We’ve entered the second half of December this week, which is unofficially the slowest of the business year for tech (on the product/engineering side anyways). Everyone’s realizing they haven’t spent their annual vacation days, and are using up all the days they can’t roll over. Other people are taking time off because of children being out of school at the end of the year. Normally, many people would’ve been traveling to visit family over the holidays but it’s 2020 and I hope people refrain.
All told, this makes for a very slow time of year. The few people who aren’t on vacation face a lot of quiet time because coworkers are away. So things are generally fairly relaxed once we hit this point, both at work, and at home. Phew.
That’s why, 4 years ago, my much younger, pre-parenting self decided it’d be a good idea to organize a small conference for game developers around this quiet time. I hadn’t thought I’d still be doing this in 2020, but here we are.
Since I’ve been busy doing conference organizing stuff the past few weeks, it’s been dominating my out-of-work brain cycles, so y’all get to go on this ride with me.
Visual;Conference, aka VNConf, what is it?
The conference itself is super niche and quirky, but has managed to survive the test of time. January 23rd, 2021 event will be the 5th conference (O_ o). Ostensibly, it’s a game developer’s conference, where people interested in making games come together and give talks about developer-relevant topics.
The unique spin is that it’s targeted towards the tiny, indie, visual novel game developer segment. You’ve might have stumbled across VNs while browsing Steam over the past couple of years, where they stand out as having a decided “anime” aesthetic, may or may not contain mature risque/adult content, and generally feature less gameplay elements and more reading.
Out of all digital entertainment genres, VN has probably the lowest barriers to entry. Since it’s a book with additional pictures, music, and possibly voice acting, anyone who can write a narrative story and draw some static character art can put together a game. Good ones largely rely on the strength of the writing and artwork. There’s no need for big budget 3D models, and little in the way of programming (amusingly one of the major VN game engines is Ren’py, written in Python).
This low cost barrier leads to lots of small, often young teams, even lone developers, experimenting and playing with the medium. There’s a huge variance in game style, quality, and market success. The space is mostly full of amateurs who may contract out work from professional artists, programmers, musicians. Since it’s a fledgling non-industry with very few full-timers, there’s also very little industry community. There’s very little cross-pollination of ideas like we’d see in tech. Obviously, no one has ever thought to have a conference about the topic either.
Everyone is also scattered around the globe, and no one has money to fly to a physical conference. So from the start, the whole conference needed to be online. With all these constraints, I got bored of waiting for someone else to put something together. One (probably drunken) night, I talked to some community members about the idea, saw some interest, and blindly plowed ahead. I did work for almost 6 years at Meetup doing data stuff, so organizing groups online wasn’t entirely foreign to me anyways.
The conference itself (vnconf.com) has a very silly name: Visual;Conference. A completely nonsensical semi-colon shoved between two arbitrary words — a nod to the amusing non-English found in some Japanese game titles that the genre owes it’s roots in.
Incidentally, all the talks are posted freely after the conference, so you can see the archives on YouTube.
Since the conference is essentially run by me alone, no one else knows the chaos and oftentimes pure arbitrariness that goes on behind the scenes. It seems fun to share that information a bit.
Conference Format
The conference is very much like the tech conferences I occasionally attended. It’s single-track of speakers that have been selected by the organizer (me). Talks are either 20 or 5 minutes long, with 5 minutes of Q&A in between that doubles as setup time. All of the tech stuff was handled on GoToWebinar, a somewhat clunky, but “full featured enough for our purposes” webinar service.
Next, we needed to recreate the more interesting interactive component. I opted to avoid the built in chat system (which is horrid) and use Discord instead. Speakers and attendees are all mixed in together to interact and comment while people speak, as well as ask questions during Q&A. It’s a giant free-for-all, but is probably the most fun part of the experience. Everyone is commenting, asking and answering questions live. I feel that it’s that sense of “doing things together” that’s the most important part of the whole conference. Otherwise we could just record talks that no one watches on YouTube.
The whole thing has been slowly getting longer over the years, but currently runs for a little over 7 hours. There’s a lunch break and other short breaks during the thing, but otherwise we take advantage of the online format to let people eat and take breaks as they need.
Tickets
After years of being a Meetup organizer, I’m firmly in the camp of putting a very small bit of friction to gatekeep events. I’d rather have people excited and engaged in the audience than a very high attendee count. Other organizers argue that events should be free if possible, for valid reasons, but I prefer the paid model.
So we charge money for attendee tickets — $15 a person to attend, $10 if you buy a ticket early, w/ a bulk ticket discount. It’s deliberately designed to be almost trivial to attend, but enough to make sure people are motivated.
The money helps pay for operating expenses like server space and webinar software (usually around $150 for a month… unless I forget to cancel the subscription…). Any extra left over money we have goes towards fun things like commissioning logos from pro artist, recording equipment, mics, etc..
All told, the event pays for itself, but isn’t a significant source of income in any way.
