Golden Whale Productions: https://www.goldenwhale.com/
Episode Information
In a riveting new episode of Player Engage, host Greg Posner sits down with Thomas Kolbabek from Golden Whale Productions to demystify the complex relationship between data and game design. They agree that while data is a powerful tool for objectivity, it’s not the be-all and end-all. It needs to be balanced with player feedback and expert opinions, especially when many of those making design decisions are not even players themselves.
The duo also tackles the sensitive subject of data sharing between corporations. Thomas underscores the principle of trust and introduces the concept of a “Chinese wall” to keep sensitive information secure while allowing for mutual benefits.
To round off the discussion, the importance of multi-system data integration is emphasized. Whether it’s marketing, product operations, or other domains, aligning data across the board can significantly boost player retention and monetization. Thomas shares how Golden Whale Productions leverages insights from a range of sources to create a holistic strategy that delivers on both fronts. Don’t miss this insightful episode of Player Engage, where data meets the gamer’s heart.
Transcript
00:00 Intro Welcome to the Player Engage podcast, where we dive into the biggest challenges, technologies, trends, and best practices for creating unforgettable player experiences. Player Engage is brought to you as a collaboration between Keyword Studios and Helpshift. Here is your host, Greg Posner.
00:16 Greg Posner Hey, everybody. Welcome to the Player Engaged podcast. Greg here. Today, I’m joined by Thomas Kolbabic from Golden Whale Productions. Golden Whale Productions does insight in player gaming, and it’s going to be exciting to hear how we can use this data to make smarter decisions when creating a game. So Thomas has been the CTO at Golden Whale for about the past 18 months. Before that, he was at at GreenTube for over 20 years. He did a lot of advising gigs in between there. And before I steal all of your thunder, Thomas, thank you for joining me. Is there anything you’d like to go deeper on about yourself and introduce yourself?
00:49 Thomas Kolbabek That summarizes it perfectly. Thank you. Thank you for having me. Hi to the audience. Yes, maybe just a word on GreenTube. So we started out as early esports in 2000. So that was when I was completely into Quake, Counter-Strike, and all these other things that were current back then. and wanted to do actually a player engagement portal by allowing clans to play against each other. So today are esports leagues, obviously way too early by then. So then we transitioned into subscription games, into pay-to-play games, all the way into creating in-game advertisement systems, which are having a renaissance today again. Yeah, so literally went through all the different categories in engagement, in retention, ran platforms with upwards of 70,000 current players across the globe, so global audiences. And yeah, happy to be here and talk about retention.
01:42 Greg Posner Yeah, before we get there, let’s start super simple with an easy, easy-ish one. Maybe it’s not as easy as I think, but Thomas, you are a CTO for a data collection insight company. You, I believe, have a medical degree background. When you were younger, what did you want to be when you grow up and how did you end up here?
02:05 Thomas Kolbabek Good question. I actually had no clue whatsoever. When I was 18 in Austria, you joined either the military service or the social service, which sort of gives you a gap year between school and university. And I was sort of struggling between studying medicine or computer science. Both of my parents are doctors, so that was sort of the obvious choice in terms of talking to my parents, not that they wanted me to become a doctor, but it was just, I was relating to those topics. Also very much interested in, so all in nature science across the board, so physics, chemistry, biology, and computer science. And what mesmerized me back then was actually in the 90s, that you could show data from a table, from a database table, from the early databases on the website. Back then with HTML and like really, really old school techniques compared to today. And that prompted me to build websites, to start to build websites as a webmaster. Back then, I think it was called. A check of all trades in the end. And because I did so much gaming and was interested in computer science, I applied to the gaming company. And then the next 20 years sort of flew by. So it was I would say a bit coincidence, but also I was looking for a job in gaming with the skills I had, which I think is also a common topic, how to get into gaming. Actually, if you like gaming, gaming companies are so huge, no matter what you can do, even if you do accounting or anything else, HR, you can always work for a gaming company. Yeah. So sort of with the interest of data into a gaming company, I would say.
03:42 Greg Posner Yeah. So that’s good. Right timing, right? learning how to do the web became kind of in the master of that and then kind of. built out there. That’s awesome timing.
03:53 Thomas Kolbabek What was- Although to add to timing, I joined the company in 2000 before the dot-com bust. So that was actually an interesting learning. So I went through both 2000 and 2008, obviously, and all the last years, which are like a continuous crisis. Like it feels like.
04:09 Greg Posner Every other week now, it’s a crisis, right?
04:13 Thomas Kolbabek So it started off on a high, went quickly to a low. So monetizing games with advertisement. My first job was to actually try to sell advertisement in games in 2000, to beer breweries after all, who like all hung up on me, like all 160 I called. But eventually that was a learning to not sell advertising back then and go into subscription, go into in-game advertising, in-game monetization and so on and so forth. So it helped steer the way.
04:43 Greg Posner Monetization is this whole fascinating subject, right? I mean, we saw recently the Unity news of them trying to force their ads. And that didn’t get quite the reaction I thought they were. And I think we have a whole episode based on that. But, you know, it’s interesting because let’s kind of pivot this with the information that you’ve learned about kind of in-game monetization and different methods you can do it, right? We want to take a look at player insights. That’s what Golden Whale Productions focuses on, understanding kind of the insights in your game. So let’s start high level and start breaking down what this actually means, because it’s fascinating. So first, let’s talk about just the data landscape. I love data, most people I talk to in this industry love data, and people want to collect as much data as possible. The problem is once most people collect all that data, they’re like, Now what? I have all this information. So let’s talk about let’s pretend we’re starting an indie gaming company, right? Is collecting data from the beginning important? And when would you actually start collecting data? And kind of what do you do from that beginning stage?
