grokludo – Détails, épisodes et analyse

Chris Simon is a technologist who's given talks about AI, specifically LLMs, or Large Language Models such as ChatGPT. 

AI is going to be an increasingly big topic in games. From things like art generation, code generation, to chat moderation, to dynamic difficulty systems, and all the way to engineless games that use a neural network to generate images based on user input. And there are many more.

So it's a topic we'll probably return to, but one of the biggest ways people are using LLMs is in story generation and dialogue. The idea here is that one could chat to an NPC indefinitely, or let a robotic Dungeon Master do all the work.

Chris Simon says it's not so simple. And I'm kind of glad he's the first person on grokludo to talk about AI. Because just as many games and NPCs will use the enterprise models as a foundation, all our future AI chats will use this interview as a foundation, about the hidden costs of AI in games.

Timecodes:
00:00 - Intro
01:45 - Epistemological statements
06:12 - Can LLMs tell infinite stories?
16:42 - Are locally hosted LLMs ethical?
18:00 - Harmful human reinforcement training
21:30 - Inherent biases in LLMs
24:45 - Jailbreaking - intentional and otherwise
30:00 - The Sycophancy Problem
33:10 - How biases could seep into games
39:40 - The silly and the tragic ways to break LLMs
42:00 - Human brains vs neural networks
48:48 - Who wants a regression to the mean?

Follow Chris Simon:

https://chrissimon.au/
https://bsky.app/profile/chrissimon.au
https://www.linkedin.com/in/chrissimon-au/

grokludo website:
https://grokludo.com/

YouTube:
youtube.com/@TheJunglist

YouTube Playlist:
youtube.com/playlist?list=PL3cxb6pzUE9Z7PXjIVbmr5qiohoq5wTdo

Follow on X:
https://x.com/TheJunglist

Follow on BlueSky:
https://bsky.app/profile/thejunglist.bsky.social

Minh Le, also known as Gooseman, is best known for creating the world-conquering Counter-Strike. It started out as a small Half-Life mod, and now sells out the largest stadiums around the world as one of the greatest esports of all time.

In 2006, Gooseman left Valve to make Tactical Intervention. Since then, he's worked on Rust, Black Desert Online, and now he's got a new game, Alpha Response, marrying his traditional semi-realism with co-op PvE heavily inspired by Left 4 Dead.

I never miss a chance to talk to Goose, and since we've talked before about tracking the genealogy of game mechanics on this podcast, I also wanted to quiz him on the industry's trend towards static recoil patterns in gunplay, which Counter-Strike popularised.

Timecodes:
00:00 - Intro
01:15 - Alpha Response
03:06 - Challenges in retaining players
06:39 - Stop Killing Games
11:31 - Taking inspiration from Left 4 Dead
17:18 - The style of game Gooseman likes to make
19:10 - Playing and watching modern CS
20:50 - FPS moving towards static recoil patterns
25:35 - Leaving Valve, making artistic leaps

Alpha Response is in Early Access now on Steam, and you can get it now.

Get grokludo in your inbox at grokludo.com, or catch it on the big podcast platforms!

Margaret Anderson became Director of the Australian Classification Board in 2013, a time when it still made opaque decisions and wasn’t prepared for the tidal wave of gaming content that would come in the following years.

She talks to grokludo about declassifying the classifications as it were, and dealing with the multiple challenges that games created as a fast-moving technology that vastly outpaces the laws written to regulate it.

On the way we cover Australia’s debate over whether or not to have an R18+ rating for games, the quagmire of loot boxes and gambling content in games, and some fun stories about what it was like at the Board in these big moments.

Chapters:
00:00 - Intro
01:00 - How games are rated differently
12:00 - The R18 Rating Debate in Aus
14:58 - What Margaret wishes people knew
20:37 - How classifying games works
25:08 - Publishers changing games after classification
28:47 - Diversity is the Board’s strength
34:30 - Watching disturbing content
38:40 - Funniest thing the Board was blamed for
41:08 - Margaret hates the C word
43:45 - Defending anime to the Aus Senate
46:43 - How do we get meaningful change on loot boxes?
56:31 - What responsibility do industry bodies have around loot boxes?
1:00:20 - An idea for ratings based on types of fun
1:05:08 - What Margaret misses about the board
1:08:03 - Today’s Classification Board
1:09:08 - Prisoner’s Aid NSW

The latest Digital Australia report by Jeff Brand:
https://igea.net/wp-content/uploads/2021/10/DA22-Report-FINAL-19-10-21.pdf

Prisoners Aid links:
prisonersaidnsw.org
matesonthemove.org

Follow Junglist at @TheJunglist

grokludo.com to subscribe and get emails!

