What is a computer science conference, really?

Most computer science conferences expect that, except in extraordinary circumstances, at least one author will present the paper in person. The paper is also, in most cases, published in the proceedings, possibly on paper and distributed to some set of people, such as a special interest group.

The reason for this is that we highly value collaboration and face to face conferences give us a chance to sit and talk as well as interact with authors.

If you publish in a lot of conferences you need to plan for the travel. Grant funding is one source, as is university funding. I think that personally funding your travel is not especially common, but possible.

Note, also that many conferences attendees work in industry, not academia, where they do research of various kinds. It is common to see folks from IBM Research, or Microsoft Research, for example at many ACM conferences. They are funded by their companies.

One option for a prolific publisher (we should all be so lucky) is to do a lot of collaborative work so that if you can't attend, say, SIGPLAN POPL 2021, then one of your coauthors will be able to.

But no, it isn't just another way of saying "journal".

I should also note that quite a few people attend such conferences even when they have no part in the program. They are just there for the collaborative opportunities and for informal meetings.

I echo the other answers: Yes, the standard is that an author needs to attend the conference and physically present the work. If you don't have the time/money/energy/childcare/visa to do that, then hopefully one of your coauthors can go. In a pinch, a non-author may be able to present. However, ultimately, if no one is able to attend and present, then you have to withdraw the paper and publish elsewhere.

This situation is problematic and there have been plenty of complaints about it. I personally have had to miss conferences due to visa issues and due to childcare constraints.

However, I would like to add that this is rapidly changing. The COVID-19 pandemic has made conference attendance impossible for almost all participants. Many computer science conferences have decided to become "virtual" or at least have the option of virtual attendance. It remains to be seen how this will work. But there is a good chance that this will result in a lasting change beyond the pandemic.

Computer science conferences are confusing because they involve two entirely different things:

• A process that works essentially like a journal, where you submit papers, get reviews, and accepted papers are formally "published" in conference proceedings which is just like a journal. Differences with journals:
  • there is usually one submission deadline per year (aligned with the conference);
  • reviewing is speedy (around 2-3 months) because the conference gives a firm deadline;
  • papers are relatively short (around 12 pages) though in some fields this is worked around by having an unlimited-length appendix (which reviewers do not need to review);
  • the number of accepted papers is limited (by the conference capacity);
  • you can publish an "extended version" of a conference paper into an actual journal later;
  • sometimes, papers are selected based on whether they are expected to make a good talk, but this is only done by a minority of conferences and not the main deciding factor.
• A physical meeting, where people come, present their work, discuss and socialize: the program of the conference consists of what was accepted to the proceedings in that year. And yeah, there is a requirement that some author of the accepted paper will travel to present it.

To confuse matters further, computer science also has "informal" conferences, also called conferences: their program can be chosen by invitation or by a light reviewing process. At these informal conferences, you only do the second point (there are no formal proceedings) and presenting work there does not "count" as a publication in the bibliometric sense, so you may allowed to present work that you have published or will publish elsewhere. Some of these conferences will have "informal proceedings" to circulate the presented works but which does not "count" as a publication. An example are Dagstuhl seminars (invite-only) or Highlights of Logics, Games, and Automata (open, light reviewing).

I say that these two things are "entirely different" because the reviewing process on the scientific paper is usually only concerned with the paper and not concerned at all about the suitability of the work to be presented as a talk; and because the physical conference could exist without these proceedings (as exemplified by the informal conferences). This forced marriage is unsatisfactory because some people want to formally publish their work and do not care about presenting it at the conference, but need to show up no matter the cost, time, inconvenience, CO2 footprint, etc. Conference proceedings accomplish a valuable job (get formal publications of short papers with speedy refereeing and the option to publish an extended version later) but there's no good reason why they are tied to the physical meeting aspect.

There are some rare CS conferences which have started to de-couple the two aspects, e.g., the VLDB conference is attached to a separate journal PVLDB for its proceedings volume; and the POPL conference is attached to the PACMPL journal. One can only hope that the current COVID-19 crisis, which is forcing some changes in how conferences are run, will encourage further evolutions of this model.