05:43 Thomas Kolbabek So that actually, there’s a very nice, I think it’s a Chinese proverb. It is probably, you know, it’s like, when is the best time to plant the tree? And usually the answer is 20 years ago. And the second best time is now. So the same applies to data. What we see in the industry is everything from fire and forget games, like I would call them. There’s nothing wrong with them, by the way. So these are games that are built sometimes in shorter timeframes, let’s say a couple of quarters or a year or two, deployed and then monetized quite quickly after the launch. So there’s a huge marketing campaign, a classic marketing campaign, then they are sold by either online channels or boxed, and then they have to earn their money within a certain timeframe and then they are forgotten. And that’s part of the business case. So a bit like a movie minus the, it used to be DVDs, videos, now Netflix, long tail. So this is one group of games which still exists. And some of them don’t collect any data at all. So they might record crashes or anything else that happens to the game, but those might not even be player related. So they don’t have a player identifier. And then you go into games that collect at least some, let’s say, log data, so in terms of game loads, game starts, statistics, so basic data, not in-game play. And then you move over to operational financial data, like how much was earned, how much was paid, so beyond whatever you find in an app store, because you can get most of those reports also out there, but track it in your own system, on your own servers or virtual cloud, so within your control. And then obviously you have games mostly in the mobile landscape, but also in the forever games category. So games that last for a very long time, but everything is online. Sometimes you can only play them online, but you can play a majority of the features online. And obviously these collect everything online. It might be real-time online anyhow, so latency is important. It was important in 2000, it still is important. to have a good gameplay experience. So these are sort of the ranges of data. And of course, if you ask me, but that’s a bit biased, naturally, you should collect it as early as possible, even in the development process. So I’m not saying that you should build your development process around collecting data. So you’re building a game after all, you’re not building a database. So the product is still the game, but it doesn’t hurt. Quite to the contrary, it’s a huge benefit to track metrics, to track events. So that doesn’t need to be a complex database, by the way. So usually an event stream, so event sourcing is, for example, a very nice technique if you build a new game, to just have a stream of events recorded. with maybe even the purpose to replay the game. So we built games deliberately for cheating prevention, actually, because those are not online games, so you could play them offline. But we recorded all these events encrypted with a vector, let’s say in the beginning, to have every game a bit different, to make it harder for the computer. But we still found people that managed to actually create the bots to play the game through. And then we caught those by replaying the game and figuring out which game was played by a bot or which game was cheated completely. So they sent, let’s say, a race time that didn’t fit. the recorded route, and then those games were discarded. And by the way, we thought the game was pretty solid, but that was actually done within sometimes minutes of a release. So people already waited for it. So again, as early as possible, in my opinion, if you have legacy systems, because some game providers out there, on the other hand, have games that are around for a long time, they have classic databases. And then there are loads of techniques to get data out of it, to make data fluid. You can use change data capturing and other tools. So there are very nice tools out there like Apache NiFi, for example, which also have a low impact on the operational, on the resources. Because typically, if you have a DB and it’s used for online transaction processing, you don’t want to slow it down at all. But again, it’s very, very specific, typically depending on the age of the product overall, and obviously also on what you want to achieve. So what kind of game do you want?
10:00 Greg Posner So a couple of questions to kind of build off that, right? Two separate questions, but maybe you can answer them in one, right? The first one is going to be, why create a fire and forget type of game then, right? Is it maybe just to test out the market to see if the market is accepting it? And then the second part of that is, I assume collecting data in the beginning is going to be easier because I can start to define the metrics I want to save and the KPIs that are important. It’s easier to stick a poll and then say, hey, I’m doing this right now in the beginning. But what tools actually collect the data? Is it the Apache tool you were just talking about or people just exporting the data to a database? So kind of fire, forget versus starting to collect data. And if I want to, where am I collecting this?
10:39 Thomas Kolbabek You have to differentiate between actually accessing the data or, let’s say, sending the data, so the data source, which can be a function trigger or it can be a function call in your code that just sends data to a server or that stores data in a log file. It can be a third-party product like Apache NiFi that you attach to a database. So that just gives you data. You sort of get data input. Then how you store it is a different decision or discussion. So you can store it in a classic database, you can store it in the cloud service. There are like gazillions out there at the moment. So since the NoSQL movement or NoSQL introduction, you can literally store data and everything. There is also a cost aspect of it. So some of those cloud services are very, very useful, but they also scale very well for the companies offering them naturally. They also provide a lot of service. So managed databases, automatically optimized databases. So it’s always a trade-off also on how you monetize your game and how much you will earn essentially based on the data that you send. Nowadays, it’s depending on where you send data. You can almost send, you can send a lot of data. without running into any issues. But then processing that is also a cost factor. Just moving data around, if it reaches, let’s say, terabyte regions, can be challenging at times, especially if you want to dig through all of that. So it’s just, it gets more and more inconvenient and more and more an effort. And that all has to be balanced with what to earn per player. What’s the budget of the game? So that can be very little, actually. Maybe it’s an indie game you create for yourself and you don’t even want to monetize it. So maybe a game is built to transport your opinion on how this game should be. So this is also games can be a work of art, and then you might not want to monetize this work of art, let’s say directly at least. At the end, you still want to create the next game and have some funds to do that. So it will be a push-pull. Actually, some games don’t want to go into that direction, but you still have to have something to eventually build those games. And yes, then other games you actually need to build off of that because there is so much competition or because there are simply so many games in the market. So when I used to play games as a teenager in the 90s, the only source of games were discs that I got in magazines. So before the internet, before at least I had internet access. So I got this magazines which had a floppy disk and those had shareware games on it. And those were the games I played. And I played all of them because I had no access to other games. But nowadays, if I want to play a billiard game and I search on my phone, I find like hundreds of them or thousands of them. So once I download one and install it, it’s even more important to have the data to make that game entertaining for me because I have 9,999 other options.
13:31 Greg Posner And this is interesting, right? Because it leads into two different types of conversations, which will probably take this within the conversation. One’s going to be user or player acquisition. How are we going to find new players to play this game as well as one I think you’re heading more towards is player retention. Are my players staying in my game? And both are essential to the survival of a game to be able to look at this stuff and monitor this stuff. But before we jump into that, because I feel like that’s a big bulk of it, is When you’re working with golden whale or you’re helping a customer are there specific metrics or guidelines that you’re pointing your customers to say hey you want to be able to collect this from the beginning and does it apply to different verticals as well right obviously retail or entertainment might not be capturing everything the same but is there a lot of crossover.
