BITKRAFT Summit 2025 Day 3 - EVE: The Longest Game with Hilmar Petursson (CCP)

Hey everyone. Um, good morning. Good to be here. Uh, Hillilar and I were talking a few days ago about how to approach this. And he said, "Let's do a random fireside chat so we can talk about whatever makes sense at the time." So, I have probably three questions that are going to be very open-ended and you can maybe answer them or maybe just talk about whatever you're going to talk about. Anyways, So, one time you told me a story about how one of the most pivotal moments in Eve came from a player defrauding a bunch of people, pissing off most of your community, your game almost going bankrupt, and then a story being written about that scam and then all of a sudden Eve was back. And I wanted to start by asking you when you think about the 22 years of Eve so far and the luck versus skill spectrum and how much of it has been as you say the business of staying

in business versus these very random things that have made the business what it is today. How do you think about luck versus process? So um [clears throat] I I I don't really think it's it never really felt like lock. Um it uh it felt more like okay there's some intuition about something and now you're like faced with like a a fork in the timeline like are we are we going to go that way or this way um and uh and when these moments come to you they they don't have any instruction manual. There's no real kind of way to do it. And um and I I think it's largely been this one just feels right. Um and then you go that way and that is kind of how you prove whether your intuition is correct or not. Obviously this creates massive survivor bias. This is not a very helpful teachable moment of like okay

here's a process to figure out uh uh what is right etc. But I mean this is generally why it's hard to start up new things and whatever. You're generally just wrong and then there is survivor bias and some people come through it. So I think um at some point we just had this murky idea of what we wanted to create. We wanted to create this world that was kind of belonging to the players that were playing it. And I mean even online often is called like a web three game before there were web three games or even ideas of it because we very much wanted to abdicate as much as could be done to the players um and make the the game kind of operate as if it was a real world. So every time these moments came and there were several moments we just always came back to that. Okay, in this case, here's a person uh that person uh defrauded 500 people inside the game uh because they

had joined their corporation, gained trust, moved to the top, got in the keys to the coffers, and then took all the assets away. Um and uh and when it came time to decide should we undo it, which we easily could have done. we could have just rolled back the server state and now it's like as if it didn't happen. Um it just felt wrong because okay we will have we will have um we will have broken the world like the the timeline of the world would have been broken. Um so um and and then that was kind of the reaction of like okay we can't really interfere even if we can it just feels wrong because the world should and the community should somehow resolve it themselves. And then just to fill you in on on the details of the story. So these 500 people they they quit the game in protest of us not doing a very simple action of just rolling back one day uh or even just printing

the money and giving it to them. Um and uh and the game at the time was teetering on profitability. So um 500 people uh stopping playing the game out of a player base of 30,000 people was the difference between the company being profitable and losing money. So um so when those people leave, our revenue drops, we're now losing money. We don't have a lot of money to lose. Um I now need to fire 10% of the company, which was three people, not to over traumatize it. uh it was still 10%. Uh and some of them were were childhood friends and then like okay there was revolt with the players now there's revolt with the employees this is stupid like just just make the change and it was just no it's not how it it's not how it works I can't explain it it's just feels wrong we're we're we're staying the course and then a few months later PC gamer writes eight pages

about the the first heist in a in an MMO and how it all played out and all that. And 5,000 people joined because they want to belong to a world where the nanny is not always watching. Um and uh and and and it wasn't like a very metaculous like, okay, let's sit around the table and make some very balanced decision of pros and cons and whatever. It was just nope, it feels wrong. We're not doing it. But then the company will die. Yeah, we'll [snorts] also die if we do it. So um so there there is something like in that and again this is not very helpful because we are the people that uh that survived that path. So but that is kind of how it is. You just have to go with your instincts and then you just fail or you succeed. Uh and and trying to be very methodical and very recent and all that. I just don't think that is the exercise of trying to create something

new that has never existed. You're so royally off the map anyway. You just have to No, we're going north. Why? Because it's there. I know it. You you you just have to channel that.
Cool. Thank you. Not very helpful, very entertaining. Um, so we were talking earlier today about how the meta of game distribution has shifted even in the last three years and now you're, you know, 22 years in. Are there major seismic moments where you say three years ago this happened and we had to make a big change. when you think about the big process improvements or changes of direction that you've done in Eve, is it every year just trying to adapt to like in these little little moves or have the changes been way more around hey we're going to launch this new game and we need to do one big change or platform switch or I don't know internet or distribution platform does change for

you happen incrementally or in step changes
um I think it is the endless tactical refinement there is like a tactical layer which is kind of changing I don't know what to say quarterly half a year yearly it's like a tactical layer of like okay now this particular tactic this channel this message etc this way to manage reviews on steam this way like that's kind of a tactical level that happens now it feels more like there's some strategic shift um in um in how it is kind of to get to people and make them excited about joining a game. I think we can all throw out some straw men about like what is going on from Tik Tok eating so eating minutes and of attention to 18,000 games on Steam coming out just fills the game fills the

