MiniConf: A Virtual Conference in a Box
mini-conf.github.io
mini-conf.github.io
I'm a professor, not a software engineer, so happy for any comments or contributors!
The mini-conf tour is a good overview for generalized application. Your ICLR pages are great examples of mini-conf in action, e.g.
https://iclr.cc/virtual_2020/workshops_8.html https://iclr.cc/virtual_2020/papers.html?filter=keywords
A few production questions.
1. Sync/playback is relatively fast. Are you using a streaming server or web-hosted video files?
2. On pages with video & slides, how are the two frames synchronized, e.g. are those two synchronized video streams with known timecodes for slide changes, or is the slide frame showing static images that are synced with video timecodes?
3. Did you chose Python because of existing code that you wanted to reuse?
4. Which existing software was closest to meeting your needs, and what were the shortcomings that motivated miniconf?
Answers:
1 + 2. We used https://slideslive.com/ an external conference provider. They do a lot of optimizing behind the scenes, and have that cool sync.
3. Python is the de facto standard language in ML. We use it for everything. We thought about js/react or something, but we didn't think it would be maintainable.
4. A lot of the existing software assumed there would be a physical event or even worse tried to pretend there was a virtual physical event. We wanted something that was async-first, desktop-oriented, and very simple to browse.
The result is a static website where the participants meet at time, but all interactions are handled by the third-party providers.
The main features are ease of configuration (CSV files) and templatability (with Flask).
Is my understanding correct?
I was wondering if it offered similar features to EasyChair, like participant registration, review process, notifications etc.
As you mentioned there are a lot of good solutions for review, notification, and reg. Common choices are OpenReview/CMT/HotCRP/EasyChair/softconf... They're all pretty mature and complex.
The thing that people suddenly need is ways to easily customize their own pages for hosting virtual conferences and workshops. That's the problem miniconf targets.
But your simple solution for virtual events is very much welcome and appreciated! Many thanks!
What researchers want is discoverability and credit. Some researchers are already famous and own their own distribution via arxiv or Twitter but for new researchers publishing a paper at a prestigious conference is a way to bootstrap a reputation.
If you really wanna get rid of large publishers what you really need to solve is discoverability and you can solve it with a combination of virtual mini conferences and callouts on Twitter and Hacker News.
I've gotten some of the best constructive feedback on my blog posts from Hacker News directly, I don't see how that's different from peer review. I don't find it very likely that someone would even be willing to give me feedback unless they themselves are already subject matter experts.
But really the impact of a piece of computer science research is directly correlated to how many people directly use the OSS inspired by the research. People in industry know this, what's left is is for tenure committees to also take software impact into account.
Software should be citable. You can make it so by publishing a paper describing the software. Usually the authors then request anybody who uses the software in their research to cite that paper.
(No reason for doing it, I was just curious if I could.)
A lot of people think they're subject-matter experts and may comment on your article, but may be mistaken.
In all the peer review processes I've been involved in reviewers review each others reviews before they go out. Other reviewers will challenge a bad or misleading review. If there aren't people who know enough about the subject they'll go out and find someone who is.
You could literally do this process in a web interface that looked like Hacker News, but they already pretty much do.
I think if you built your dream review system... I think you may find it looks pretty much like what we have already.
> callouts on Twitter and Hacker News
Sounds like a popularity contest.
It’s kind of like arXiv with voting and comments, but also supports collaboration for research in progress.
> You could literally do this process in a web interface that looked like Hacker News, but they already pretty much do.
I would love to hear your thoughts about how the current systems might be improved and if we’re on the right track.
Virtual conferences are terrific. I just helped host the AirMiners conference [1] on carbon removal on May 13. Because it was a virtual conf we were able to achieve top speakers, international attendees, and just seemed so much more effective. (KSR's keynote is here: https://youtu.be/9Pw0n0CeK0k)
It's going to be an uphill battle to get me to attend an in person conference again.
As someone who’s currently in industry rather than academia, I interpreted “virtual conference” as “a Twitch or YouTube stream, and likely some sort of chat space like a Discord”. That landing page does a great job of explaining the technical underpinnings of your tool, but doesn’t actually explain what sort of site/services the web server actually serves!
Even just including the GIF from the beginning of your Twitter thread would be super helpful.