14:18 Thomas Kolbabek Yes, so in general, the overarching principle is an onboarding assistant, you could call it, and the retention assistant. So whatever you see with that. So whatever helps the user to enjoy the game in the first minutes or maybe even seconds. Because some of those games, I see it for myself, I download the game, I don’t like it, I install it after 30 seconds. So I delete it after 30 seconds, and I never download it again. So that already meant that I downloaded that game out of 1,000. So that already costs a couple of euros at least, maybe even dozens of euros, depending on the game category. And the other one is retention. So how do I like this game now for the next couple of years, if that is the intent of the game and the purpose of the game? And out of that, yes, you go from generic to very specific. So you have generic things like starting a game, starting a level, starting a game session, you could call it, so attempting something, so doing a race, starting a round, starting a fight, starting a puzzle, whatever it is. than engaging with the game. And that might not only be gameplay data, but also interaction data. So how does the user interact with the device? Where does he touch or click? How often? So you might find impatient clicks, for example. So customers pushing the start button multiple times, which is an indication that they might be bored. So they might just go on, go on. You have to make the game faster for them. Others might click around and they want to see the tutorial. But the other guy is really annoyed by the tutorial, because if I see another tutorial for a match three game, then I actually need to uninstall the game again, especially if I can’t skip it. And so that is all across the board generic. And then you have a genre specific, which might be the difficulty level of a race game. So we’re actually working with a race game developer. to balance the difficult level for the first couple of races by either lowering or upping the assistance factor. Obviously, this is a single player game. Otherwise, it will be more difficult. But so how much does the game support you in gameplay? So in steering corners, obviously, when you have more assistance, you can’t go as fast as there always needs to be a balance, but that’s part of the game. So the game would be at, let’s say, 50, so 0 to 100, game would be at 50 by default, and the game developer hopes that it works for all. But for some players, 30 might be right, or 10, or 99. And this is where we can help to take all that interaction data before the start of the game, like clicking in the menu, starting a game, maybe choosing a character, whatever information you have. And this is also why Even in your menu, it’s important to already collect data. This is all data before the first gameplay, before a player experiences the game. And then, ideally, having a great first race and an even greater second race, and so on and so forth. Obviously, you don’t want to interfere usually right in the middle of the race, because that might confuse the player if you go from 99 to 10, maybe. But again, maybe for some players that works. So that all can be part of an experiment. And then you go into very specific, which might be environment interaction. So this is completely different for open world scenarios than for closed world scenarios or level-based games. Player evolution. So role-playing games. So how do players evolve? What do they use to build? These are then time series problems. So what does a player enjoy doing the most? Watching my son play Age of Empires, he has a completely different sense of importance than I have. He usually picks whatever he thinks looks nice over what actually works better, in my opinion. But again, he plays the game differently than I do.
18:02 Greg Posner It’s interesting. And you know, it makes me think back to when I was younger, one of my favorite systems was the Dreamcast and the whole it’s thinking tagline that I had going on. And I remember playing NFL And it was one of the first games I think I played that had this dynamic AI that if you were getting better at the game, the computer itself would get better. And it wasn’t just, hey, we’re going to go from easy to medium or medium to hard. It’s going to say, hey, we’re going to put these players a little higher and these players a little… That seems like a perfect example of taking data, seeing what they’re visualizing, seeing how I play the game. And this was Dreamcast. And this was back in 99. Right. And this isn’t something that you really see that often anymore. Is it because it’s expensive to do? Is it because it’s hard to do? Is it because not enough people saw the benefit of it? Well, why would people turn away from seeing dynamics? And maybe it’s not even the right question to ask.
18:55 Thomas Kolbabek No, it’s perfectly fine because yes, this has been around for ages actually. So you have NPCs that act dynamically or in a dynamic fashion since the 90s or the 80s, since forever. You could play Pong against the computer. So the challenge always is to make a game engaging. You need some sort of some element of luck, ideally. Because if it’s a purely skill-based game, eventually there’s a very, very small fraction of players that will win at that game or excel at that game. And then you either have to fix it via matchmaking or via balancing the game appropriately. And as you said, like on NFL Dreamcast, raising that or making that dynamic in an engaging fashion can be done very simply. It can be done with a rule. So you can implement heuristics to just say if the player I played Need for Speed for way too long. So if the player always wins by 15 seconds, then you gradually make the cars faster or the opponents faster until it’s really just a fraction of a second. But actually, Need for Speed, after a while, initially, it was really, really frustrating for me. But after a while, you can always play it with your eyes closed. Or Mario Kart. So what I actually don’t like about Mario Kart is that if you do it well, you can very easily get to first place from wherever you are. It’s almost impossible to lose. And I would wish that this is harder, especially in the player versus player. And I believe that even in a multiplayer environment, if you advertise it or make it part of the game, you can create handicaps. So why not make my card heavier or whatever? Because I want to. Of course, you have to opt into such things. But yes, ideally, the game adapts And we believe still that heuristics are one way, but heuristics can never target every player specifically because there are so many factors in the game and so many data points you can record that heuristics only takes you so far as you as a human being can comprehend data. And that typically ends after the third or fourth dimension. So when it stops being visible in an Excel file. If anyone ever used a tool, I assume yes. Once you get into five, six, seven, 30, 300 dimensions, it’s something that you can’t comprehend as a human being. And that’s where AI is so powerful. And since lately, it’s also very powerful on an algorithmic level, not only on the hardware level, as we’ve all seen. So the usual suspects make that very obvious in the past 12 months.
21:26 Greg Posner interesting. It’s kind of, you know, I think the way you’re kind of talking through some of this is, I mean, the whole aspect of luck also coming in, right? I guess, this comes down to like, if you’re playing a game of Call of Duty, and all of a sudden, you’re so good, or you’re, you’re bad. And all of a sudden, you’re tearing it up, because somehow you landed in a party that, I guess, helping you out, right? I mean, I think a lot of this is kind of above my head on what I’m understanding. But it just seems so insane that the game’s truly adapting to my style of play, because someone’s taking a look at data on the back end saying, Hey, look at these numbers, right? They don’t match in the appropriate place. And I’m clearly dumbing this down more than it deserves to be by like, trying to put the right people in the right matches.
22:11 Thomas Kolbabek Yeah, exactly. So matchmaking is a very nice way to balance a game without interfering with the game. And even if I’m now, let’s say if I’m an amazing chess player, like using an ELO score, which is the usual way to do it, or other algorithms, I might have a bad day. So why don’t I just match an amazing player on a bad day with a medium player at a good So that player had a run, he’s really in it. The other player maybe sapped bad or whatever. Just match them. And then suddenly it becomes more exciting for both. So the intermediate player might win against the master, and the master actually needs to really beat that player, otherwise he will lose more points or credibility. And worst case, actually someone streams the whole match and makes fun of him. So there can be public embarrassment in that too.
23:00 Greg Posner I’m thinking back to the big chess.com boom we had during the pandemic. I feel like I’ve heard of a bunch of different scam type of podcasts about different people trying to scam chess.com and how it becomes very apparent right away.
23:15 Thomas Kolbabek We actually built a chess game that had issues with cheaters using AI, actually. using computer algorithms. And what we then did, we replayed those games and played it against the AI. And the AI was able to grade, in this case, the player. And if the player was playing superhuman, it meant that it used the AI algorithm. Because otherwise it wouldn’t be marked as superhuman. So the AI in this case graded the skill of the player and we used that to discard cheaters. Because obviously chess is very easy to… There’s no luck involved. So you can very easily use an algorithm to cheat.
23:55 Greg Posner So when you talk about that, right, I mean, you talked about multiple dimensions that computer can monitor, right? Like when someone’s cheating in chess, right, I can be looking at a YouTube video while I’m playing, I’d be doing something else, right? Like, is there something internally that I don’t even recognize that I’m doing that it’s going to signal that, hey, Greg might be cheating here? Is it moves that 99% of the people won’t make? And all of a sudden, Greg makes this move? Or like, yes, I have to imagine some crazy computer running on the back end that’s taking a look at all that information.
