An organizers’ report on FPW’26. The goal is to preserve participant feedback and organizers’ impressions and data about the event (attendance, budget, etc.) to help other organizers that might consider running similar events in the future.
A quick summary of key points:
FPW’26 was a series of functional-programming workshops organizers together in Paris in August 2026, between Monday 24th and Friday 28th. The following events took place:
HOPE and ML also had a physical presence in Indianapolis, colocated with ICFP 2026.
See the main FPW’26 webpage for more details.
FPW was run by a handful of volunteers (see Acknowledgments below) and the general approach was to do things simple and cheap.
Participiation was free and thus lunch not included. We had a list of recommended restaurants and a long break, so participants formed into groups organically to go have lunch together. From a small budget of 2k€ we paid for coffee breaks and snacks for the participants.
We believe that it is possible to organize satisfying scientific events, at little cost and with a reasonable amount of work, and hope to encourage other people to run more of these small and simple events in the future.
We ran a survey of FPW attendees, 33 people answered who had all attended in-person.
1 person (3%) had a “bad” experience, 13 people (40%) had a “good” experience and 18 people (56%) had a “very good” experience.
We asked: “If FPW’26 had not been organized, would you have attended ICFP in Indianapolis in-person?”. The 33 respondents answered and none of them (0%) would have attended ICFP in-person.
We are very glad to confirm that we did not, in fact, take anything away from ICFP organizers and participants by running FPW, but merely offered an additional experience to other people.
out of 23 responses, 21 people (91%) had a “good” experience during the watch party, 1 person (4%) “very good”, 1 person (4%) “bad”.
We asked whether they would have watched the same amount of ICFP talks online (or less, or more) if they had not attended our Watch Party: out of 26 responses, 3 (12%) would have watched the same amount of talks, 12 (46%) would have watched fewer talks, and 11 (42%) would have watched no talks at all.
This confirms my personal impression that watch parties are more motivating than watching talks remotely by oneself, and that they increase the visibility of conference presentations.
We asked whether the lack on on-site lunch and the cheap coffee made the experience worse. For 22 people (71%) it was “just as good”, for 4 people (13%) it was “better!”, for 5 people (16%) it was “worse”.
Participants gave various feedback in their free-form answers, in particular an encouragement to organize more in advance and communicate the event more broadly – we agree.
Finally, we asked for testimonials, and people said for example:
I’m very grateful to the organizers: experimenting new formats is always challenging, and FPW turend out to be a great success in my eyes. I had a great time!
I had a lot of fun attending FPW! It was also an interesting experiment to see in what ways we could organize our conferences in other ways that are better aligned with the needs of people and our planet.
FPW was amazing: it didn’t cost anything, there were lots of friendly and talkative people around. A lot of the presentations were nice as well. There was also enough time in breaks to catch up with the people around.
FPW’26 was a very valuable experience, with an excellent scientific program and a great opportunity to meet and exchange with researchers working in functional programming. I particularly appreciated the welcoming atmosphere and the quality of the workshops and talks.
… and finally a snarky one:
The experience was well-worth the $0 price tag.
We had a lot of fun participating to FPW! Our general impression was that people enjoyed the event, and we were impressed by the scientific quality of the talks.
What went well:
People had fun! Talks were interesting!
Lunch outside the venue went well. We had planned at least 2h30mn for the lunch breaks. This was in fact too conservative, and a 2h break would have been fine.
We organized a last-minute conference dinner with 20 people in attendance and it also was very nice. (Note: if you organize an event, do think about a conference dinner in advance.)
The video-conference platform we used for remote talks (BigBlueButton, an instance hosted by the French government) was just fine. We had a couple AV issues but they came from on-site configuration issues, not the online software.
The Watch Party was a nice experience. We were not sure people would actually be willing to attend one full afternoon of remote talks, as it is typically less engaging than in-person talks and the pull of the mailbox is strong. People did in fact listen to the talks!
We quickly realized that just listening to the talks and QA would be a bit too passive: to stay engaged it is better to give attendants the opportunity to be active. (In theory one could ask questions to ICFP speakers remotely, but in fact questions from online tend to be ignored, as in previous years. For the ML workshop specifically Sam Westrick did an effort to forward questions and it went well.)
So we organized the following: for each paper to be presented during the session, we nominated one person in attendance to get on stage and answer question. We would mute the QA session from Indianapolis, and have our own locally. This was fun! For keynote talks we kept the QA, as the answers were typically more personal, and also there would be a coffee break right after for interactions anyway.
None of the small issues we had during the event felt like they would have been easily solved with just more money. This supports our belief that “registration is free” is a reasonable constraint to set ourselves when organizing scientific events, and we suspect that it could lead to better, more diverse and more sustainable organization of academic group events.
What could be improved:
The online-attendance experience for the OCaml workshop was not great, especially in the afternoon. Our impression is that, to have a good remote experience, the deciding factor is to have one person in the room who is in charge of monitoring the online platform, ensuring that any technical issue reported by the online audience is fixed promptly. There was a lapse in OCaml in the afternoon; some of us were taking care of HOPE which had a more complicated remote setup (due to taking place also in Indianapolis). Oh well.
