Home Creators Posts Import Register Favorites Logout

Content

Hey all! As is customary, I like to wrap up the GMTK Game Jam with some thoughts, feelings, and lessons for future jams.

If you somehow don't know by now... GMTK Game Jam is a community event where thousands of game designers try to make a game in just 48 hours. And the game has to fit a theme - this year, that theme was "Roles Reversed".

Once again, this jam was bigger than the year before. 23,193 joined up the jam, and they collectively produced 6,836 games.

It was almost 7,000 at the very end of the jam, but people are free to pull their game out of the jam, so the number ticks down over time.

(And yes, we broke Itch's servers again).

It also had the most ever ratings: 160,000. Wow!

What went right: Clearer rules and directions

I've gotten much better at making the rules as clear as possible. The jam page on Itch has all the relevant details, and I now change the page as the jam goes along. When people are submitting games, for example, I add a piece at the top about how and when to submit.

I also send details in an email that goes out to all jammers. And we communicate things well on the Discord.

The main thing that's tripping me up now is just weird stuff to do with the flow of submitting games through Itch. It should be easier to submit your game to the jam and harder to remove it - that way I wouldn't have to spend so much time explaining this to people!

I'm always working with Itch to implement new features, squash bugs, and improve the flow. I can't get everything I want - it's their website, not mine! - but they've been great at taking my requests on board. Hopefully we can get these last few issues resolved and make the whole process friction free.

What went right: Zero tolerance for issues

If there's one thing that makes the jam horrible for me, it's dealing with people who miss the jam, delete their game, or need help updating a broken game. It takes a lot of extra admin time, people get impatient and argumentative, and it makes me feel miserable to turn people down.

However, it's simply not feasible to help everyone, and not fair to pick and choose eihther.

So the best thing to do is simply have a zero tolerance policy of "sorry, better luck next year!". I also spend a lot of time stressing how important it is to upload your game on time - and make it clear before hand that there will be no late uploads.

This year there actually weren't that many people who missed the jam, considering the size, and they were mostly fine with my "sorry!, nothing we can do!" response. Another thing that really helped was pushing people to the Post Jam Jam, as a sort of consolation prize.

The more years that go by where it's clear there's a hard cut off date, the easier this will become.

What went right: More moderation

The GMTK Discord is the community hub for the Game Jam. Every year thousands of people enter the server to chat about the jam, get hyped for the theme, troubleshoot problems, and so on.

We've always done a pretty good job of moderating it, thanks to server stalwarts like Slam and Adam. But this year we went hyper mode and hired a bunch of extra helpers. That meant there were always at least 2 or 3 people watching the chat, 24/7. They could quickly answer questions and shut down spam, intolerance, and unhelpful arguments.

Getting new people up to scratch on how we like to moderate things is always tricky, so there were a few bumps and problems. But nothing major, and I think the server ran more smoothly than ever.

We'll definitely have a big team of mods going forward.

What went right: A good theme!

Last year's theme, Roll of the Dice, had its issues. It was difficult, limiting, and led to lots of same-y games. This year's theme, Roles Reversed, was much better I think.

It was definitely tough, but people rose to the challenge and made some incredible stuff. And while many games followed my lead from the theme announcement video (essentially - take a popular game or genre and then let us play as a different character), many others did wild and unexpected things.

This was the first year that I checked the theme with another game jam expert, to ensure it made sense. That was a good idea - I felt much more confident heading into the jam.

What went wrong: AI woes

Ah. Sigh. So, as the jam was approaching it suddenly dawned on me that I would need to make some kind of ruling on whether jammers could use generative AI models like Midjourney and ChatGPT.

It was tricky, because I was juggling lots of different arguments.

For one, generative AI does allow people to focus more on game design - the most important bit of the jam - and less on presentation or code. We already let people use preexisting assets, so why not AI-made stuff?

And also, it's really hard - if not impossible - to police this stuff. Can you know for sure if a background was made by Midjourney or a human? And, my word, what about code? Game Jam rules should be easy to follow and effortless to impose - AI really muddled that.

But then there was the more ethical side of things. Many people are wary or outright critical of AI. There are serious concerns about how these models learn from human-made stuff, and thorny questions about how unscrupulous publishers could use AI to replace coders and artists. Do we really want to support this stuff in the jam?

I'll cut a long story short - I initially announced that the jam would allow AI code but ban AI art. That made a lot of coders mad. So I change my mind and banned all AI stuff. It was a messy announcement and I regret it.

