's additional 30% of papers could appear in a SalonDesRefuses session.Gush. I attend paper sessions only if I've seen a particular paper that catches my eye. Mostly I do it so as to be prepared to chat with the author afterward without having to admit I didn't go to her session.
If what we're saying here is even partly true, it's the very ACM / academic "quality" philosophy that is damaging the conference. It's a puzzlement. -RonJeffries
But let's be clear: this is a side effect of OOPSLA having succeeded: we have established a discipline - perhaps one that functions differently to Mathematics or Physics (where conferences are unreferred and journals appear weekly) but similar to biochemistry - another resaonably recent evolving discipline.
I go to more ECOOP technical sessions more (when I'm there) but that's simply because there is less competition at ECOOPTechnical sessions will ever only appeal to a small, specialised group of people: if we want other kinds of people to come along, we should first look at improving the other parts of the conference.
Matthias has a writeup of a concrete proposal: BuyPapersWithMinutes. -rpg