Getting speakers is full of surprises
The first year, getting speakers was the biggest worry. Will anyone even come speak? Without speakers, nothing else can happen. It’s an especially big concern for a brand new conference with no brand name to begin with.
Luckily, I’ve been around the gaming community long enough that I have a modicum of credibility. At the least, it’s not likely I’d be staking my public reputation to scam speakers somehow. So I put out a public call for speakers, using just a Google Form. Then I tweeted constantly about it as well as posting on various forums where VN devs hung out. That was it.
Despite my general incompetence in marketing and promotion, it worked. Word spread, people got excited about the idea. About 16 people submitted 1-paragraph talk proposals. NUTS.
To balance out the potential schedule, I also went and personally invited a couple of “big names” to see if they were interested and, to my surprise, a few actually agreed. I even presented a talk myself just so there’d be at least 1 face on the web site’s lineup early on.
I honestly don’t know if this level of positive response is typical or not for such activities. I think it speaks to how much pent up demand existed for a conference like this. No one else was crazy enough to do it. And for those keeping score at home, I think there’s still room and pent up demand for lots of niche data-based conferences too, so it’s worth considering if you have an idea.
In subsequent years, I had to personally invite fewer and fewer guests because we had enough talk submissions to pick a good talk lineup from that it wasn’t necessary. I sorta feel bad for having to reject a good talk for scheduling reasons because I had invited someone “famous” to speak already, so I don’t do it much.
Probably the biggest surprise out of all this was that people started telling me they were planning out their talk ideas before I even announced the call for speakers. People also started asking me “Hey, when are you posting the call for speakers form?”.
There’s. So. Much. Data. Entry.
My conference lives in, and dies by, spreadsheets. Almost everything is tracked and organized on Google Sheets. You’d think that an experienced data scientist and engineer would have built out some kind of website and scripts that would do a lot of the grunt work. Nope. Almost everything is manual. There’s so much data entry and management, I feel like I’m in a pre-internet office job.
There’s a big spreadsheet with the submissions, which gets augmented with talk ratings, who was selected, have I sent out their acceptance email yet, did they respond w/ their info, etc. It just spawns into a monster as event prep rolls around.
How manual does it get, you ask? I actually manually copy/paste the acceptance emails from a template in notepad. I also manually transfer the profile/talk info they give me to the web site. It’s all too squishy and free-form to build too much ingestion scripts for, and I don’t want to spend time building a giant intake web site that would collect the same messy text data, but just put it in a database instead of auto-populating it into a spreadsheet for me.
There’s a lot of risk in doing all this manual work. I’ve caught myself on the edge of accepting the wrong talk before, or worrying I might’ve accidentally over-booked the speaker lineup (which you can’t really take back if you do). You have to be very careful and check work multiple times. I wish there was a better way to do everything, but I honestly don’t see any scaling solution that doesn’t involve a ton of dev work.
Picking talks is hard
Perhaps the weirdest feeling of running a conference is picking talks and speakers is having to choose talks. Who the heck am I to judge if a talk proposal is good enough? Essentially, the conference quality lives and dies by my internal sense of quality and ability to visualize what a talk could be like and how it fits w/ everything else. It’s fertile ground for imposter syndrome to haunt you.
I decide on talks mostly blind. I just read the proposal and judge how much I want to hear that talk as an audience member. If I’m excited to hear something, it’s likely to get in. You’d think that picking interesting talks out of single paragraphs submitted by random people would be hard, but it’s surprisingly easy.
Here’s the reason for that — the barrier to entry for giving a talk is very high, since you’re potentially standing in front of an audience to do it. There’s very few trolls and outright dumb proposals because it’s hard enough to have the guts to step forward and propose a talk. On top of that, you can very often get a preview of how articulate and clear-thinking a speaker is just from a single paragraph. Long story short, well-written proposals mean a ton and are a strong predictor of future talk quality.
Proposal quality is also how I take risks on talk descriptions I don't fully understand but sound competent. Since I can't imagine what those topics will be like, it's a risk to put them into a lineup while building a flow of talks. But at the same time, the world is much bigger than my ability to imagine, so it’s necessary to risk accepting a potentially bad talk to hear something unexpected. There’s always one or two of these included in every lineup I’ve put together.
After filtering down to “talks I think are good”, I have to then balance the entire conference program. This is the tricky bit and takes a few days of debate on my part. It involves breaking things down into topics, making sure we’re not overloaded on a single topic (too many programming talks, no business topics, etc). I often go back and forth quite a bit on this point, tentatively accepting talks, finding the balance to feel weird, then shuffling the acceptances.
Since we’re all sitting together to listen to talks for 7 hours straight, we need to keep things fresh, so I think this is the most critical part of designing a good lineup. It’s just so disgustingly squishy.
Other considerations
The one thing I very luckily don’t have to worry about is making sure the speaker pool is not just a bunch of men. Gaming, similar to tech, has a strong bias to being male-dominated, but despite the inherent bias the speaker ratios have hovered around 50/50 most years. There was one year where there were more women speaking then men.