In the end it was fine - people followed the rules, there wasn't much moaning, and people were mostly happy for the jam to be AI free. But rules need to be crystal clear and consistent, and that was not the case, thanks to my flip flopping. Whoops!

Plans for next year

Next year things will likely stay largely the same! We've got GMTK Game Jam down to a science at this point, so it's only little twists and tweaks from year to year.

It's only weird curveballs - like the boom in generative AI - that mess things up. On that note, my plan for next year is to keep the rules as they are with a firm ban on AI-made stuff. But I'll keep an ear on the mood and get more feedback closer to the time.

One thing that I'm considering changing is the weights of the criteria, so presentation counts less towards the overall score.

The argument for this is that sometimes really clever games get shut out of the top 100 because the creator(s) are not brilliant artists. And some games can score highly because they look amazing - even if the gameplay is just so-so.

The argument against this is that while GMTK Game Jam's focus is always on design, I wouldn't want to devalue the efforts of artists and musicians, or dissuade them from entering the jam.

I've got a year to make up my mind, but any and all feedback will be useful!

I also want to do more to clear up broken games and submissions. It sucks to rate games and then get stuck with dead games and busted EXE files. I tried moderating this myself but it was an impossibly large task. I'm impressed by the efforts of the mod team on the Discord, so I could definitely get an army of volunteers to help me sort through game reports.

Stats

Let's finish with a look at some stats.

We can get an idea of team sizes, based on the number of people assigned as contributors to a game. This is optional, so it probably doesn't give the full image. But from the data we do have, we can see that...

The vast majority of games are made by a single person.

And having more people on your team will improve your rankings.

However, there probably is a limit! Look at Normal Activities - a game made by a whopping 24-member team, including pro game devs like Jonas Tyroller. It certainly did well, but puts paid to the idea that a huge team will guarantee game jam success!

Thanks to Jamalyzer for this info.

For game names, the most common words were dungeon, tower, reversed, reverse, hero, space, revenge, and simulator. The most common game name was Tower Attack (18 games), then Tower Offense (14), Dungeon Master (11), and Quest Giver (9).

The longest game names were...

"I am a Monster Saving the Princess I Captured While Trying Not to Let My Boss Dragon Notice Because They Are Unpleasant"

"PLEASE HELP ME I'm a lich in a dungeon but this knight wants to kill me and each time he dies he comes back stronger"

"The night I went to convenience store won a lottery but got kidnapped by alien then started to escape or die"

Every game (barring the disqualified games) got at least 6 ratings. Making the Game had the most ratings at 1504. The average number of ratings per game was 23.5 and the median was 16.

Roughly 80% of developers made their own art (compared to 20% who used preexisting assets). As for music, the split was essentially 50/50.

And then in terms of engines being used, Unity is the big winner as always, followed by Godot, and then a small chunk of Game Maker and Unreal Engine games (plus a few other engines like Construct and PICO-8).

But what's interesting is to look at the historical data. I've been collecting this info since 2020, and so we can see how things have changed over the last four years.

Basically, Unity is slowly shrinking every year, and Godot is quickly growing. I saw someone on Twitter do a graph showing, if this trend continues, Godot could overtake Unity in 2028. But while Godot is slowly biting away at Unity, it is demolishing Game Maker.

Probably because Godot is in the same niche, and because Game Maker switched to a controversial subscription pricing system in recent years.

GMTK Alumni

One of the great things about running a game jam is seeing jam submissions take on a second life as fully-fledged games.

So I thought I'd add a new section to these post mortem posts where I can shout out new games that can be traced back to GMTK entries. These were released since the end of the last jam, in July 2022.

The most notable game is certainly Rollerdome, which came out shortly after the last jam, in August 2022. I covered this one in a GMTK video, so I'll let that do the talking for me.

Another big game was A Little to the Left, a cozy "puzzle game that lets players sort, stack, and organise household items into pleasing arrangements". It was released in November last year on Steam and Switch. It was even featured in the Wholesome Direct.

A 2021 winner, Factori, has been expanded into a massive new Steam release, as Word Factori, and will be released on August 3rd. Good luck to Star Garden games.

At GDC this year, I was thrilled to see that Slider - a GMTK Jam game - won "Best Student Game" in the Independent Games Festival. Good luck to them on turning it into a bigger game.

