Really interesting one!!

I'll be interviewing with him today!

I wish this was a longer video

@Reed your twitter link doesn't work anymore?

A modern version of Shandalar would be incredible. Co-op two headed giant mode would go crazy.

Played the crap out of shandalar, very stoked to hear Gavin mention he would like them to make something like that again

The father of Diablo himself. David Brevik. Bro, are we waiting for a new game from you 😭😭

Damn how tf did u get gavin, thats sick!

Very thoughtful insights! Very excited to read the book

Great talk!

HOW IS THIS TALK SO GOOD! Andrew had me laughing throughout. What a nerdy, thoughtful talk. I am never going to stop thinking about this. Roguelike Celebration is such a treasure trove.

OMG I love your web page name!

Gavin, make an awesome proposal of a roguelike version of Shandalar? Seriously, you can probably convince them about making one.

Gavin "play more lands" Verhey!!! Excited to watch this!

Whao, when he started talking about "capabilities"; it's like he was talking about Rust's trait system lol. Composition in general is far superior. And that "fireball-shooting sword" is a great example where the class doesn't adhere to its supposed hierarchical ancestry and thus inheritance breaks down, causing a lot of overhead and possibly a lot of bugs by inherited features that were never needed. Traits are much more superior.

12:23 The way I have been using word generators, having some ugly things pop up hasn’t been a problem. I’m using them to generate names(toponyms, ethnomyms and personal names) to use in human-generated or curated content. As I focus on a procedural planet generator for a game, I find the ability to fairly-reliably generate good sounding names is more important. If the game comes up with a name like Rliqwyzxd, I can’t just ignore it. I will have a mechanism for the player to rename planets, stars and such, if stinkers show up, but I’d like to minimize that. Mekipi may not be perfect(experience tells me it won’t be), but I appreciate that it was able to generate 60 names, that while I like some better than others, none are terrible. I can figure out how to pronounce all of them and none of them are unreasonably ugly. There’s a good chance I don’t pronounce all of them the way you would, but they all have reasonable pronunciations that aren’t hard to come to.

Sounds like you would hate nethack... too bad

I’ve heard all these stories before, but it’s always amazing to hear David talk - Diablo being one of my all time favourite games:) Thanks for the interview :)

I'm currently into godot-rust and I love it

which programming language is this ? i didnt understand it it looks like c# but it is not c#

leigh and brian, two of my creative heroes. so happy they're having fun with this awesome project

Sorry, but I looked away when Leigh was playing the game and casually talking about the various celebrities and who's doing what, and thought I was listening to someone's retelling of "i had a strange dream".

This helped with my project, thanks

It's funny; Diablo 3 and Diablo 4 are both much closer to being MOBAs or perhaps Gauntlet likes than the action-based roguelikes that Diablo 1 and Diablo 2 had been. I think the folks to whom classic action roguelikes appeal to and gauntletlikes appeal to are not perfectly intersecting populations.

This is so good lmao

I saw a sonority sequencing principle!

Loved the talk. Incredible. So weird I only found this after 5 years! Learned a lot. Thank you!

This is the wildest thing I've seen using Google Sheets. Very cool!

@18:10 While I generally like the command pattern---I've used it in a editor with multiple undo and it works great there---I think you have to watch out. As in his WalkAction command you can end up with a bunch of IF statements such as "if (actor is Hero)" and the need to query actors. You're trading off virtual functions for if statements; just slicing things a different way. It may end up being just as unorganized.

You could definitely have a game thats moreso focused on graphical immersion and mastery instead of experience, for example. Look at dark souls.

SUBSET GAMES!!! RELEASE A FULL GAME BY THE END OF THIS YEAR, AND MY LIFE IS YOURS!!!

Why when people think about roguelikes they think anky about permadeath and procedural generation?

this is where tarn met kitfox ?

i was in the talk. it was an awesome one!

You might want to actually look at game code from the 90s. They were doing composition long before you kids thought it was cool. This stuff isnt even remotely as new as you seem to think it is. 90s game devs were way better than most people today.

I found this fascinating, but it covered the ideas fast enough that I needed to watch several times and pause on each slide, before I felt like I understood what was going on with the technology. I'll try summarise: - Inspired by tarot cards and simulations like The Sims, Jasmine Otto made a narrative generator for murderous druids, where the druids would take random actions such as hurting each other. Those actions would be listed in order, for the player/querent to read through. - However, the large number of independently-acting druids meant it was hard to create meaningful plot payoffs. If one druid murders another druid, Jasmine wanted there to be a dramatic accusation for the crime. But I'm guessing that when you're waiting for one character to accuse a murderer, the random selection of which action happens next means you'd see unrelated characters' plots interwoven. It could feel messy if a setup and a payoff are separated by a lot of unrelated characters' actions. Jasmine implied that it was hard to control the structure of the narrative using this system. - They decided to stop generating actions randomly. Instead she started with a social graph (i.e. network) of character relationships, which has many cycles in it - for example, character A harms character B, and vice versa, in a loop. Each edge (i.e. relationship) in the graph eventually becomes an event, but they aren't listed in a random order anymore. Instead, the narrative instrument "walks" the graph by following edges. If A just did an action to B, then B would take an action next, and so on. - Eventually the walk would reach a dead-end, which is a character for whom all of their outgoing relationships have already been turned into events. At that point, the chapter would end, and the walk would restart in a different part of the graph with a different character, until all the graph's edges are used up. Each chapter would have internal continuity because it would reuse the same characters. You would also get narrative payoffs because of the cycles in the graph: every harm has a corresponding revenge, and since those relationships are near each other in the social graph, they'll usually not be too far apart in the walk of the graph. (I'm not entirely sure if each relationship can only turn into one event, or if it can become multiple. It's also possible that each chapter is restricted to a certain subset of characters to make it even more cohesive.) - The list of events (termed 'verbs') has a nice enough overarching structure for a narrative, but the detailed prose itself was generated by retargetable 'storylets', which are chunks of writing with gaps in them. The gaps get filled in by the elements of the verb (or perhaps by elements of a 'graphlet', which can have more than one edge). Storylets are picked randomly, and depend on the state of the story (i.e. the effects of previous storylets). For example, if a character has been killed, then from that point on, they can only participate in storylets that have a ghost-shaped hole for them to fill. - The storylets give the story a texture made out of randomised and contextual prose, even though the storylets originally come from a much simpler social graph which doesn't even store any state. - Further work might run this narrative generation process on multiple different scales. For example, at the high level it might generate a social graph of factions, but instead of turning that directly into storylets, it might zoom in on the individuals, generate a more detailed social graph for those druids from the factions, and show how the factions interact through individuals. That way you could give high-level faction interactions a narrative structure without compromising on the detail of individual druids. - Jasmine also looked into story-sifting (i.e. automated searching for patterns of events that a player might care about) but it was not in scope for the talk.

this video and the person explaining are golden!

thats was amazing! i feel so inspired right now

VERY useful.

required viewing

We're slowly getting there

7:16 bro was having vietnam flashbacks ON CAMERA thinking about physics... What things did he see while trapped at EA?!

My player objects ARE monsters... With a different controller implementation

The fans want multiplayer/coop ! THANK YOU TEAM ADOM !

Keep that lazy plagiarized garbage away from games. No one wants it.

total heroes

I really enjoyed this talk. It’s a good resource :3

I actually DO have a new found knowledge and appreciation. Thanks.

I love learning about the history of these games

10:20 "Baboonsboon"! 👌