world with um some abundance feeling where uh where all that or the black hole game theory where people are just so stuck in their current black hole games etc. I mean I I think we can all throw out straw men about like what is going on but something is going on. So I mean as we have been taking this game we're now making to market we just threw out the playbook of just what would we do for Eve? What we would we have done three years ago and we just like we have to just re reinvent everything. So uh and that has been a a long experimentation of like okay what is um what is the whole p performance social landscape like what is the whole viral growth landscape like what is the whole kind of built it with the community and and uh engaging the community very early on what is the meta like there and um I mean uh some of it is building up to something that scales

but we very much kind of okay even if We've been doing this for 30 years. We're just going to assume we are no know nothing. Um and we're just going to relearn it from basics. We're just u going to figure out a bunch of stuff, see what uh works out, have theories about why it works out, see if we can scale it and all that. And we've spent this whole year doing a lot of these kind of things. And I would say I mean it's not like we have a playbook on how we scale it up next year, but we are getting to a point where it is developing into a playbook. Um, but I mean the the the very intentionally we started doing this very early. I mean we're in some Fun access with a pre-alpha version of the game, but also like even if the game is like that and still I mean it's kind of a platform for modding and all that. So I mean people have plenty to do but just how how do we message that? And and just to give you an example like first when we made this

development platform which is built on top of a blockchain where we not use the smart contracts in a way to make a service site modding engine. So first we thought about like okay it's for third party developers. People didn't really understand that. Then we went into okay it's it's you can mod the server in an MMO. And then people were like, "Uh, okay." And now, okay, you can mock the server in an MMO without our permission. And now we have some like server peri per permissionless modding. It's kind of the the pointy edge of the spear. And like every time the message kind of and and there was a lot of other thrown at the wall, but when we got into the okay, you can change the server in MMO as the game. It's different for everyone and you can do it without our permission. Then people came kind of ah okay that and and this has been a long journey of figuring out like how can we make that value proposition which is

fairly unorthodox very sharp and so that it cuts through the very confusing landscape of all the endless amount of slop that is just filling up the world.
Cool. So, [clears throat] um, probably three years ago, I asked you for some career advice, and you told me that up to that point, which was, I don't know, 2023, 2022, um, the job of a VC had been to find really talented leaders that could scale 50, 100 person organizations underneath them. And that from that point forward, the job of a VC was to find a team of three people that could use AI to do everything they had to do. Um, now if I had followed that advice, we probably wouldn't have invested in you because it certainly hasn't taken three people to make e Frontier so far. Um, lot has happened in AI in the last three years. Is it time for me to actually follow your advice? And you know where where do you think we are in AI for games today?

[clears throat]
So I I think um we are I mean we're maybe at like the tool states kind of like uh I mean you probably all see it like okay now if you need to make a mood board instead of going to our station and Google and copying pictures and make a mirror board etc. U now the LLMs have already done it for you. they've already copied all the art station pictures and all the Google images and whatevers. So now you can just make the mood board like that instead of having somebody spend a week to make a mood board for something like now that is gone and you just get the results from printing. Um I mean there's a lot of just efficiency and managing information and and and these kind of things documentation coordinating teams. Um, I mean, I've seen people do some like really cool sorcery with AI and Jira mixed together to kind of get a sense of of where

projects at, etc. So, I I think we're in sort of just the whole world is in this kind of AI can make stuff efficient. Uh, then I I think we're all having this kind of okay, it's going to be making assets, it's going to be making things, it's going to be VIP coating and all that. And that's probably all going to happen. But then we are making efficient a system that is kind of more or less like the system we used to have. It has just taken a lot of the sort of bloat out of it and the redundancies and it's become more efficient. And I think that is just happening automatically. If I look at my own company, we've not really been pushing any massive AI agenda. It just happens. People are just doing this because we need to ship games and make things and it takes a while and there's a way to make it faster. And then if you look at what's going on in the AI overall, then we've been following like okay now AI can write text, it can make a picture, it can make a video. Um and now right now we're in

the sort of world model states where people are trying to make world models to train AI inside their own latent space of embeddings which they've derived from VTO's etc. And I and I think in the year that we're about to roll into we we we are going to be seeing the thing where now people are using AI to make a game that is not just more efficient than it was before. It's kind of a similar process, similar game, similar outcome, but it just gains efficiency. We're going to enter a world where humans could not make this game like um and I I think that will come out of this world modeling uh concept which um which has a lot of like struggle to kind like okay how do you make stable semantics how do you express topology in an embedding in a latent space nobody can understand the latent space I mean it's not for humans to understand and how can you make like deltas to it and

just predicting the next frame is not enough like you saw in the sort of Minecraft hallucination demo which by the way now is way better. Uh but that showed both the the the strength and the weakness of this and it is about to get cracked how you can make stable semantics and then sort of hallucinations on top etc. and and this is going to happen next year because it just happens so quickly. And then I think we're in some world where um where it's very hard to predict exactly what it is, what does it feel like, etc. because it's going to be a new format. It's going to emote like a game just like Tik Tok emotes like a video but but Tik Tok is not a TV station. It's not a YouTube. It is something different. It involves videos, but there already you're seeing the mixture of media and AI. Tik Tok is that there is an AI, not like an LLM AI, but an AI at Tik Tok