Paul Tozour is saying exactly what the industry needs to hear right now. Backed with data from his 2015 study titled The Game Outcomes Project, he's fictionalised the data in his new book The Four Swords: A Parable of Leadership, Video Games, and Dead Dragons.

Tozour goes through his lessons for studios, publishers, managers, and creatives, able to definitively describe what leads to a successful studio and point to the data that proves it. In this moment of post-largesse layoffs, these lessons are even more important.

And there are plenty of wacky stories from his time in gamedev to boot!

The Game Outcomes Project Part 1: https://www.gamedeveloper.com/business/the-game-outcomes-project-part-1-the-best-and-the-rest
The Game Outcomes Project Part 2: https://www.gamedeveloper.com/business/the-game-outcomes-project-part-2-building-effective-teams
The Game Outcomes Project Part 3: https://www.gamedeveloper.com/business/the-game-outcomes-project-part-3-game-development-factors
The Game Outcomes Project Part 4: https://www.gamedeveloper.com/business/the-game-outcomes-project-part-4-crunch-makes-games-worse
The Game Outcomes Project Part 5: https://www.gamedeveloper.com/business/the-game-outcomes-project-part-5-what-great-teams-do

Paul's series on decision modeling:
https://intelligenceengine.blogspot.com/2013/07/decision-modeling-and-optimization-in.html

The Four Swords: https://www.goodreads.com/book/show/195019998-the-four-swords

Follow and support grokludo at grokludo.com!

Follow Junglist at: twitter.com/thejunglist

Malcolm Ryan from Macquarie University speaks about his study that shows how suggestible we can be if a game's morality meter tries to nudge us in a certain direction... provided certain conditions are met beforehand.

Even when players think they're ignoring the morality meter, the results say different!

Read the study here: https://journals.sagepub.com/doi/full/10.1177/15554120211017040

Play the game, The Great Fire, here: https://moralityplay.itch.io/the-great-fire

Ryan et al's paper on the four-component model of moral psychology can be found here: https://press.etc.cmu.edu/file/download/924/b557cd42-6151-4ecf-a8f8-d2c18bdfd27c

Follow Malcolm Ryan's work at Morality Play:
https://moralityplay.org/

Follow Junglist:
twitter.com/thejunglist

Sign up at grokludo.com to get new episodes in your inbox!

Raph Koster has helped forward our understanding of game design for decades. 

He's the author of A Theory of Fun, a must-read for game designers. He was lead designer on the pioneering MMO Ultima Online, and led the creative team on Star Wars Galaxies.

Now he's back with a new MMO called Stars Reach, built on 3D cellular automata system that simulates everything in the galaxy. The game's not out yet, and already the stories emerging from this kind of simulation are hard to compare with anything else.

So strap in, we're going pretty deep into game design theory - that's what grokludo is for! - and in the second half, we cover what to expect from Stars Reach.

Timecodes:
00:00 - Intro
01:05 - Most games are copies of others
02:58 - What videogame designers can learn from board games
06:18 - Separating mechanics from aesthetics in policymaking
13:00 - Different kinds of randomness
16:05 - Kriegsspiel, Ur, and other early games
25:50 - Do modern games satisfy modern compulsions?
31:20 - Games criticism focusing on aesthetics
34:55 - The art game movement of the early 2000s
39:15 - Using ludo-narrative dissonance as a tool
43:20 - The Pascal's Wager in A Theory of Fun
47:20 - Don't get Junglist started on Dark Souls
50:15 - Do game design lessons line up with cognitive psychology?
56:00 - Studying government styles for MMO design
59:40 - The "designer's dream game" trope
1:05:50 - Stars Reach's cellular automata system
1:13:13 - Sustainability despite player terraforming
1:18:30 - Solving the problems of biology
1:21:50 - Terraforming and PvP
1:24:20 - Scaling the galaxy with adaptable sectors
1:27:45 - The dangers of fire and water

A Theory of Fun:
https://www.theoryoffun.com/

Stars Reach:
https://starsreach.com/

Sign up at grokludo.com to skip the algorithm, or catch it on Apple and Spotify podcasts!

Kirk Sigmon is an attorney at Banner Witcoff , and an expert in patent law.

He joins us this week to make sense of the wild, and complicated situation in which Nintendo and The Pokemon Company are posturing towards US litigation against Pocketpair, the maker of Palworld. Litigation in Japan is already underway.

In the process, Nintendo has secured patents for game mechanics that look a lot like what we've already been playing for the last 30+ years.

There's patent activity relating to mounts and the quick-switching of mounts. There are patents about throwing an object to capture a character. And very recently, a new patent was secured that covers the summoning of a character.

This is a complicated and niche field, so we first cover some basics of patent law, before getting into the dirty details of what's been called "by far the most aggressive patent enforcement any game maker ever attempted against a rival."