I try to list all of these games on Steam and on Backloggd.

Thanks for reading! I'll see you next year!

Mark

Files

The Best Games from GMTK Game Jam 2023

Full GMTK Game Jam Playlist - https://www.youtube.com/playlist?list=PLc38fcMFcV_vnAZjugCRdKr8_8d_y_rRl The GMTK Game Jam for 2023 was - once again! - our biggest yet. In this video, Mark runs down his 20 favourite games, from a jam where every game fit the theme “Roles Reversed". Support Game Maker's Toolkit on Patreon - https://www.patreon.com/GameMakersToolkit Play the GMTK Game Jam 2023 games - https://itch.io/jam/gmtk-2023 === Winning Games === 00:00 - Intro 01:20 - Princess Paladin - https://woutervanvugt.itch.io/princess-paladin 02:10 - LedgePainter - https://charlie-wagner.itch.io/ledgepainter 02:53 - Bready or not... - https://aramilion.itch.io/bready-or-not-unsurvivor 03:49 - Inside Job - https://featurekreep.itch.io/inside-job 04:23 - Sneakerdoodle - https://pulni.itch.io/sneakerdoodle 04:55 - Bricker Breaks Free - https://angry-armadillo-games.itch.io/bricker-breaks-free 05:52 - VERSUS SNAKE - https://pizzart.itch.io/versus-snake 06:21 - Descriptionary - https://fingerclap.itch.io/descriptionary 06:53 - Last Piece Standing - https://sandroleon.itch.io/the-last-piece-standing 07:27 - Get Packing - https://dawnscythe.itch.io/get-packing 08:06 - Crabjuice - https://dev-dwarf.itch.io/crabjuice 08:42 - Walkies - https://tosmaster.itch.io/walkies 09:19 - One Day Better - https://aquanoctis.itch.io/onedaybetter 10:07 - UVSU - https://dietzribi.itch.io/uvsu 10:49 - Reversal Rooms - https://taelsdafoox.itch.io/reversal-rooms 11:08 - RAGDOLF - https://weentermakesgames.itch.io/ragdolf 11:40 - Pause to Play - https://ecremprown.itch.io/pause-to-play 12:22 - Making the Game - https://kindanice.itch.io/making-the-game 13:11 - Our Dearest Player - https://blazerchlopiec.itch.io/our-dearest-player 13:32 - Streamer Screamer - https://organzola.itch.io/streamer-screamer 14:15 - Honourable Mentions === Credits === Music provided by Epidemic Sound - https://www.epidemicsound.com/referral/vtdu5y (Referral Link) === Subtitles === Contribute translated subtitles - https://amara.org/videos/hKlrG1vUeVQd/

Comments

Antonín Maloň

I read the graph about team sizes very differently: in the top 0-10%, there are less solo jammers and more larger teams. So larger teams overall tend to do slightly better. Am I wrong?

Chartor

It'd be really interesting to see this graph normalised to unit area! (Although that might be a little discouraging for solo'ists!)

Liliana

Really wonderful reflections. For what it's worth I think you're making the right call with AI. Out in the world there's all sorts of questions to be answered about it, but I feel like a game jam is a great opportunity to shake that off and just test some creative skills under pressure. Presentation scoring is definitely a trickier problem, made extra fuzzy by the use of pre-existing assets by some. Go with your gut, and when all is said and done it's your call! Thanks again for hosting an awesome community event!

Yegor Zhumikov

Our team would really like to see AI being allowed in next Jam. These days even using Photoshop means using some of AI. And when we google... Well, it has some AI inside of search engine for sure, but people simply try to close eyes on that. Even my ear-phones advertised as some think with AI to enhance sound quality (yes, it's not this type of AI, but I believe borders between all of that would blur in next few years if not months) It was really hard for us sometimes not to ask GPT about some code advises or simply when discussing something. And we saw no logical reason for such restrictions, which made us feel bad for Jam And those arguments against shady use of others work looks strange personally for me, because people do the same – they don't mind using answers from stack overflow or looking at other people art and latently use it in their work I would really hope that next year any AI would be allowed and I really think that won't change how Jam games look much or how good artists/coders/others better in their job – that wouldn't change in my opinion at all

Yegor Zhumikov

Besides that, also. It's not possible to see whether someone used AI for brainstorm/code. Simply no access to that information and instead of allowing to use it freely without doubts it allows to use it shady – just don't say to anyone and everything would be fine With that, saying that not a lot of people used it is actually strange. Maybe not a lot of people who were accused of that were caught with evidence, but we can't say anything about how much people used it for real (I personally think not much and most of us played honestly), but that doesn't make situation any better

Two Separate Gorillas

Its a good point that certain applications are utilising ai in them already to optimise workflow, as opposed to just asking midjourney to make an entire image.