24:24 Thomas Kolbabek Yeah, exactly. So either it might be that we have enough historic data on you and suddenly your delta is just way too good. So no one improves like that. Again, a rule could say now if you improve by 30 points, then you’re a cheater. But that might be unfair because maybe you went to some chess summer camp and you actually became that good. So, that’s where all the data points before come into play, where then your likelihood of being a cheater is assessed by a model, by a machine learning model, which is much more accurate because it allows for this fuzziness. So, again, machine learning, you all have to understand that I think everyone that uses ChetchatDB or something else knows it. So, it’s not a concrete science. So, it’s not a fixed science in the sense of I get the right result. I get an approximate result that is likely to be correct. And that can also be an advantage, so that you don’t judge someone too early. You also might, of course, judge someone in the wrong way. So it goes both ways. And in this case, or in this particular case, we used the AI and the AI graded each move. So it would really replay all your matches in the background. And if it said, I would have done the same turns, then it was obviously the AI, because there is no coincidence in that. So if you really, you make whatever, 50, 60, 70 turns in chess, and all of them are exactly the way the AI would do it, and your skill level was not at that level, then you have obviously cheated. You might, of course, complain on forums or Discord or wherever, but at least you have some basis on it, or you put someone on probation. Similar tools are now used to assess unfair behavior in chess. in gaming. So whether it’s chat or whether it’s just people being annoying. So doing friendly fire, friendly shootings where they are not supposed to do. So this is also something you can track. So is someone bullying other players? And by the way, that also happens unfortunately in games like it happens in school.
26:21 Greg Posner And that’s something we talk about a lot and it’s just kind of toxicity and how that affects your game. Let’s go back to our fake indie gaming company that we’re building here. And I think a lot of our listeners are building these indie games and player retention again is something that’s always top of mind. Right? So we will always this balancing act of player retention, right? I think on our pre-call, you and I talked about, or maybe it was less about player retention, but you can’t ask players for too much, right? You can’t monetize by having a monthly pass and then also say, Hey, here’s a daily reward you can pay for it. Hey, here’s a pay, whatever, right? It becomes too much when you’re going to scare your player away. When you first sit down with a new customer and they come to you with their concern of, hey, player retention is my biggest concern. Where does this conversation start? And what do you start looking at or recommend you start looking at?
27:10 Thomas Kolbabek Okay, so we typically start is like you should do any scientific research. So we start with formulating an hypothesis. So what does the developer think that will happen or considers at the moment happening that should be avoided or should be mitigated? This might not actually be the real problem, but it’s important to have this starting point. So this also might be surprise me. So I’ve had my share of projects where someone gave me the briefing of surprise me. or excite me, in the sense of challenging us, which we then call the exploration of unknown unknowns. Because as we both know, there is like lots of stuff we know, lots of stuff we know that we don’t know well, or don’t know at all. So I don’t know how to do knee surgery, but I know that it exists. But then there are stuff that I don’t know anything about and I have never heard of. And this is the really important one where you don’t want to… I also played football for a while and I’m painfully aware of a blindside hit. That’s really, really bad. When you suddenly lie down and you only see the stars. and someone’s screaming in your face. So you want to avoid that. So we start with a hypothesis and then we take the data we get and analyze it. So we actually go make predictions that might prove or might understand better the hypothesis. That might be why are players leaving or why are players not taking my bloody season pass, I made such a nice artwork, and so on and so forth. And then we either validate the hypothesis or sometimes we also have to tell, no, it’s not effective, or it doesn’t work at all. And we also show what we don’t know yet. So that might be that a certain group actually likes the season pass, but another group hates it, like literally hates it, uninstalls the game immediately. And then we offer usually an algorithm that either provides a segmentation of the players into the characteristics that are important or provide suggestions on what to do next. So you either categorize your player base and then the developer makes a decision of what to do with those categories. So create heuristics. So you can’t create a heuristic as a human being easily at least from 300 parameters. But if I give you out of the 300 parameters, 10, or maybe even just four, that describe your player base, you can actually start to create an Excel file in the rule set and maintain this. Or, this is the next step, we would actually recommend the action to you. So, offer season pass now, or offer, what was the other one? Daily bonus offer, daily reward. So knowing when the user would appreciate an offer, and actually more importantly, when the user would appreciate also something else, which might not be monetary. So it doesn’t always have to be engaging a player through a sale. At the sale, you want something from the player, money, and you give something, but the player doesn’t know that usually in advance. So he hopes or he has expectations of something. I have the expectation that the season pass is awesome. But that might also be a disappointment afterwards. If I’m buying anything else, like the pizza at the restaurant, it might be awesome, but it might be really, really bad for me. My spouse might like it.
30:22 Greg Posner So you’re basically helping them break down their user base into these segments, right? Then you can start to understand how different segments are adopting, adapting to the different offers or different things that are out there. When Golden Whale works with their customers, if it’s something like, hey, we should offer a season pass now, do you provide them any insight and maybe marketing things that might work? Or do you partner with companies that might be able to help with that? Or you just give them the generic, we think here’s where you want to do this, and here’s where you want to do that.
30:53 Thomas Kolbabek Yeah. Next to analysis and predictions, which is our main business, obviously, we also do benchmarking. So if companies want benchmarking, they also agree to provide a certain set of data themselves, anonymized, obviously, and aggregated to a very high level. We don’t share the data initially or specifically, but we would share, like you said, findings. So what are best practices? What are potential conversion rates? Similar things like that. But obviously, only if you give data, you also get some of that. So we might do that. But we never share otherwise best practices or data or even specific insights, because naturally, this is also an edge that you have, and that your data provides. So by default, there’s a Chinese wall, and you don’t get anything, but also you don’t give anything, which I think is very important for safety and for trust to start this relationship. At the end, you give data to us, usually, ideally, all or most of the data. or at least access to the data. But of course, over time, you also might want to share some of that to get something in return. Some companies actually like to start off there. Some others, they more like to get their own operation to a certain level and then look beyond the plate within German, but I think it’s beyond the Ocean or lake or river.
32:08 Greg Posner Does, and this is going to be a silly question, right? Cause I know the answer, but I want to be told I’m wrong, but does player acquisition come before player retention or do you start thinking about how you save your players even before building a, building a game? Like, Hey, I want to make sure that when I have a player, they’re comfortable.
