This is my brain dump of SIGMOD 2020 conference experience.
First of all, I really like virtual conference. I have been to conference once in the past. One big lesson learned from that experience is that there is no way to attend all sessions simply due to it’s not physical possible. Sessions are running concurrently at the same time. It’s cumbersome to navigate through the venue and get around the crowd to reach the session you like in time. However, with Zoom, the magic happens. I can open up all sessions I’m interested in and mute the speaker via drop audio setting in Zoom. If I find the topic I want to hear more, I can instantly switch to the desired Zoom window, reset the audio setting, and listen to the talk. Analogously, the experience feels like watching some streaming marathon on Twitch or watching the International from your laptop. Also, virtual means no visa headache 🙂 Another good thing about having sessions in Zoom is that I can easily ask questions whether via directly chiming in (Thanks Boon Thau Loo for promoting me to the panel to ask question live) or through typing. Asking questions in person offline can be challenge but being able to type questions online creates a relax environment for me to interact with speakers.
Another big advantage of virtual conference is the cost. Thanks to the COVID, this year’s SIGMOD is free. All I need is to sign up and then I can get into the internal system to attend any sessions I’m interested in. My perception on the academic conference is that everyone gathers at some fancy resort, enjoy the social interaction for a week, and then fly back home. I would imagine how costly this can be given the flight expense, hotel room, and registration fee. The zero cost conference means outreach; means reaching out wide audience. I would really love the organizer publish some stats on how many people actually virtually attend conference. I’ll be in huge surprise if we don’t see a huge number jump there. In addition, the free cost feels like welfare for me: I don’t have to pay a few hundred dollars to get myself motivated for the PhD journey ahead of me. I can see an accessible conference like SIGMOD this year will be a huge morale booster to someone who is struggling in their PhD and will motivate people to do good work.
Another big win for me is the recording part meaning that each presenter records their talk before hand and the session chair simply plays the videos one by one. In some session, I do experience some technical issue like there is video with no audio. But the issue is fixed within 5 minutes and the small distraction doesn’t impact the whole session experience at all. Pre-recording means high talk quality. The speaker can give his best performance for his talking. I think many of the speakers probably record their talks several times to pick the one they think can best deliver their idea in their work. Another great part is that the presenter is actually standing by to take questions and can give replies in the Zoom chat and even in the Slack channel several hours after the talk. This feature is very nice because we can keep the discussion in asynchronously fashion; both question and answer can be written out for further digest. Talking about Slack, I see PC chairs in SIGMOD and PODS are quite busy: you can see them across almost all Slack channels. Constantly saving questions and comments from Zoom to Slack to spark more discussion; spread Zoom link and session context. I think they deserve some kudos.
Being virtual means there will be a lot of writing communication: whether it is through Zoom chat or through Slack. This is huge benefit. I assume important information can easily get lost in the offline conversation. For example, it’s really too early for me to get up at 7am Friday to attend New Researchers Symposium given an 8 hours work ahead of me. However, thanks to being virtual, lots of discussion actually happen both during the Zoom live and more importantly, on Slack. People use Slack to ask questions and some panelist is nice enough to write their answer on Slack thread as well. This is good for me because now, during the lunch break, I can scan through the Slack channel and get some information from the past discussion. I figure this would not be possible if the conference is offline.
I’m not sure how the conference is run in the past but I think this year’s organizer puts huge effort to organize all-in-one page with Zoom link, Slack link, and schedule in one page so that I can easily find the information I want.
As a first time academic conference attendee, matching actual people with their name in paper is a huge win for me. In some way, this does feel like handshake events for Japanese idol. Indeed, those “Japanese idol” are in fact quite approachable. I really do enjoy Anastasia Ailamaki’s smiley face from the camera and her persistent typing to answer the questions both from Zoom chat and Slack channel. There are several sessions I really like from this perspective are SIGMOD Plenary Panel: “The Next 5 Years: What Opportunities Should the Database Community Seize to Maximize its Impact?”, Mohan’s Retirement Party, and Industry Panel: Startups Founded by Database Researchers. So many names that I saw both from paper and from internal codebase. It’s very cool to see some roast and teasing online.
Social wise, I really like Zoomside Chat series. For example, “Zoomside Chat with Jian Pei”. It feels like coffee break and the topic is very relaxing. This is the place where some “ungraceful” question get asked. Also, “Zoomside Chat with Tamer Özsu” is also fun. I really wish there would be more time allocated for this type of social events.
Research wise, taking a look of the accepted papers beforehand is really helpful. On my laptop, I have a list of papers written on the Notes. I write down my thoughts and comments for each corresponding paper on the list. Due to my personal interest, Wednesday and Thursday sessions interest me the most. Luckily, my targeted papers spread out quite evenly through two days. My biggest regret is to not go through the papers I’m interested in beforehand. The result of not doing so is to get lost in sessions that may seem tangible to my research direction. This is worth improving for the next time to make most out of conference. Having said that, I’m still able to learn some useful benchmarks that I can run related to my research. Also, sitting through the talks (even lost) help me to further refine my paper list for future reference. Another thing I notice is that workshops are much nicer for learning. Research sessions usually have only 10 minutes for each speaker. Speaker has to move very fast and cover part of material on paper. However, workshop speaker has 25 minutes (at least for aiDM workshop) and the pace is much slower compared to normal research sessions. Lastly, attending sessions is a great way to discover the knowledge gap: even they may not relate to my research direction, it is still fun to learn for pleasure.