MMG

I definitely do not agree with scoring presentation lower. As a musician, I really enjoy the jam and if my work was valued less to the score I would be seriously dissuaded from participating. I don't see how it is fair to lower the value of the contributions of artists and musicians because some do not make good art or music. There are many games in the top 100 with less than great music and art anyways. I really hope this doesn't change next year, since there are many musicians and artists who are not game designers and still enjoy participating in jam. This seems like devaluing our work.

MMG

I'm not a programmer, but I do have to ask, what did you do last year? Sure chatgpt can make things easier but the bottom line is on 2022 no one was able to ask chatgpt anything and we still had an astounding level of goo games. I think at certain points ease has to be circumvented, but correct me if I'm wrong here.

Yegor Zhumikov

If you mean 2022 as last year – this year was my first GMTK jam. You can see my work here http://yogurtthehorse.itch.io/, but I don't see how is that connected to our discussion :) And don't get me wrong – I don't say that all of arguments are nothing or shouldn't be considered. I even said that generative AI won't change Jam much and definitely it's not possible existence of AI would make games dramatically better than they were and (at least for now) forbidding AI doesn't make Jam worse All I am saying that in my opinion and in opinion of my team where we do all the assets ourselves, we think that using is AI is better than forbidding to use it in overall comparison and considering all arguments

Lauren

I think scoring should be changed so that the weakest score is weighted less heavily in the overall ranking so as to be fair to those who are about the art and those who are creative at the same time As for the AI debate, if you cannot enforce a rule you should not have it. A rule you cannot enforce just rewards cheaters. So if you cannot enforce AI bans you are hurting your honest participants.

Lauren N

While I understand the point about presentation scores causing good games with bad art to score lower than they perhaps should, I think it's important to remember that the jam is not a fair match and it likely never will be. There are plenty of unfair reasons one game idea does better than another. After all, you have teams of 20+ versus individuals. Even if you were to lower the weighting of presentation, people will still probably rate a game higher in all categories if they enjoy it's appearance, because it's hard (if not impossible) to separate the different parts of your experience with a game. So I doubt it will fix the problem much. Besides, it already feels like a loosing game as a solo artist. No need to dash our hopes even more. Anyway, good job and thanks to you and your team running the jam! It's always a lot of fun every year.

Wayfarer Dev

The issue is specifically with generative AI - the AI in search and headphones etc is not that. It's all about how much content it ingests and spits out, and where that content comes from - it's usually stolen and used without the owner's consent. You specifically say it won't change the jam much, of at all, so what's the point in adding it? 😉 I don't think it's possible to ethically run a game jam like this and allow the use of generative AI, it's a question of morals more than anything. THAT is the crux of the issue. It basically devalues the entire creative process, and this jam is about human creativity after all.

Yegor Zhumikov

> it's usually stolen and used without the owner's consent I already told my thoughts on that in text, so I won't repeat > You specifically say it won't change the jam much, of at all, so what's the point in adding it? Well, this one I also mentioned in text > It basically devalues the entire creative process It's not something that replaces humans, so I don't agree

Willhart

I think this jam could have been a great opportunity to discover more benefits and use cases of AI made code, art, and ideas for those who have been experimenting with the new tech a lot. I also read that at least one good game got banned for using AI art too (one where you forget the lines when going on a stage and have to improv).

Willhart

Like an option for it would have been nice.

H.B.

The reversal on the stance of AI when it comes to code made no sense to me. E.g. check out the StackOverflow survey on AI tools being used (https://survey.stackoverflow.co/2023/#most-popular-technologies-ai-dev), more than half of respondents are using GitHub Copilot, a paid tool; that should give some indication how much people actually are in support of such tooling. It also is not like people use such tools in the vein of "write a game for me". Most of the time it's just a fancy auto-complete that writes *exactly* what you would have written yourself anyway. And as someone else noted, if the rule cannot be enforced it does not make much sense. Especially for code, it is virtually impossible to tell and it is not like the source is being published or reviewed by anyone.