I don’t take any credit for things winding up this way. I'm very bad at identifying and personally inviting people to speak, so manually manipulating the speaker pool is an extremely difficult task for me. On top of that, people submit talks with random internet handles and pen names all the time, so it’s never clear who’s actually doing the speaking, even if I looked closely at all the identifying information of the submitters.
Thankfully, I don’t have to intervene because the genre itself has attracted an extremely diverse array of creators. It’s not all just a bunch of dudes in dark basements We’ve had voice actresses, artists, programmers, writers, full-stackers, all speak because they had cool stuff to say.. At most, my insistence that there be a balance of talks covering topics outside of mere programming, including writing, art and design, acting, business and marketing, made more room for everyone to participate.
Scheduling is a paaaaaain
Time zones
Oh look, Randy’s complaining about time again.
As I mentioned before, the audience and speakers of this event is global, and that makes scheduling an interesting challenge. Normally for a physical talk, since everyone is in the same physical space, you can just arrange talks however you feel like. Organizers can put similar talks together into a block, which encourages cross-pollination of ideas and questions.
With a global speaker pool, block scheduling becomes much more difficult. We run the event from 11am to 6pm (Eastern Standard Time). That chunk of time was deliberately chosen to be a balance for a global audience. It’s not too early for people on the US West Coast, it’s not too crazy late for people from Central/Eastern Europe, and the beginning is an awkward, but not horrifically late hour for East Asia. That covers most of the major areas we get speakers from.
Still, to make things bearable, I have to schedule speakers according to where they are on the planet. Speakers from Asia go first since the conference starts around midnight for them, then anyone from Europe since it’d be getting late in the evening. Finally people in the Americas can be organized more by topic.
One year, I invited two industry speakers from Japan, and despite them being the first speakers of the whole event, it was 1AM already local time. They gave their talks, went to bed, and had to view recordings of the rest of the event afterwards.
The first year, I realized all of this somewhat late, and had to have everyone give me their time zone via email — hello ridiculous data issues like “Central time” vs “UTC+8”, etc.
Now, I ask for timezone info front in the proposal submission form, which still allows for some data issues due to the box input, but I give an example of “UTC-4” so people often follow it.
Also, Everytimezone is a lifesaver when manually dealing with time zone scheduling situations.
Breaks!
Despite being online, with no camera requirement, we build in a couple of breaks into the schedule. Sure, people can run off to the bathroom or grab a snack whenever they need to, but it still takes concentration to pay attention to talks. Breaks were a welcome place where people could relax for a bit to refresh before another chunk of talks.
We usually have a 1 hour lunch break early on, then two short 10 minute breaks later in the day. We use the break periods as a free-talk period where all the speakers can just turn on their mic and talk about whatever they want, often responding to the Discord chat.
The break periods are also strategically placed later in the day because everyone gets progressively more tired and attentions start to wear out quicker. So the schedule looks something like this: 2hr talks - 1hr lunch - 2hr talks - 10min break - 1hr talks -10min break - 1hr talks - END.
Organizing is a game of predicting the future, and fixing then it
Organizing is a complicated act of juggling that’s very hard to describe, but I think the above heading is the best I can put into words. You need to have a very good imagination about what could go wrong, and then work to prevent that disaster. Everything done is pretty much dedicated to solving those future problems.
Tech checks to head off tech issues. Hosting “presenter practice sessions” so people know how to use the software ahead of time. Keeping a site updated so people have a source of truth to refer to. Collecting time zone information. The list is endless, and you can always do more.
I guess this is why organizing usually falls to more experienced people who have seen things go wrong before. But if you have a refined sense of paranoid negativity like me, you can probably do it too! *coughcough*
Guess what, we should have more miniature data conferences
Waaaaay back in Feb/Mar of 2020, before pandemic quarantine threw a wrench into everything, I had hosted Drinks and Data Rants, an ultra informal speaking event that was more a “get together and rant about data topics”, I even threw together a tiny little site at http://datarant.link/ in case I wanted to do more of it…
Then pandemic hit and I just couldn’t muster the will to drag even more people into more video conferencing… but y’know. if someone wants to help out and organize a few, I’d gladly help them do so!
Between newsletter writing, a rampaging toddler tornado in the house, and 2021 probably looking like 2020 for the forseeable future, I’m a too overbooked to run more events right now.
But we should. Because data folk need to talk to data folk more, for more sustained periods of time instead of brief tweets.
So if anyone’s interested in helping out organizing, ping me w/ a message and I’ll gladly help.
About this newsletter
I’m Randy Au, currently a quantitative UX researcher, former data analyst, and general-purpose data and tech nerd. The Counting Stuff newsletter is a weekly data/tech blog about the less-than-sexy aspects about data science, UX research and tech. With occasional excursions into other fun topics.
Comments and questions are always welcome, they often give me inspiration for new posts. Tweet me. Always feel free to share these free newsletter posts with others.
All photos/drawings used are taken/created by Randy unless otherwise noted.