32:28 Thomas Kolbabek Yes, it also starts because your data trail can start there. So you have acquisition data, depending on what you can track, obviously, and depending on the type of game. So obviously a mobile game is different than a web game is different than a game on Steam or a boxed game. But you can track a certain extent potentially of stuff that happens before the game is even started, especially in mobile and web. And then you can use this again to learn or to group players and to aim your attention strategy on that. So yes, it’s important, but in the sense that you also want to combine that data. And what we see in many companies, also historically, that sometimes those two areas are very much separated. One is marketing. product operations, and they sometimes even have separate systems. So what we then do is also combine data from multiple systems. So there might be a gaming platform, there might be a player telemetry or game telemetry server, there might be a marketing server, there might be something else around it, a commission server. So maybe someone has to pay commission to someone else. And all that factors in because you want to steer ultimately retention also to what helps you in your monetization strategy. And if you have content that’s IP based, you probably get less of a share than if it’s not IP based. So yes, this is also important, but there is lots of tools out there to optimize acquisition. So we are not focusing on that part at the moment, but we believe that we don’t see many tools out there that focus on retention. And by the way, every player you retain, you don’t have to acquire. Just do the math. If you can retain 10% of your players, calculate the CPA, calculate the player base, and you know what you can save. And then compare that to your marketing budget.
34:13 Greg Posner And that makes sense, right?
34:15 Thomas Kolbabek And sometimes it can be extraordinary. Yeah. So. I’d say thousands or tens of thousands of euros.
34:21 Greg Posner A lot of companies will collect insights, but you need to send their data to them. How do you collect data from your companies that you work with?
34:28 Thomas Kolbabek In any way they want to. So we appreciate that games are different, that platforms are around. So again, I was involved in building platforms that lasted two decades. The backend lasted two decades, the frontend, naturally not, which is also learning from being in the industry for long. So we started with Java applets, by the way, which are gone now from the internet and Flash. But the backends remain. So you have to be appreciative of the backend. It’s battle-tested usually. So we actually want data access. We don’t want data integration. We don’t need specific APIs. We have them and we can use them. But we would take the data in any way, shape or form that it is and do then the heavy lifting, which is the data engineering. and data storage and everything else, because typically the data you need to make predictions and to use machine learning models is different than the data you need for your financial auditor anyhow. So the more you monetize, the more your database probably looks like a balance sheet. And the more you focus on your game, the more it might look like your game or your game mechanics. But again, that’s a different access point. So ideally access, which can be a data mart or something else, and it can be files. So we get CSV files, we get event streams, ideally, of course, real time. So real time is always superior, but it has its challenges in terms of consistency and processing speed, naturally. But nothing you can’t handle in 2023. So that used to be an issue in 1998, but not anymore. And yes, of course, you have to consider costs and you have to build it in a way that it doesn’t kill you, or at least you don’t make the usual suspects even richer. But so it’s just data at the end of the day.
36:12 Greg Posner Yeah, it’s interesting with the API, right? Sending data real time, you know, the question is, are your insights coming back real time, which I think real time data is really awesome to see. And even then, you so you get your results real time, how actionable can they be? Right? That’s the other thing is just to consider like, what, some data is probably more valuable real time than other data, right? So being able to understand what’s my concern right now? What do I want to measure? What am I looking at from my player base versus I want everything right now?
36:39 Thomas Kolbabek Yeah. And which action do you want to take? So yes, if we get real-time data, we also give real-time feedback. So real-time predictions or real-time interaction recommendations. Because otherwise, what’s the point of sending real-time data? There can be daily data, there can be weekly data, you can have a hybrid. So you send data daily, but then you act on it based on events. So you have some sort of lazy load actually, if you want to engage, it depends on the use case again. So if you want to improve session one retention, if you want to call it like that, so even day one retention, but session one, so first game start, play as long as possible and get to know the game, then you need real-time data for that use case. If you want to improve retention of players that are with you for a year, you don’t need real-time. They will eventually come back, usually. And another advantage of predictions and recommendations is you can actually, with a high degree of certainty, act before a player leaves. So you don’t act after 30 days or after 14 days when you haven’t seen the player, but you act the day before they leave, most likely. So you might not hit all of them, so you might not get all of them right. But you can at least target 60, 70, 80% of them right. Obviously, the earlier, the less accurate it becomes, but this is also an important learning. So accuracy and other things are very often discussed, and especially in a sales process, those might be emphasized, but they ultimately don’t matter. So the impact matters. If I have something that improves retention by 10%, No one cares how accurate it is. Of course, as long as it doesn’t do anything else, any other harm or uses more budget, but if the net effect is plus 10%, it doesn’t matter whether the model is 80% or 90%. You still use it as a metric to assess models, obviously. You want an accurate model at the end of the day, but it’s not the primary factor if you run a gaming operation. It’s of course different if you do something in the real world, like I don’t know, running a factory or making a medical decision, then it’s a different story. You don’t want to just wait on the output, on the outcome. But that’s the beauty about games. You can actually track and review them very nicely. You can use control groups, so control and target groups. So yeah, it’s actually the perfect environment to utilize AI.
38:51 Greg Posner It’s crazy to think that. you can predict when I’m going to stop playing a game. And it makes sense, right? So if we think about kind of the lifecycle of a player, there’s user acquisition, there’s player retention, and odds are that eventually a player will churn after enough time. And if you can identify when they’re going to churn, There’s a lot you can do, right? You can send notifications, you can give them rewards, you can give them prizes. There comes, I’m going to skip a little bit ahead here, kind of the ethical considerations. Like, I’m thinking about the movie Minority Report, if you’ve seen it, right? Like, trying to create or trying to prevent a crime before it actually happens. Are there ethical considerations with this, and maybe not in that aspect of it, but in different areas that you see your customers talk about?
