The world’s leading publication for data science, AI, and ML professionals.

Data Science Coding Meets Esports

Building community through memes and live competition

Community Spotlight

In the Community Spotlight series, TDS Editors chat with members of the data science community about the exciting initiatives that help push the field forward. Today, we’re thrilled to share Elliot Gunn‘s conversation with Nick Wan and Meg Risdal, the creators of SLICED: an exciting new live coding competition that’s "like the TV Show Chopped but for data science".

_Nick Wan is co-host and co-creator of SLICED. He is the manager of Data Science at KFC and streams SLICED and other data science content at Twitch. You can follow him on Twitch and on Twitter._

Meg Risdal is co-host and co-creator of SLICED. She leads product at Kaggle (a Google company). She lives in Los Angeles. You can find her on Twitter and LinkedIn.


If I had to summarize SLICED in three words, it would be: playful, welcoming, and professional. I think this vibe comes across through SLICED’s aesthetic, production values, and hosting style. Could you speak more about how this project originated and evolved?

Nick: I was doing these things called data blitzes, where I’d give myself 30 minutes to get through a dataset I haven’t seen before while also making a presentation and then the chat would grade me on my presentation. It was pretty wacky and silly but people would really get into it. I always wanted a version of it where people would create data science work competitively like in the data blitzes. I think of how esports breathed new life into video games and thought maybe there’s an opportunity to breathe a competitive fire into the existing idea of a hackathon. So after Meg and I talked about SLICED for a while, we got four people from my community to participate in a 4-episode pilot season and it was a big hit.

Meg: Nick and I started talking about early concepts for SLICED during a roundtable chat on his stream. We made a joke about Chopped but for data science (we’re both into cooking!) and the idea kind of stuck with us. And we thought, why not just try it? So we did! And it ended up being a great partnership as we developed the idea. I brought a lot of my experience working for Kaggle (which is of course also all about competitive data science and community) as well as my network for building hype, recruiting contestants, and bringing in an audience and initial sponsors.

And I’m glad to hear that the SLICED aesthetics are eye-catching! We put a lot of thought into the branding brief for SLICED when working with our designer, Veronica Casson of Salt & Fog. She did an excellent job of helping us convey something we hope is approachable, welcoming, and a lot of fun. For the most part we’ve been pretty scrappy, doing a lot of the video production and social media marketing ourselves, so big kudos to us, too. Recently, though, we brought on Tony Pelleriti to create YouTube cuts for our episodes that have really taken things to another level. We’ve gotten a lot of great feedback on the production and want to keep investing in this.

Do you see SLICED as an initiative to innovate in the data science online ed space? How would you situate something like SLICED alongside existing content?

Nick: There tends to be three layers of educational media: educational content (like how-tos or lectures), edutainment (like Bill Nye or Planet Earth), and general entertainment (like BattleBots or MythBusters). My stream has been labeled edutainment. I don’t personally see this as directly educational because we aren’t really trying to instruct or teach anything, even though it may come off that way at times. We’re not simplifying any concepts on SLICED (or on any of my streams); it really isn’t for beginners. Take when NASA launches a rocket. Are they trying to educate people about astrophysics? No. Not that SLICED is at the level of sending people to fucking space, but this is similar. We’re showing off the fruits of labor that a lot of people put into learning data science. There’s not enough of that in data science. A lot of data science streamers are aimed at educating. Like David Robinson, Julia Silge, and Greg Matthews who live-streamed his courses on Twitch during the pandemic. There are these educators and I’m not trying to walk in their steps at all.

Meg: Like Nick said, SLICED is meant to be fun and entertaining for our audience and the broader data science community. It’s not meant to be directly educational in its own right, but I hope it can inspire people to find data science fun which might motivate them to want to learn something new. It’s been especially cool to see SLICED encourage people to participate alongside our contestants and try their hand at the weekly challenges, experiment with techniques they see on the show, write blog posts (like Julia Silge’s post about implementing a custom RMSLE metric in tidymodels), etc. But at the end of the day, it’s the community building and "doing" (versus learning) aspect that excites me.

