Does anyone learn something new at conferences? Measurement of intuition and dissemination of knowledge at scientific conferences

The development of digital networks and conferencing platforms in recent years has led many to question the value of conferencing in terms of environmental protection and accessibility. However, one frequently cited example of their value is the possibility of random, unplanned interactions. Using evidence of conflicts in conference scheduling and subsequent publications, Misha Teplitskiy, Soya Park, Neil Thompson And David Karger consider such chance encounters important and measurable.

Academics and many other industries are investing heavily in conferencing. The costs are significant in both cases. financial and environmental terms. For example, one scientific conference may represent 7% of an individual’s annual income CO2 emissions. What are the benefits of conferencing and are they worth it? The evidence to answer this simple question was surprisingly limited and uneven.

One area where evidence is accumulating is the networking function of conferences. For example, in an experiment at a medical school Boudreau and colleagues found that placing two scientists in the same room for a networking event—a common occurrence at conferences—increased the likelihood that they would co-sponsor a grant proposal by 75%. Repeating this result, Campos and colleagues found that when a major political conference is unexpectedly cancelled, the likelihood that a potential participant co-wrote an article with another potential participant is reduced by 16%.

So do conferences lead to distribution beyond alternative access, and if so, how?

What is less clear is the diffusional function of conferences—how much do attendees actually learn? This issue has become even more relevant in recent years with the emergence of alternative distribution channels and conference formats. Articles that were once only available after publication through personal connections or as conference presentations are now often available as publicly available preprints. Quality and popularity of virtual and hybrid conference formats also increasedwith a concomitant expansion availability. These large-scale changes are redefining what constitutes the choice of attending conferences – it is increasingly not a choice between having and not having access to information, but a choice between accessing it online, that is. reading a preprint or watching a video stream rather than synchronous face-to-face communication. So do conferences lead to distribution beyond alternative access, and if so, how?

We can divide distribution into two types: “directed distribution” when visitors see presentations they are interested in, and “random distribution” when visitors accidentally encounter presentations they did not intend to see. The history of science offers numerous cases from intuition, but how important is intuition compared to “normal” directional diffusion? Is random spread common enough for people to take it into account when deciding whether to attend a conference?

Our last preprint offers extraordinarily compelling evidence that conference attendees do indeed learn a lot from presentations and that much of the spread happens by sheer luck, that is. the participants did not plan to explore the ideas. We use data from conference scheduling software Conferwhich allows participants Like negotiate and create a personal schedule. The data includes the personal schedules of 2,404 participants in 25 computer science conferences held between 2013 and 2020. We link these graphs to the participants’ future publications and, as a learning proxy, measure whether the participants have cited submitted papers.

Our recent preprint offers extraordinarily compelling evidence that conference attendees do learn a lot from presentations and that much of the spread happens by accident.

Of course, it is very likely that the conference participants Like and attend presentations of interest to them and subsequently quote them more than did not like those. Because Like articles are not random, we do not compare citations Liked against. did not like documentation. Instead, we only target a set of documents for a particular user. Liked and exploit the fact that sometimes the conference schedules some of these paper presentations at the same time, which on average reduces a person’s ability to see the presentations. If viewing presentations is important, we expect the person to quote Liked There are more papers without scheduling conflicts than those with conflicting time slots. If viewing presentations isn’t that important, scheduling conflicts shouldn’t affect citation. In addition, we take advantage of the fact that users often Like only one or two presentations in a particular session, but once they get there, they will also probably see other presentations in the same session. We call it in-one-session-but-not-Liked paper “serendipity”. If serendipity is important to dissemination, we expect participants to cite articles about serendipity more frequently in sessions they may attend.

Using this “scheduling conflict” study design, we find that the conference propagation function is essential. Compared to talks presented at time slots that people couldn’t attend due to scheduling conflicts, they cited talks in presentations without conflicts about 50% more often. About 22% of the diffusion caused by presentations is due to articles that did not receive likes “by intuition”. To the best of our knowledge, this is the first time that the contribution of intuition to diffusion has been quantified. The evaluation shows that even when serendipity is defined in this limited sense, it represents the majority of what participants can expect to learn at a conference. Accounting for all channels of insights, such as when participants learn unexpected ideas from hallway conversations, can further increase the perceived contribution to dissemination.

Our paper uses a new study design based on scheduling conflicts to answer the old question of hosting conferences Job to spread ideas? At least at the computer science conferences we study, the answer seems to be yes, and much of this proliferation is a fluke. This work also raises clear questions for future research. Can I get the same benefits of in-person conferencing with other conferencing formats? Do other formats provide the same (or even more) intuition? And, in the long run, are the benefits of in-person conference worth the cost, especially given affordability?

The content created on this blog is for informational purposes only. This article represents the views and opinions of the authors, but does not reflect the views and opinions of the Impact of Social Science blog (blog) or the London School of Economics and Political Science. Please see our comment policy if you have any concerns about posting a comment below.

Image credit: adapted from Michael Christenson via Unsplash.com.