39:36 Thomas Kolbabek Yes, of course. Once you come to larger companies, you also have a discussion with the data protection department or with the legal department. So what is key for the whole process is for any more, let’s say, regulated software development, let’s consider banking or medicine or whatever, you have certain standards that you have to uphold. So you have to make sure that the code can’t be tampered with, or at least not by a single person. You have to know that if it was tampered, you have an audit log or a proof that it wasn’t tampered with. That’s particularly difficult with databases, by the way. So your DBA can theoretically change your balance sheet. or change your cancer diagnosis if that system wasn’t implemented in a proper way and you don’t want that. You also don’t want anyone doing favors to his friends or her friends. So if Greg plays, he’s always amazing. You don’t want that, obviously. So this is the base consideration that you have to deploy it in a way that is auditable, if need be. And then you also must not exhaust the player. So there are freemium games and there are even examples that end up in front of court. So actually, in an Austrian court, there was this legal dispute over the loot packages in FIFA. which I think Blaster checked it was twice ruled against, so that they’re illegal, they’re considered gambling. And so monetizing too aggressively and monetizing on, let’s say, the 1% of the high spenders is not sustainable in the long run. So even if someone wants to spend tens of thousands of euros or thousands of euros for playing some match three game or some other game that probably you would never pay that much upfront. You shouldn’t do that because it will, in the long run, ruin the player, either financially or also in terms of time. A player invests resources in the game, time and money, and you can consider those and you have to consider those when doing models. The goal is not to over-engage a player. It’s always a fine line. How do you call it? The razor’s edge between making the game too boring for a player and too exhausting. So if I as a kid played too much and did my homework bad, my mom told me not to play anymore until my grades got better. And that doesn’t happen with adults. So it’s particularly important that the models also consider that. You can do this actually quite easily by assessing the potential spending power in zip codes. So you can get databases where you know what the average person earns in a certain zip code. So not getting personal data, of course. So we never use personal data, but you can use demographic data and commercial data. to actually identify baselines. Obviously, someone, I don’t know, in San Francisco, off the top of my mind, would probably spend more potentially on the game and can spend more than someone in another part of the world. And you can and you should consider this. Naturally, and ultimately, those limits are to be enforced by the decision of the operator. So they define the boundaries and ultimately regulate them. So we also see, or I would expect that eventually also games are regulated more. Because people are spending more and more time on it. The same might apply to social media at some point, scrolling through feeds. I think we talked about it. So I easily get lost in LinkedIn feeds, which is not good for me. Radical. Yeah, yeah, exactly. I just want to look at the contact and suddenly it’s 20 minutes ago. So it’s key. And then following that transparency in the process, hence this process of hypothesis, analyzing data, presenting results. So until you can actually then deploy something, it’s still a process that you want to review closely. And then you want to deploy it with a control group in place. Like you do medical studies, to have the analogy again, you use a placebo group and you use a target group, maybe with different doses. So you might use a 50 milligram, a 100 milligram, a 200 milligram group, and you separate your client base like that. And then you assess whether what you do is really effective. So maybe the AI makes it worse. And you don’t want to do that. So either makes players spend more worst case, like more than they should, or maybe even makes them leave the game, which is what you don’t want at all. So the key is to make the game more engaging, more entertaining within the limits of a player. And also knowing this is crucial.
44:11 Greg Posner You keep talking about AI, and I like the ethical considerations there. It makes sense. One of the things that you brought up and kind of blew my mind was the idea of looking at zip codes and understanding, hey, what’s the average spending on a zip code? Internally, we believe, from the data we see, Japan is the highest spending country when it comes to actually spending money in games, but that doesn’t break it down by zip code. So being able to use that data and understand that, and I think I’m going to start talking about AI in there and saying, Hey, what can AI help assist with this data? Again, I don’t know, again, where we’re talking about ethical or not, because whether you own the model, or you don’t own the model, but do you see AI as being a big stepping stone here to help allow us to understand different insights from within our data? Or do you think it’s a
44:56 Thomas Kolbabek Yes, because it allows you to add all these different dimensions. So whenever you find a dimension that you want to explore, AI makes it possible to just add it to the model and then to see the effect of it. So depending on the model and the algorithm, you can actually find out which parameters or which input parameter defines the output most. So some parameters have a high impact on the output and some don’t have any impact at all. which might also then be data that you don’t want to collect anymore, both for performance and for privacy reasons. Even though, again, we don’t collect personal data and we don’t want it, because whether you’re Greg or Thomas, it doesn’t matter. Your gameplay matters, which is crucial, especially in Europe, GDPR, and all the other considerations you have across the globe, rightfully. So the beauty is it’s not about having a discussion, having a meeting. Can we now introduce this into our system? Where do we have to store it? Who has to make the decision? Then we make another meeting to make the decision on how to actually use it. Then we need to present the rule, a study, and so on and so forth. So this whole process in the classic sense for multidimensional data is very cumbersome and just very long. Lots of meetings, lots of documentation. Having just that input data into a model and then assessing the outcome mathematically makes this whole process much less biased, assuming, and that’s very important, that the data is not biased. So if you send me data of, let’s say, just 30-year-old males to train the models and then use it for another game that’s maybe played by 45-year-old females, you must not do that. So you always must send a representative amount of data for the player group. You also must be aware that, for example, if you have a game in the US and suddenly you market it in, as you said, Japan or Turkey or somewhere else, then maybe your models don’t apply as much anymore. Maybe some do, maybe some don’t. But again, the beauty is you can measure things like data drift. So the models actually have metrics that you can monitor continuously to identify where the underlying data structure or data shape changes. And then you have to retrain the model. But again, all of this is what summarizes machine learning operations, which is a beast in itself. So creating a model is one thing, getting data is one thing, running it in production is one thing, actually monitoring it and keeping it up to date is one thing. So all of this together creates the magic that something automatically works in an AI process.
47:30 Greg Posner As not your target person, because I don’t develop games, a lot of this sounds intimidating to set up or get started, right? Am I overthinking it? Is it, is it depending on how you set up your game? Like, well, one of the, I asked my team, like, what are some questions that you find interesting and kind of, they said, what would be the top five data points that you would want to collect if you were starting your indie company? Right. And again, it’s, is, is that something hard for me as a developer to kind of set up my feed to you? Is it like you said, just tag what you want and export it. But, but I guess both of those, what would be the top five you’d recommend if you were going to start a company that you’d want to capture? And maybe it’s not stuff we even talked about. I mean, just for our listeners, things that we haven’t discussed on this podcast that are cool things that you can get are things like when you’re starting a game, right? Where’s your player in the world? Is it going to be something that scares people away? If your buttons are over here, is it going to be like Golden World as a whole? Just take a look at it. I’m not going to say this right, but like the mentality of a person and how are they going to absorb all this when the game actually starts? I think back to Mario 64, right? How like it just plops you in the middle of nowhere. You’ve never played a 3D Mario before. What now? And you just kind of go from there. And probably Starfield is probably quite similar because there’s no tutorial. And what would you come back and say about that? So Golden Whale does all this cool stuff. And I’m sorry, I’m jumping all over the place. You can plug that a little more if you want, but it comes down to, again, what would be the top five things you would recommend if you were going to do it that you’d want to capture? And is it hard?
