What are Benefits of Physical Conferences?

There is a lot of discussion now about online conferences.   This of course is motivated in the short term by Covid-19, but I think there is an opportunity here to make a lasting change in the “conference culture” of NLP and AI. Below are some personal thoughts and preferences.

Drawbacks of Physical Conferences

When I was a young post-doc who was working full-time on research and had no family commitments, I could attend 5 conferences a year if I wanted to.  But now, I am supposed to only spend 20% of my time on research (the rest is on university teaching and my work at Arria).  Hence spending a week travelling to and attending a major international conference amounts to 10% of my “research days” for the year.   Also if I am away during a weekend, then I cannot help look after my autistic son (who is normally in a residential school during the week and home during the weekend), which makes life stressful for my wife unless we make expensive alternative arrangements for him.

So the upshot is that I can physically go to one or (at most) two major international conferences each year.

A lot of other people also find it difficult to go to conferences.  I know researchers who find it difficult to travel because of health issues, disabilities, caring responsibilities, problems getting visas, and lack of funding.  These people have good ideas, and the community as well as these individuals would benefit if it was easier for them to make scientific contributions.

Jet-setting around the world also of course has real costs to the environment and global warming, and indeed may make it easier for new diseases to spread.

Benefits (Value-Added) of Physical Conferences

Meeting people: Conferences give me a chance to meet and catch up with colleagues.  Its really nice and useful to find out what people are working on and interested in, get their feedback and suggestions on what I am doing, and discuss possible collaborations.  Its also a great way to meet new people, who often have new ideas and perspectives.

Of course I can also meet and interact with people via email and social media, and some interaction is possible in online conferences.   But I think there is a lot of “value-added” in meeting people at physical conferences, at least for me.

Learning about new developments: In all honesty, I dont learn much from most oral presentations, especially in a conference context where I’m expected to listen to a huge number of presentations in a short time.   I usually find that I’d be better off reading the paper than attending the talk (and then I can also focus on papers I am interested in).  Poster sessions are better, because I can focus on what I’m interested in and chat to the author; sometimes I’ve had really good discussions at poster sessions, especially if other people who are interested in the paper also join the discussion.    Discussion sessions can also be good, especially at focused events such as workshops.   Keynote talks can also be interesting and valuable, but since there is limited interaction, I could just watch these on Youtube, being physically at the conference doesnt add much.

So with regard to learning about new developments, I think poster sessions at physical conferences work well and “add value,” as do discussion events in small focused events.   But otherwise, I’d learn just as much by reading papers and watching videos.

Presenting my own research: I personally prefer to present results in journal papers instead of conferences.   However, I do find it useful to present ideas, work-in-progress. initial results, etc at workshops or (small) conferences; it is a great way to get feedback and suggestions.

I think I get better feedback *from participants* at physical workshops.  However, I get feedback from a *wider range of people* at online events.  So its probably good to have both kinds of events.

My Preferences

I guess what I would like to see is a mixture of events and venues in the future.

• Proper results published in journals.  This is what the rest of the scientific community does, and we should do likewise!   Journal papers should have associated discussion boards (again common practice elsewhere).
• A smaller number of large physical conferences, which focus on posters and discussions, and more generally encouraging participants to interact with each other.  Again, this is what many other disciplines do.
• Specialised online events, more like workshops than conferences, where participants present and discuss ideas, work-in-progress, etc.  Perhaps discussion is done using written comments (like blogs) instead of oral discussion; this is something we’ll need to experiment with.

Above probably wont work for everyone, but I think it would work for me.