that tricks your brain into engaging with it. And the game version of that is going to be some like sorcery. Um, and um, I think I've seen enough to to uh to think like I I I I I think the the the best way to describe it, it is like navigating the mind of God. That's that's that's what it's going to feel like. And it is to some extent already feeling like that. So I think there's a lot of you know B2B SAS incumbents that look at vibe coding as a major threat as you can build purpose-built software for an organization and does intermediate that whole thing when it comes to games or more broadly interactive entertainment. What do you think is the role of vibe coding? What do you think that eats the market share of? And how do you think about shared context,

virality, playing things with friends versus the opportunity to fragment content and entertainment into ever smaller bits that are hyper customized to who we are? So I I think I mean it is it's very dangerous misleading to to make these signifiers labels on AI like saying it's wipe coating and then now we have been saying wipe coating for I don't know eight months let's say that and maybe eight months ago it was wipe coating now it not that it is something different and uh and when the tens of trillions of parameter models comes out the context window will be so long that um your whole code base is now the context window I mean even line is like 10 million lines of code at some

point it is just a context window all of all the code in line and then you you you you will gain perspective it's again no human can have uh unless they're more at an architect level etc. But they can still not nobody can take uh no no can operate on 10 million lines of code to transmogrify them into something and because the context windows have been short and it's been kind of frustrating etc. Once you're at that level you you've now gained the tool for example with online where we had 10 million lines of code
that 4,000 people have written in over 30 years. I mean it's a lot to deal with and the world is full of these kind of systems
and nobody can really
nobody has any agency in that we're about to gain agency over it. Um, and that I mean when when when people are making like a dated concept like it's wipe coating again just take the it's

evolving so quickly. You you you should take great care in not getting stuck in a in a model you have made of this that is now half a year old. It's just doesn't apply anymore.
Yeah.
So [snorts] I mean when we think of that I mean wipe coating I mean we used to do paper prototypes now we just don't. with just somebody makes a prototype and it takes a few hours and now we're playing a prototype of the game like a paper prototype except it's client server, there's a website, there's a leaderboard and we're already playing a prototype of the game. I mean I mean that's a massive super brow. Oh my god, how these paper prototypes were cute. They are not like super productive. Not compared to okay, here's just a website version of the game. We're playing it. Um somebody spent a weekend of doing it. We're we're having this moment again and again. We make like okay somebody make up pits for a game no few concepts of game. It's like okay here are 10 ideas okay these three go wipe code it and we'll play a

week later
like that process of like discovering what to do and all that has just compressed into
like and the quality of the output is way better
point of view.
So and again okay you can use the current coding to do that. what you will be able to do the next year is something else and that needs another name
uh and that will be very different and in this world what will happen and that's the
change it because after I spoke to Jasper
that's the I don't know part you would then
all the people which are
with what I say instead
okay at jobs will not have jobs
uh the people which are great at their jobs
they will just have 10x the value that they used to have so the the where we're going to end up with this
how does a junior becomes a senior it's very hard to see frankly um and how do the people which are kind of okay which definitely have a have a role in modern gay game development it's a lot of work that just somebody needs to do but that

work is going to go away it's already kind of going away
last question I'm going to steal Yenza's question
and we're going to see if we get a different answer on stage today
in the clicker.
As someone who's been building games for 30 years or 20 well 30 years I guess 20 years for the game to come out
would you rather be a builder building a new game today where you have the black hole games and the incumbents and the budgets are out of control and all that stuff but you also have AI and whatever else or 15 years ago when you started which part or which uh time frame do you think was a more fertile ground to go build some cool I mean absolutely today I mean I've been there done that I don't need to do the other thing again like uh there's no there's no challenge in that I mean for sure there was challenge in that but it's just kind of
okay
now it is exciting because the
distribution is broken creation is

broken attention is broken it's all broken and that like requires some like real innovation to figure out that just Like when we were I don't know.
No, it's you.
13 nerd. 13 nerds in Iceland that had never made a game.
We hadn't even seen a game developer from afar like like when Line ships nobody on the team had like pointed at the game developer like
the and we just had to invent it. And that's why I mean Ean Line is a weird ass game. Like I can say that with all my love for E line. like the we people often say to me like, "Helmar, why don't you make it more like a like a game? What do you mean?"
Yeah, it's kind of like where's all the game stuff in this in this game and them being offended, etc. But, but I mean, the reason why Line has like gone on for, I don't know, 22 years by now.

Thank you.
20 million people have played it. It's made $1.5 billion. The reason why is because it is a it is a game that was invented against all the all the norms. So now it feels like that all the norms are gone and even if I know them some of them by now I now know a little bit about how to make a game. Not so much though like it's still kind of weird what we do. That feels like okay what a time to be alive. Now we now we need to dig deep and come up with some craziness. Um, and uh, that feels a lot more exciting rather than, okay, here's a yet another genre that we're about to break into and and all that and have to follow some conventions about how that needs to be. Um, I think it's quite exciting to be in this moment where where none of what we used to know is very valuable right now.
What a time to be alive. Thank you so much.
Thank you. And with that, I'd like to

invite Lee, Emily, and Lou to talk about scaling journeys.