Our impression is that the online experience for other workshops was fine. The most delicate aspect is to include remote viewers in QA sessions: make sure that they can follow questions from the in-person audience, and ask their own questions. To support remote questions it suffices (again) to watch the online platform and forward questions to the speaker; easy (if you have someone in charge of this). Sharing in-room questions with the remote audience is harder: people will not have a microphone, the speaker will not think of repeating the question, so we just tried to repeat the questions in the online chat in writing, and this was probably a bit jarring to online attendees at times (for example when the speaker starts answering before they can read the question).
(Note: our experience is that current SIGPLAN conferences are not doing a better job with QA sessions and online participants. We remember the experience being better a few years ago, when there was a stronger push for hybrid conferences (before people got sick of it during COVID). During a few conferences everyone, in the room or outside, was invited to ask questions in writing in the online chat, and that worked much better to include online attendees.)
We were a bit disappointed by the relatively low number of online attendees. With an event that is clearly advertized as supporting online participation (for free), one could have naively expected to see more online attendees than in-person attendees, or at least similar numbers. In fact the number of online attendees was always much smaller, with 10 as the peak number of online attendees for the whole week.
There could be a chicken-and-egg problem: if the remote experience is typically not good, people will not consider remote participation. It could also be the basic issue that sitting on your computers watching talks from elsewhere is a bore, and distractions abound. Maybe it is unrealistic to expect most people to attend more than two or three talks online in a row (we can all see that hybrid seminars tend to work better than hybrid conferences).
The seniority profile of the academic attendees was a bit surprising: very few non-local senior researchers attended. We had plenty of PhD students and post-docs, even some undergrads. There was a strong participation of industry people; they were drawn probably by the OCaml workshop, but also attended some other events (… and were also a substantial fraction of the no-show).
We are not sure why non-local senior researchers were not interested in attending. It may be that the date conflict with ICFP prevented from joining (being on the same week was unfortunate in some respects, but it was a request of HOPE and ML to take place in both places on the same day). Or maybe senior researchers attach more importance to the colocation with a “prestigious” event like ICFP? Or maybe the fact that FPW was a one-shot event with unusual motivations was discomforting. Oh well. One theory: juniors are attending their first few academic conferences, so it may be harder for them to “just wait for next year” and they are more in demand of local alternatives.
We should have set things down in writing more in advance, to help people plan to attend, especially those that need to wait a few months to get a European visa. We started discussing FPW in Fall 2025, settled on a location in January 2026, and should have communicated more widely at that point instead of communication widely only in May-June.
In general, we have come to appreciate the value of the “rules to organize conference Foo” passed from year to year in established conferences, with strict dates on when to do this or that. If you don’t have them, you are going to do things too late. (For future organizers: take an established conference, look at how many months in advance they do [X, Y, Z], and imitate it.)
Thanks are of course due as well to the speakers (in-person and remote), whose scientific content powered this event, and to the audience who spontaneously created a vibrant and welcoming crowd.
(Note: the list of people providing assistance on-site is relatively small. Within our self-imposed low-profile organization, it was possible to run a satisfying conference with only two or three people responsible for each event on-site.)
Note that these attendance counts are noticeably lower than the registration counts we had, probably due to the fact that attendance was free. We had: - 96 people register for the OCaml workshop (so only 75% actually attended), - 37 people register for HOPE (but only 14 who registered for HOPE and not for OCaml, which happened on the same day), - 53 people register for TyDe (56% attended) - 28 people register for PLMW (50% attended) - 44 people register for the ICFP watch party (68% attended) - 53 people register for the ML workshop (60% attended)
In retrospect we could have estimated that 66% of registrants would attend, and that would have been a good approximation. (Some people showed up without attending, but this clearly did not compensate no-shows.)
In terms of geographic provenance (when we had the information): - 48 attendees were locals from the Paris area - 15 came from other parts of France - 14 came from the UK - 7 came from Germany (with a strong Effekt presence) - 7 came from the Netherlands - 2 from each of: Belgium, Canada, Denmark, India, Switzerland
Attendance to FPW, in-person or online, was completely free.
Our costs were small:
venue: we booked the main amphitheater of the INRIA research center in Paris, which can host up to 100 people, for free
lunch: we did not provide on-premise lunch, people spread in small groups to various restaurants around the venue and paid with their own money.
coffee breaks: we spent 2K€ for coffee breaks (bad coffee, meh tea and okay croissants)
So our total costs were 2K€, funded by the research budget of a few academic colleagues in Paris, with support from the OCaml Software Foundation. (In typical French-administrative fashion, the event only happened last month so we do not really know yet who will actually pay.)
FPW’26 was not a bi-localized satellite of ICFP’26, but it gave us a glimpse of what a multi-localized conference could feel like, and we liked it a lot! The attendance was around 30 people every day, with a spike at 100 on the first day. This number is high enough for meaningful, diverse social interactions, while remaining low enough that we could host everyone for free.
We believe that having “secondary sites” for larger academic conferences could be helpful for many people, for example:
Our colleagues in the US who are currently worried about taking the plane to another country and coming back.
Our colleagues in South America who cannot afford conferences organized in expensive places.
Our colleagues in Asia and Australia who attend international conference in smaller numbers due to the cost, fatigue, and environmental impact of long-distance travel.
30 people showed up on three consecutive afternoons to watch live ICFP talks together. In future years other people could organize this in Santiago, Philadelphia, Chennai, Kyoto and Melbourne; especially if there is moral support from SIGPLAN.