49:06 Thomas Kolbabek Okay, there were multiple questions. So I think it’s like what to collect, how to present it to the user, so that it neither alienates nor doesn’t make any sense, and how to deal with probably onboarding in open-ended games. So collecting data really depends on what you need. So again, as I said before, if you want to focus on session one retention, you probably want to track everything from the install of the game and ideally even before that. So the less time you have before opt for looking into this one event. So this, why do players leave after they arrive? Do players uninstall after the first try? Then you need as much data as possible in the process before. So track everything in terms of your marketing campaigns, pass something to the game if you can do. So technically and legally, of course, because there are tracking laws in place. Again, depending on the country. then track literally every hit in the menu, send everything to us with a time code so that you also have this cadence sequence and literally every single data point in real time. So this is one use case. Another one might be you have a game already that’s most games or almost all games have a certain peak at some point. So you usually have one hockey stick in your lifetime in a product. That’s amazing. But it’s rare to have two or three hockey sticks. So usually all the games have this cycle of being peak use and then having a long tail and then how large this long tail is. That’s where I finally learned why differential equations make sense in school because the area under the curve is the important part. And that changes a lot if you just change a bit of the functions. So it’s under your control how big that long tail area is then back then, and also how long then your long term monetization is. And coming back to the very, very like 10 questions before, some games just choose to just create the next game, and the next game, and the next game, to adjust for that fall in demand essentially. But if you have such a game that you have released, you had a lot of budget, even you made a profit, but why not create data points to then retain players longer? And maybe not even by changing the game, but just interacting with them inside the game. And then can be things like game start, starting a level, losing usually recording all the good and the bad experiences. So having an amazing win, having a really, really bad loss, streaks, but then they can all be then reconstructed off these peaks. So let’s say peaks and valleys then, but again, that very much depends on the game. Also performance data, that can be frames per second, that can be delay, latency, so anything that involves a new gameplay experience. So the closer you can mimic gameplay experience, the better you can judge how someone feels inside the game. There have been companies, by the way, where you can track this with and where they combine game tracking with and EEG. So scanning brainwaves. Obviously, this is more for development purposes. So scanning brainwaves while playing to assess how you like a game. But you don’t want that in real life, obviously.
52:26 Greg Posner Fascinating. Try me.
52:27 Thomas Kolbabek Going too far. And the other question was then, so which data points? What was the other one?
52:35 Greg Posner Would you be capturing? But it sounds like it’s also dependent on what your use case is.
52:40 Thomas Kolbabek And then there was a second question that was sort of tied into that.
52:43 Greg Posner It was the data capturing and then is it complex to get started? How do I actually choose to send that? Where does click data exist in a game?
52:57 Thomas Kolbabek everywhere in the menu in, again, click data in the sense of that you interact with the game UI. Yeah. And that can be a key press. It can be a mouse move. That can be a touch.
53:08 Greg Posner I played this new match three game today, right? Like you had this tutorial and just banging the forward button, but it doesn’t actually do anything. Like even though it doesn’t do anything, are they capturing those?
53:19 Thomas Kolbabek they should in this case. Yes, they should because they actually annoy you and they should know that they annoy you. And they should really close it and never show it to you again. But again, it depends. Maybe this particular game takes a lot of pride in their onboarding process and they wanted to go through it. And this is also maybe that ties into that. So how much do you want or how much should and can a game change. Also like the onboarding experience, like you mentioned in large open world games, it all depends on what you want to achieve. So you can have a very, very fixed opinion about your game. So this is my game, like a movie in the end. So it’s my game, you watch it, you like it, or you just leave, or you leave in the middle, or you go and get three beers and still watch it. I had that once with Open Water, I think it was called, like watching people swim in the ocean for like two and a half hours, which wasn’t my type of movie, but I couldn’t change it. I just sat there. At least in Austria you can. I don’t know if it’s possible in the US. But you have the beauty of games can interact with the player. But fair enough, you might want to create a game that is what you as a developer wanted, or what maybe can’t or shouldn’t be changed because it’s a very competitive multiplayer game. So I would hate a Quake 3 arena if that suddenly makes the opponent faster. I would really hate that. I would really, really hate that. I’d rather have them matchmaking better. or a mode where I can play with a handicap to get more players. Like in the 90s, it was really difficult to get a proper Quake server with proper players. It was either boring or good. But nowadays, most games have enough audience. And then you might choose to have an opinion about the game. That’s your default settings, so your default balancing, default game properties. But then you allow for a certain bandwidth that the game can adapt. So you can either go this way or that way, and you can change those levers. And you could have 200 settings of a game, again, multi-dimensional, that you can then optimize for every player. I think I sent that to you. There’s research from Tomb Raider about different player types. So players that enjoy more like the riddles, jumping, or the fighting. And there’s even research papers on classifying those group of players. So players enjoy the game, but they might enjoy it for different reasons. Also, I’m now playing, I think another guest of yours said that once, so me playing Quake today would be different than me playing Quake 20 years ago. So I would like to probably enjoy it more, or like immerse myself more. Back then it was like kill, kill, kill, or get most headshots in Unreal Tournament to get that. to get all these funny voices and sayings and recommendations there. And that’s also the point. You can’t change a movie like I watched, what was it called? About an oak. I think The Heart of the Oak, which is a movie, an hour and a half, about watching an oak grow for a year. Sweet. Yes, but that’s a very different movie to what’s the other one, Bullet Train. So I enjoyed both of them, but they were completely different experiences. Bullet Train is just Kill Bill on steroids. I liked action movies, but this was beyond for me. But I also enjoyed watching The Heart of the Oak with my daughter, because it was just almost like meditation, slowing down and immersing myself. And at different days, I might have a different tendency to enjoy it. the game and the game can be different if you want to, but the movie can’t. So if I watch the Heart of the Oak with the wrong audience on the wrong day, I might really hate it. I really recommend that movie if you want to take the time.
57:06 Greg Posner So we’re coming up to an hour here, right? And I appreciate your time. I want to ask one last question, maybe two, we’ll see where it goes. The first is something I found interesting is I have a lot of friends that work in film and since they started working in film, they don’t love going to movies because they nitpick everything, every shot choice, every angle. And now that you’re working on kind of the insights on the gaming world, do you look at things and wonder why they made these decisions? Does it frustrate you? Or do you still have the pleasure of turning that side off and just playing a game?