Do you see SLICED as a way of bridging that gap by allowing viewers to watch experts in action? Does SLICED have a meta-goal of showing students to learn how to learn or ask good questions?

Meg: SLICED is not a very realistic setting for doing production-grade, professional data science (thank goodness). For new people to data science, SLICED is more an opportunity to see how professional data scientists aren’t perfect, don’t have every API memorized, etc. Our contestants know very well by now that it can be a very humbling experience! And it’s really, really meaningful as someone new to data science to be able to observe experts making mistakes and iterating. SLICED definitely provides a unique opportunity to witness this. But I wouldn’t say someone should expect to learn how to do data science from watching the show. For people who do want to learn, I’d encourage them to participate on Kaggle since there’s no better way to learn than by doing.

Nick: People have said this to us because it gives a window into a legit industry professional’s process. We don’t have any goals for SLICED to be educational, this is just fun and a way to grow the data science community and culture in a new medium. Community is far more important to us than education. SLICED is like watching people do trickshot basketball or pool. You don’t do that stuff in an actual game. SLICED is trickshot data science. It’s also introducing more people to Kaggle in a really unique way. That’s really exciting. I think that’s good because it’s using existing tools to create something new in data science. We already had Kaggle and we already had Twitch. Use both and turn it into a game show and people who have not been privy to the Competitions section of Kaggle, now they’re exposed to things like other Competitions on Kaggle. Eventually we’d love to see more people do live competitions on Kaggle. There’s more trailblazing that has to happen. Interesting and creative people could be inspired by seeing these dimensions to these platforms and really do something exciting with it.

The show does a walkthrough of each contestant’s notebook a day after the competition is over. Can you share more about the value in watching experts talk through their process?

Meg: During the live show, Nick and I can only speculate about why someone makes decisions or approaches a problem the way they do. Only once you hear them walk through their approach do you understand their full thought process. It’s been very cool to see how our contestants pay close attention to each other’s approaches, too, between episodes. There’s a lot of conversation happening not just in the day-after live streams but also on Twitter using the #SLICED hashtag, in Nick’s Discord, in blog posts, and in contestants’ own streams. Many of the contestants like Jesse Mostipak and David Robinson have shared their mistakes and how they would approach things differently given more time and hindsight. There’s a lot of learning going on outside of the show itself which is yet another thing that makes SLICED’s budding community so great!

Nick: Last season we did this during the episode. Now that there’s four people instead of two, there’s not enough time to get deep into everything during the episode. There’s people who watch SLICED who want these deep dives. And contestants are more than willing to walk through their code. So it’s a good, happy marriage that’s an easy thing to do. From the contestant point of view, even they have better ideas about how they want to convey something the day after instead of right in the moment. So it’s both a time/pragmatic thing, but it’s also that it’s better and more succinct after they have time to relax.

Do you have a specific target audience for the show? Do you see any differences in how beginners or more advanced coders can draw value from watching it?

Nick: Our specific target audience is people who know about data science. Whether it’s someone who just started or it’s someone who’s in the industry for years. If they understand data science, we hope they enjoy it. Our audience is very driven by me and Meg and our contestants’ communities. I’d say half of the people watching SLICED are coming because they know one person who’s coding. 30–35% may be coming because they saw someone they respect promoting the show on Twitter (use #SLICED!). The remaining 15% or so are probably the current, existing SLICED regulars. A lot of these people are not necessarily regulars on my other streams. And that’s cool!

Meg: Dads and moms are also welcome in chat (data scientists or not)! But seriously, I don’t want SLICED to become so super competitive that it becomes an unwelcome niche. It’s obviously still competitive at its core and we want to be true to that reality, but we want it to feel fun and exciting to everyone who’s familiar with data science. As a competition (or a sport, really), we want to make it really rewarding for people to follow the action week-to-week and get bought into the progression of the show. But at the same time, we want it to be entertaining and fun for people who can only drop into a show here and there.