Some downside of this year’s virtual conference is Gather. I don’t know how useful it is for others but it is not quite useful for a working professional like me. First of all, my company “bans” the website: I can get into the room but it will take forever to load the venue floor plan and see other people. If I really want to use Gather, I have to disconnect from company’s VPN. I want to walk around the venue while waiting for the build. However, VPN-unfriendly Gather is not quite helpful here. Another disadvantage for virtual means I don’t have to participate “fully”: I can run errands; check work emails; fix some code bugs for the work. I don’t have to give out full energy to the conference. I guess this is really my bad.
Overall, I’m very grateful for this virtual experience at SIGMOD this year. The overall experience is excellent. I’m hoping they will do something similar next time; maybe partial virtual? However, I surely will miss the chance to see people and attend sessions. That motivates me the most to do good work because I want to attend next time (maybe as a presenter).
Received an email from organizer
The last workshop has finished, and SIGMOD/PODS 2020 is now history. We suspect it will be a landmark in most of your minds, separating SIGMOD/PODS Conferences into those pre-2020 and those post-2020. Even before all the adjustments brought on by the COVID-19 crisis, we planned to stream more of the sessions. Our registration of ~3000 shows that there is high demand for online access to the conference. If our community is serious about fostering diversity and inclusion, then remote participation should become a permanent option.
Looking forward to the remote participation in the future.
This section collects some useful comments I gathered from Slack. It is for my future reference; might be useful to you as well.
From SIGMOD Plenary Panel: “The Next 5 Years: What Opportunities Should the Database Community Seize to Maximize its Impact?” on whether researcher should be “customer obsession” and solve real problem:
I would never discourage work that is detached from current industrial use; I think it’s not constructive to suggest that you need customers to start down a line of thinking. Sounds like a broadside against pure curiosity-driven research, and I LOVE the idea of pure curiosity-driven research. In fact, for really promising young thinkers, this seems like THE BEST reason to go into research rather than industry or startups. The best reward is the joy of the idea.
But I think I share some of Anhai’s concern about improving the odds of impact and depth in our community.
What I tend to find often leads to less-than-inspiring work is variant n+1 on a hot topic for large n. What Stonebraker calls “polishing a round ball”. The narrow gauge for creativity in a busy area makes it really hard to find either inspiring insights or significant impact on practice; but at the same time the threshold for publication is often low because social factors in reviewing favor hot topics (competition, familiarity, and yes — commercial relevance of the topic, which can lead to boring research too!) That’s something we can try to address constructively.
Now I am guilty of going deep and narrow sometimes myself, e.g. in distributed transactions and consistency in the last many years. But it’s been rewarding and fun, and I like to think we had a new lens on things that let us hit paydirt a few times. Certainly outside my group, work like Natacha Crooks’ beautiful paper on client-centric isolation in PODC 17 demonstrates there is still room for major breakthroughs there. So some topics do merit depth and continued chipping away for gold.
Bottom line, my primary advice to folks is to do research that inspires you.
To align with @AnHai Doan a bit more, if you are searching for relevance, you don’t need to have a friend who is an executive at a corporation. Find 30-40 professionals on LinkedIn who might use software like you’re considering, and interview them to find out how they spend their time. Don’t ask them “do you think my idea is cool” (because they’ll almost always say yes to be nice). Ask them what they do all day, what bugs them. I learned this from Jeff Heer and Sean Kandel, who did this prior to our Wrangler research, that eventually led to Trifacta. It’s a very repeatable model that simply requires different legwork than we usually do in our community. http://vis.stanford.edu/papers/enterprise-analysis-interviews
As for where to find the customers, I really like your suggestion. Another thing I may add is that one can go talk to the domain scientists in the SAME university. Many of them now have tons of data and are struggling to process them. These domain scientists are often sitting just ten minutes from one’s office, and they are dying for any help. Talk with them to really understand the kind of data problems they have. Often at the start those are very mundane basic problems, such as querying a big amount of data. But if one can help them solve those basic problems, then often many more interesting problems come up.
Yet another thing to do is go talk to companies in the SAME TOWN. They often are downed in data too and would love to get some help. One can very quickly get to know the kinds of problems they have. This has worked at least for me. My group started out working on data problems with several domain science groups at UW-Madison. We developed solutions that were used by them and in turn they gave us feedback. Then we took those solutions to local companies (insurance, health, heating/cooling, three companies), and they helped us improve the solution. Then we got funding to do a startup. This is perhaps also a possible roadmap.
New Researchers Symposium with a question asking about how to fix “the despicable but common toxicity of the database community in the tone of their reviews and often during in-person questions/discussions?” and ask about how to write a good review.
Joe Hellerstein7 hours ago
I sense there are stories here, and I’m sorry to hear this. In my experience, the face-to-face interactions in our community have gotten more professional over the years. I’m very sorry if the questioner has witnessed bad public behavior.
But in my experience the reviews have actually gotten worse in recent years. I believe we need a process change. The root of the problem is that reviewers are generally not held accountable for what they write.
One thing I learned in my startup is that team members who voice concerns are not very valuable — startups are risky by nature, and concern-mongering just contributes to negativity. However, team members who voice concerns and propose solutions are gold. We should ask the same of reviewers.
Anecdote: Doug Terry often signs his reviews. I went through an exercise last CIDR where I decided to try that, and I found instantly that I became much more helpful and constructive in my reviewing, even for papers that I did not recommend for acceptance. It didn’t feel OK just to criticize; I felt more responsible to suggest and encourage changes.
I’ve heard arguments why this isn’t a reasonable global solution—e.g. junior researchers could face retribution for signing negative reviews. But we might consider other mechanisms for ensuring that the reviewers are (a) held to account and (b) required to be constructive.