57:41 Thomas Kolbabek That’s very easy for me because as I’m a data-driven person by not even training, I think I’m born like that. So my typical saying is, if it has colors, I don’t care. So I try to opt out of anything relating to logos and color decisions, flyers, business cards, whatever it is, whatever you need. Even in a data-driven business, you sometimes have design decisions. So I actually then enjoy that part even more if it’s done by other people. So I’m not able to create a game. I’m not able to create artwork or to think about a game. I might be able to do it if I don’t have anything to do for a while, so I have peace of mind. But I enjoy the data part way too much. But the beauty of it is, as such, I enjoy then the visual aspect and the artwork aspect of it even more. And usually, if I play a game, I’m not thinking at all about the steps behind it. What annoys me, like you mentioned too, is when I get too quickly or too frequent interactions on websites or games. Like, classic thing is you open some blog article, and before I can even read it, someone wants me to sign up to a newsletter. twice. And there is cookie consent and 15 other boxes in between. So if I’m in a rush, I might just close the page. Or in a game, when I play it for 10 seconds and I get, as you said, season pass, day pass, whatever. So that’s really, really annoying. And also, I seriously enjoy the data aspect of it. And by the way, this is also important. So as I said before, all these Data is something that can be objectively judged. It’s math at the end of the day, statistics. So this is very often very painfully sometimes discussed by creative people or people with decision power and an opinion. And then very often driven by the strongest opinion. But I’m never my player. I’m not my customer. So even though we built the data product, the data platform, I still have to listen and be open-minded to my customers because I’m not a game operator. I’m not a game designer. And equally, I often witnessed that there are discussions around everything from colors to gameplay to game balancing done by people that never play the game. Obviously, you can use focus groups, you can use interviews. There are loads of techniques to do that. There are also people that know it, that want to use a certain style. You wouldn’t do music only with AI. I wouldn’t tell Metallica with data how to play. They should just play. And they should clean the ender, please.
01:00:19 Greg Posner Exactly, right? Together, build this whole picture, but individual pieces. You might be an expert and say, hey, this piece should be changed, but unless you know the whole picture, it’s a piece of art.
01:00:29 Thomas Kolbabek Yes. And at the end, this artwork is the starting point. So you often ask, what is the beginning? The beginning is not data. The beginning is the idea of a game. the concept of a game, the first artwork, the process, the flow. And once you track it, you can then not spend time on discussing the flow, but steering the flow automatically with math. While you then can take all the time that you won by not having all these meetings to actually make the game better, or to make a new offer, or make a new level, or make a new car, or make an editor for user-generated content. whatever else is not there yet. Because what’s not there, you can also not optimize. So I cannot generate a unique new game for you. I can tell you how your current game runs better. Second question.
01:01:15 Greg Posner I appreciate that answer. It was very, very full.
01:01:19 Thomas Kolbabek And also, because I think there’s sometimes like a, like an almost like a love hate or like a, like a red team, blue team mentality between date, like more data driven people and creative people. So it’s not either, or it’s the combination of, it’s like the sales team needs the product team and the product team needs the sales team because you want your games to be played. and you want your players to continue playing your games, but you also need the game to have the data. So it’s not an either-or, it’s together.
01:01:48 Greg Posner And it’s kind of like, we don’t have to dig too much into this, but the Unity news, right? And we’re filming this for our people listening a week after Unity announced their new pricing. And It’s like the figureheads at the top of Unity knew or didn’t know what this impact was going to do. And everyone down was like screaming at the top of their lungs, like, no. And I mean, the top people definitely knew what was going to happen. They just want to try and cash out. But they have no insight into what’s actually happening. They just think they’re executives. They know the best choice, of course.
01:02:17 Thomas Kolbabek I’m not privy of any details on that and also don’t follow it in detail, but again, since I’m in there for 23 years now, so I’ve seen them come and go. So Unity is of course there to stay, but I’ve seen, especially in the mobile heyday, so 2010 to 14, there were so many companies offering free forever services, like extensive ones, like mobile ad tracking, monetization, whatever. Obviously, most of them venture funded. who had some sort of second agenda to eventually monetize it, but that never materialized. And we had multiple game development companies that built their products into tracking on those free tools, which then suddenly went bust or suddenly became very costly because suddenly they just did a free tier. So essentially, if something is free or cheap, it’s usually just a stepping stone to monetizing it eventually. That’s why we were always weary, but also back then there was no option to use like way too cheap or too good to be true looking services, even though they are tempting in the short term. Let’s see what comes out of it.
01:03:24 Greg Posner So second and final question is I usually ask what your favorite game is. I’m going to ask what your favorite gaming experience has been. Maybe it’s different. I’m not really sure. But do you have a favorite gaming experience or not? It’s a weird question.
01:03:38 Thomas Kolbabek I think it was for me, it was probably Ultima 8, I think it was, because it was so freaking hard and confusing. Like you mentioned Starfield, I haven’t played Starfield yet, but Ultima back then was to me like this open world game. It didn’t tell you where to go next and you got almost no hints. I played this freaking game for a month and never got to an end. really stupid buggy jump-and-run thing. So it was a really hard role-playing game, but also it was a really buggy jump-and-run. I think it was Ultima 8 Pagan or something, like it’s 90s game. I probably have to look it up. I banged my head against the wall so many times, but then solved it eventually, that this sort of probably was the most interesting gaming experience afterwards. So again, I hated the process. This game actually made me stay up late. And then I started actually getting up early with an alarm clock because in the morning I was better at playing, like sharper. So I got up before school with a clock and started to replay the parts that I got stuck at to then usually get the breakthrough in the morning and continue in the evening, which at the end then also prompted my parents to tell me to do my homework first. But that probably is the one that I remember most. But I played literally everything in the 90s and early 2000s. And now it all comes back because my son is old enough to start playing games with me. And I also now like to deliberately play games either first or with him to just see how they evolve. So some games really look nice, especially in the ads. look like proper games, but then eventually turn out to be almost casino-style games. And that’s not appropriate for children and also not appropriate to market it like that. But again, there is so much offering, so much monetization, and it’s probably a matter of taste, you could argue. But if you have something where you press a couple of buttons and around it, coins fly, then yeah, that’s not a children’s game.
01:05:43 Greg Posner Thomas, I enjoyed you coming on. I feel like there’s still a whole bunch we can talk about, and I’m sure our audience will have some questions as well that we can follow up with. But is there anything else you’d like to talk about or plug or talk about for Goldenwell that the audience know this time is yours?
01:05:59 Thomas Kolbabek No, thank you. Thank you for listening. And if you want to get in touch or cry murder about data-driven game development or dynamic games, you can reach me on LinkedIn. You find our page at goldenwhale.com. You can find me at tk at goldenwhale.com. So feel free to reach out with good or bad. A good discussion is always nice. I think We can agree to disagree, it’s also a good outcome. And yeah, I’m just happy to, I’m really happy to work in that area. And as a data-driven, not artwork-interested game person, it’s my dream place at the moment. Awesome. Thank you for listening.
01:06:37 Greg Posner Yeah, thanks for having it. And like I said, next time I play a game like Call of Duty and I’m matched poorly, I’ll scream at Activision to go check out Golden Whale. But I think this has been really insightful. I’ve learned a lot, and I appreciate you coming on today. And as Thomas said, we’ll have all his URLs, his links, Golden Whale’s on our Player Engaged podcast website as well. So Thomas, thanks again for coming in today. I know it’s late your time, so you’re free to go back to bed. No, I appreciate you coming on. I hope you have a great rest of your night. Thank you. You too. Good night.
