GMTK Game Jam 2021 Post Mortem (Patreon)

Hello everyone!

So, this month I hosted the fifth annual GMTK Game Jam (an event where people try to make a game in just 48 hours). And I wanted to share some thoughts, some lessons, and some interesting stats with you all.

Last year's jam was, quite frankly, enormous. 2020 received way more entrants and submissions than 2019 - more than double, in fact. It was a special year, though, with almost every country in some form of lockdown - meaning lots of bored people with nothing better to do.

And so I was unsure if 2021 would surpass it. But… it did! When the event began, over 21,000 people had joined up - another sizeable increase from 2020 (when 18,000 people entered).

Every year, only about 30% of those entrants will submit a game (accounting for people in teams, and those who don't finish). 2021 was a little down - with a 27% follow through rate. But that did mean another record breaking number of submissions: 5,883 when the jam finished (now 5,813 - games get removed for various reasons).

So that's really awesome numbers. We're still bigger than Ludum Dare (but still far behind Global Game Jam). GMTK Game Jam is still the biggest jam on itch.io and the biggest online-only jam. Which is incredible.

If you order Itch's jam list by "Most Submissions", GMTK gets all three top spots.

We'll have to see what happens in 2022 - the UK, at least, was still in lockdown during this jam but (fingers crossed) the world will be back to some semblance of normality next year.

Anyway. In true post mortem style… what went right and what went wrong this year?

What went right: better rules

In last year's post mortem, I talked about the lack of "clear, thought-through, and easily enforceable rules". So that was a key goal for 2021.

To that end, I created a website with all the jam's rules on it. It covered pretty much everything that might come up, and so the mods on Discord could just point people at that page if they asked questions about eligibility, timing, and so on.

If we were surprised by an unexpected question, the mods could get together and decide on the right decision - and then I'd add the answer to the website. That way, the mods weren't giving bespoke and conflicting advice to different people - everyone would have the exact same rules.

Trying to come up with rules is hard, as it turns out. Every single part of the jam is a decision I must make, and it's a decision that will benefit some, and disadvantage others. Make some people happy, and others frustrated.

To help with this, I came up with some internal values that the game jam follows, so I could ensure that all rules were working towards the same goal. Those values include things like fairness, a desire for solo devs to be able to compete, and not wanting to put too much of a burden on the jammers or the moderators.

That was helpful, and I'll keep thinking about them when making future decisions.

What went right: quick results

We've now run enough jams to fill an entire horizontal segment on the GMTK homepage! 

In 2019, the community's results and rankings went out as soon as the voting period ended. This was exciting for jammers, but it kinda took the wind out of the sails of my results video and led to confusion about who the "real winners" of the jam were.

In 2020, I held the results back until I made my video. This was better - a cleaner, more exciting and surprising end to the jam. But it also meant that jammers had to wait a long time to see how they did.

So for 2021, I kept the same system as last year but endeavoured to get the video out much faster. In the end, it came out 49 hours after the voting period ended - I was basically doing my own little game jam!

An alternative, unused thumbnail for the results video, this time showing the game Rift Shift. 

To help with this, I prepared for the video in advance by writing the intro and outro, and making templates for things like the game titles. And then I worked with itch to get early access to the ratings over the weekend - this gave me a heads up look at what might be in the top 100 and I could get started on playing the games.

And so when voting ended on Sunday night, I only needed to grab a few games that weren't in the top 100 in my early look. And then spent Monday and Tuesday frantically playing games, writing the script, and making the video. It was a lot of work, but worth it to get the results out in a timely manner.

What went right: team finder

A few weeks before the jam started, a bunch of Patrons in the GMTK Discord got together to make a tool. It would be a service to help people find likeminded game devs to collaborate with on the game jam.

In the end, it became the (now offline) Team Finder website: a sort of community pin board where you could post a notice about your team - and then others can find your notice and message you on Discord. 

You could filter by specific skills (like 3D art or programming), see time zones and languages, and more.

Team Finder in action

It was a truly excellent website that served its purpose beautifully. I got to see some of the development process and it was a sight to behold: very smart people building a site at rapid speed, conducting UX tests, responding to the inevitable bugs, and so on.

And it was effective at its job! We had over 2000 unique logins, and over 500 people made teams and then deleted them - suggesting that they had found the people to make their game with. (A further 800 weren't closed before the website was shut).

We also saw teams in other languages like Spanish, French, German, Russian, and Chinese - making it a very multicultural app!

I really liked this, and hope we can bring it back in some form for 2022

What went wrong: late submissions

Every year, the goodwill and fun of the jam is spoiled, somewhat, in the 24 hours after the jam ends.

Like... we give jammers 48 hours to submit, we added an extra 20 minutes to account for itch.io's annual server wobbles, and then there was a generous half hour grace period after that for emergencies.

And yet, despite this - and despite trying to give as much information as possible about how and when to submit your game - people still missed the jam and started arguing with me in my email inbox! I'd have to deal with very angry people who felt entitled to be let in.

To be honest, the stress of it this year was almost too much to bear. I had a small panic attack on Monday and had to just escape my computer for a while. It was only three or four people involved, but it felt like more than that. It was horrible, to be honest.

So I'm not 100% sure what to do next year. One idea is to have the jam be 50 hours, and dedicate the last two to submissions and bug fixes - with absolutely no grace period after that. If you miss it, you miss it.

There would be a honour system that you stop working on the game itself at hour 48 - I'm sure some people will keep going but they run the risk of not properly submitting or testing their game. So, that's on them.

I've got a year to think about it…

What went wrong: limiting theme

This year's theme was "Joined Together". I wanted to see games where multiple things were connected in some way.

I made the theme logo in Unity, so I could have proper physics on the chains!

Now we saw some really amazing games this year - with entries about relationships, telephone switchboards, ropes, chains, and so on. As always, people really surprised me with the way they interpreted the theme.

But I have to admit that some of the feedback on this year's theme was a little negative: some found it hard to think up something original, and felt it limited their options. There was a small, unofficial survey in the Discord where people were asked to pick their favourite theme from all the GMTK Game Jams… and 2021's came in dead last!

Not sure how to fix that - I'll just try again next year!

Some more stats

150,077 ratings were given out during the voting period - and every game received at least one vote. The average number of ratings per game was 25.8 and the median was 19.

The most popular game of the jam was Brain Reunion, with over 500 ratings - but that didn't help its chances on the rankings as it came in at position 782 on the overall ranks.

The most commented on game was Maze of the Mini-Taur, with almost 300 comments. This one fared a lot better, at position 6 - but that's not too surprising as at least three of the team members on the game were previous GMTK Game Jam winners!

120 game names contained the word "Joined", and 261 contained "Together". 34 contained "Joined Together" in the title, and 10 were simply called "Joined Together". Other common words in the game titles were: Connect (133), Chain (122), and Slime (74). 

I asked the developers if they made their own art for the game, or used pre-existing assets. 85% of developers made their own artwork. I also asked jammers if they made their own music. 56% of them did, which is pretty surprising!

One of the most interesting bits of data comes from asking the developers which game engine they used. Here's the breakdown:

It's very similar to last year, but in 2021 Unity lost a little ground to both Game Maker and Godot.

I did five YouTube livestreams where I played random games from the jam. I streamed for over 10 hours total in one week, and the peak concurrent viewers was 4,124!

Anyhoo, I think that's everything. A quick note that if you're a Patron and you made a game for the jam you can hop into the #game-dev room on the Discord and I'll make a video of myself playing it. Offer ends Friday.

And now... back to normal GMTK!