00:00 - Intro
01:42 - What good patent law looks like in games
05:58 - How often do mechanics get patented?
07:50 - Nintendo pursuing patents after Palworld's release
14:00 - Reading patent claims is torture
16:34 - Nintendo curving patents towards litigation
23:22 - Patenting a summoning mechanic
26:35 - Limited resources at the patent office
30:53 - Where has patent law done well, and where poorly?
33:30 - The Inter Partes Review system
37:24 - Why not pursue litigation on the art style instead?
41:48 - Attack on the mod space as "prior art"

Subscribe here or on grokludo.com to get episodes in your inbox!

Aaron Drummond and Jim Sauer are associate professors at the University of Tasmania, and recently released a paper looking at loot box spending measured against distress, when normalising for disposable income. 

The two have studied a range of issues in games, such as the effects of violent games on aggression, and the impacts of gaming on learning. But when they started researching loot boxes, things were very different.

Find Aaron and Jim's study here: https://royalsocietypublishing.org/doi/10.1098/rsos.231264

Timecodes:
00:00 - Intro
00:46 - Do loot boxes cause distress?
04:16 - Skinner box systems in games
12:00 - Disposable income and "whales"
18:39 - Why loot box research isn't just a moral panic
26:24 - The limitations of current research
31:47 - Violent games blamed regularly in killings
35:51 - Intrinsic and extrinsic rewards intertwined
38:40 - Final words

Thank you for watching and I hope you enjoy!

You can support grokludo and get updates in your inbox at grokludo.com!

Leon Xiao is an Assistant Professor at the City University of Hong Kong. After doing his PhD in loot boxes, he's released papers charting the loot box regulation landscape, and measuring compliance.

Coverage of his work has been picked up by mainstream media such as the BBC and the Guardian, and GamesIndustry.biz publishes his yearly loot box state of play report.

Today Leon gives us a zoomed out view of loot box research.

In the intro I mention Leon's thesis, which comprehensively covers loot box regulation and compliance. You can find that here: https://osf.io/preprints/thesiscommons/af8ev_v1

The last Loot Box State of Play can be found here - https://www.gamesindustry.biz/loot-box-state-of-play-2024-another-trip-around-the-world-of-regulation

Timecodes:
00:00 - Intro
00:48 - Leon Xiao's program
03:47 - Gaming Disorder and older adults
07:30 - What do we know for sure about loot boxes?
11:30 - Loot box movements to be aware of in 2025
18:10 - Low compliance to industry self-regulation
32:43 - Gambling aesthetics vs gambling mechanics
50:26 - "The science isn't in yet"
56:20 - What could researchers do with industry data?
1:04:42 - Comically obtuse disclosure of loot box odds
1:08:32 - Should we all complain to regulators?

Thank you all!

Make sure to subscribe here and at grokludo.com for updates!

Ejnar Håkonsen has studied the light and dark sides of player manipulation. He’s designed matchmaking and pay-to-win systems himself, and he’s theorycrafted in pay-to-win games with top-ranking whale guilds, to understand the strategies they use.

Years ago, he designed a multiplayer system that used manipulative practices for positive ends. It used commendations in kind of a genius way to reduce toxicity by weaponising a player’s ego against them.

He recently posted a Reddit thread about it which effectively open-sourced the design and had enormously positive feedback, which can be found here:
 https://www.reddit.com/r/gaming/comments/1m86pfb/my_job_is_to_psychologically_manipulate_gamers_as/

Although the game studio had to pivot away from that core design and the system was never implemented, game designers have been expressing their desire to implement similar systems.

Today, we talk to Ejnar about how his "friendship engine" system works, and how we can take that system in new directions.

If you're a game maker and start using this or a similar system, let Ejnar know! His LinkedIn is found at:
https://www.linkedin.com/in/ejnarh/

If you want to read the article mentioned about rolling with top-ranking whale guilds and how the maker of a pay-to-win game lost much of its community, here's the link:

https://www.linkedin.com/pulse/adventures-extreme-ethnography-how-i-got-inside-view-from-h%C3%A5konsen-a2nif/

Apologies for the audio issues in the early part of the podcast, I switched to my backup mic at around 8 minutes in.

Chapters:
00:00 - Intro
00:37 - Ejnar's history in hypnosis
07:53 - Joining BetaDwarf
11:19 - Getting investors to think long-term
16:14 - Why game designers intuit psychology
20:50 - How Ejnar's frienshipping engine works
32:15 - The parallels between hypnotism and game design
36:17 - The Brilliance and Perils of Heroes of the Storm
41:10 - Blizzard's shortsightedness in cancelling HOTS
42:15 - What kinds of loot box are toxic?
50:33 - A database of friends, PvP, and too much social proof
57:44 - How Russia's invasion and covid affected development
59:28 - The next steps for this system