Has anything about running SLICED surprised you?

Nick: The community. It was surprising in the pilot season. Before the pilot premiere, my highest viewer count was 80 something. So breaking that record was exciting. It’s surprising every week when more than 50 people show up. Stats say we’re averaging around 100 people a night, with a few points where we’d have 200 folks in the stream. Data science could be really dry content-wise. It’s not always visually exciting and it’s not usually about the personalities behind the data science content. SLICED gives opportunities for more visual excitement and to highlight data science communicators. I think that’s some of the reason why people keep showing up every week.

Meg: The drama every week is something that surprises me! It’s genuine and different with each and every episode. The leaderboard shake-ups, the tense technical difficulties, the razor-close scores, contestants not fully reading the documentation (always read the documentation!). It all plays out in a way we could never script if we wanted to. SLICED is real reality TV!

It’s also really blown my mind how many people participate along on Kaggle during the live show. Nick and I initially guessed we’d get maybe 2 or 3 people which was WAY OFF. Really, we just intended for Kaggle to be used as our scoring infrastructure, but the platform has become a core part of the SLICED community. Every week we’re seeing more people join in. We’re even witnessing people share on Twitter that they’re playing along between episodes for fun. It’s exciting that people really resonate with this extra dimension to the show because it makes the experience and community of SLICED that much richer.

You can find all of the datasets for SLICED season 1 on our website and in this Twitter thread.

Where do you see SLICED in season 2? Do you have a vision for what comes next?

Meg: My hope is that SLICED season 1 generates enough hype to really build a lot of awareness and anticipation as we prepare for season 2. We’d love to recruit more people, both from the existing SLICED participants, but new people who’ve just learned about the show. Overall, we want to approach season 2 the same way we approached season 1 from the pilot: we learned a lot about what worked well, what didn’t work, and made improvements. I want to take the same approach once season 1 wraps to evolve the show because it’s really all about what creates the best experience for our audience, community, and contestants.

I’ve also half-joked about doing SLICED spinoffs and Christmas specials when the show jumps the shark. But seriously, I’d love to do one-off episodes focused on more niche domains of data science like NLP or computer vision. That would allow contestants to participate with a lower time commitment, let us show off different areas of expertise, and let us concept-test experimental ideas to incubate for full seasons. Feel free to message me and Nick on Twitter or in Discord if you have ideas! I also honestly want to do a cooking episode of SLICED. Literally Nick and I will do Chopped in person in LA.

Nick: For SLICED season 2, we haven’t started to talk about it. We just need to get through the first season and go back to the whiteboard to figure out next steps. On my channel, we’re hopefully doing a virtual conference after the end of SLICED season, so that’s most immediate on my to-do’s. For next season, we hope to push the envelope more in terms of production. It’s been fun to see what we’ve already been able to do and people really respond well to it. We definitely haven’t reached the edge of what’s possible. We’re just getting started here. We’ve talked about ideas for how we might evolve the show. Like in-person SLICED or SLICED at a conference. What if SLICED could become like a wildfire, like TEDx out of TED talks. Where the concept becomes its own. I saw on Twitter some folks thinking of starting a competitive data cleaning contest, SLICED-style. That is the kind of stuff I love seeing – SLICED becoming more than just our show but the start of many different styles of data science content.

How can interested readers get involved?

Meg: I’d also love to see more people compete alongside our contestants on Kaggle! It’s really exciting to see the passion from our community. Also please share your #SLICED experience on Twitter! We love to see how the show inspires you and what you learn and create.

Nick: Check out our Notion site where there’s lots of information about how to become a contestant, sign-up as an intern, or sponsor us. But honestly the best way to get involved is to watch us on Twitch!


Curious to learn more about SLICED? Here are two Towards Data Science posts that share the authors’ experiences with the competition.


Related Articles