Decoding Strategic Behavior in Large Language Models: A Game-Theoretic Analysis

Recent research delves into the strategic decision-making capabilities of leading large language models (LLMs) like GPT-3.5, GPT-4, and LLaMa-2 through the lens of game theory. By exploring their resp...

https://news.lavx.hu/article/decoding-strategic-behavior-in-large-language-models-a-game-theoretic-analysis

'Sup, everyone!

Any suggestions of resources for studying simulation, in a broad sense?

Is game theory and complex systems useful for developing cooler simulations or studying them would be overhead?

Where can I start learning about tactics and strategy?

"democrats, treat republicans they way they treat you and treat trump as though everything you said about him is true."
#democrats #republicans #bullies #respect #gametheory #resist #tactics #politics #standup #standupforhumanrights
https://www.youtube.com/watch?v=g1ACmdglOSA

There are some topics that just instantly generate endless debate. The Copenhagen interpretation of Quantum Physics is one.

In RPGs, D&D and such, I think the equivalent topic is Referee, DM objectivity and the use of dice. This article by Bob Kruger describes the issue a little better than I have been able to describe it in the past (and I have tried to explain it so many times without much success).

https://web.archive.org/web/20160520122457/http://www.baen.com/danddmasters

In the mid 80s, TSR came out with a new version of Top Secret. The original game was set more in a cold war USA vs USSR kind of setting. The newer game leaned more into a James Bond fantastical kind of setting.

I played it a couple of times. I really liked the combat of the game. Each character had a paper doll with hit locations and hit boxes. Players would roll 2 ten sided dice to see if they hit. 2 ten siders can make 00-99. If you rolled under your skill rating, you would get a hit.

But what was cool was that the same roll also told you how much damage you did, and where you hit. If Suzie had a Melee score of 67, and she rolled 58. She would do 5 points of damage and hit location 8 (right let). If she rolled a 20, she would do 2 points of damage and hit location 0 (head). If she rolled 78, she missed.

I always wanted to try such a system in a medieval style rpg, but damage and body types are more varied. I did not relish the idea of making paper dolls for Dragons, Centaurs, and Mermaids.

Recent discussions about hypothetical D&D economies led me to look into Roman currency. Here is a great wikipedia image of the common currency in the 27 BC - 100 AD Roman era.

So instead of copper, silver, electrum gold (the D&D standard), the early Roman Empire used various iterations of Bronze, Orichalcum, Silver, Billon, and Gold.

And today I learned Billon is the name for an alloy of silver and gold, or silver and copper, or silver and gold and copper, or basically any alloy of silver and some base metal.

https://en.wikipedia.org/wiki/Roman_currency#Imperial_period:_27_BC_%E2%80%93_AD_476

Credit to @capita_picat

for pointing out this article about how Gold was not the usual currency used in medieval times.

https://acoup.blog/2025/01/03/collections-coinage-and-the-tyranny-of-fantasy-gold/

I like looking at real world analogues to help inform making the "physics, economic engine" for rpgs.

A lot of issues in rpgs come down to how well do we want to emulate something from the real world like; falling damage, combat injuries, or should the economy use silver pieces.

Versus the thought of, are we playing a "game" and the rules should be more gamey in nature. Usually because going too far down the emulation path becomes a slog.

Recent discussions on RIFTS Mega-Damage led to investigating what is the AD&D equivalent. If you look on pages 54 of the 1e DMG, you get information about ship warfare and hull points which seem to be equivalent to construction defensive values detailed starting on page 108.

There is something about Gygax's system here that is not very appealing to me. Too fiddly? A lot of consulting charts? I am not sure.

If you take the seige weapons table, and average and compare the damage they do to normal creatures, and then you compare the siege damage they do to structures, 1 hull point and/or construction defensive point appears to be about 2 hit points. So just double.

But it is not easy (or correct) to make a 2 to 1 conversion, because every case is a bit different on the charts. For example, a fireball does no damage to earth or stone structures!

I am kind of warming up to the idea of Mega-Damage. If you read through the rational behind MD in the Core Rulebook it makes sense.

https://palladium-store.com/1001/product/800HC-Rifts-Ultimate-Edition.html

Quoting from the book...

It is crucial that players clearly understand the concept of Mega-Damage and Mega-Damage Capacity (M.D.C.). Basically, M.D.C. indicates a super tough armor or physical structure. A structure so tough, that normal weapons will not damage it. To damage a Mega-Damage Capacity (M.D.C.) structure you must use something that inflicts Mega-Damage (M.D.).

...

You're still sitting in your tank when somebody waltzes up and lets go with a .357 Magnum at point-blank range. The bullets bounce off the tank's armored hull, leaving only large gouges in the paint and scratched armor. No damage is done. His two buddies whip out an Uzi sub-machinegun and an M-16 assault rifle, and spray the tank with a hail of bullets. The combined attack must number into the hundreds of S.D.C. damage. But the tank's hull is undamaged because it is super tough . . . M.D.C.! Normal weapons, even when combined, can not damage a mega-structure. Only a weapon that inflicts mega-damage can harm the tank.

Thanks to dmdavid's excellent articles I have been able to articulate something more clearly that has lingered in my mind for years. You should go poke around
https://dmdavid.com

It is a wealth of information and insights.

I never warmed up to 3E. I played it a little bit when it came out in the early 00s. The latest article goes over how monsters needed to work exactly like players. It is not a terrible idea. But, what I do not like about 3E is it is not "DM centric". What I mean by that is, 3E has system after system to make a coherent game where it can work a bit like a computer. But as a DM, what I really want, is a system that makes the task of running a game "as easy as possible" but still coherent. It does not feel like 3E strives to make the DM's job easier.

Lately, I have been doubting the power progression of characters in your typical D&D game. Just seems weird that a level 20 human fighter can go toe to toe with a Dragon and possibly win.

On the one hand, I think one of the appeals of D&D is the epic power gaming that can be achieved. Player's feel like they are progressing in power as they level up.

Also, in Runequest it feels like you progress very slowly, so maybe my happy middle spot is somewhere in between D&D and Runequest.

Maybe the old Epic 6 had the right idea, where you max out at level 6. Although, I get a edgelord kind of vibe reading up on E6.

Maybe it makes more sense that for the players to be able to defeat a dragon in combat, they need to find the right magic items to fill in the power gap.

https://rpg.stackexchange.com/questions/97252/what-is-e6-why-would-i-use-it#97255

Personal opinion. The 1e, 2e, 3e DMGs do not do a good job of cluing in a DM on what a DM is going to do a lot of (which is adjudicate situations, from the mundane to the completely gonzo off the wall ideas players will come up with)

Gary Gygax writes in 1E about it, but it is on page 110. You had to read 109 pages before you got this bit of wisdom (Image 1)

David "Zeb" Cook might have written about it in 2E. There is some hint of it on page 38 (first print) but it deals with the optional non-weapon proficiences. (Image 2)

Not sure who wrote the 3E verbiage (Image 3), but it really hits the wrong tone for how I like to think about running the game. But at least it is on page 9.

In my opinion, adjudicating should be the 1st or 2nd thing the DMG goes over. And the DMG should provide copious amounts of advice on how hard the job is, and throw in some easy to use modifiers and ideas about how to mechanically adjudicate things.

I think STAR FRONTIERS released in 1980 does a much better job of giving a simple to understand mechanic that can be used to adjudicate most situations. (Image 4).

Following a #boardgames podcast, I've been thinking about different types of rules: formal, informal, written, unspoken, etc.

In particular the argument that in games the formal (sometimes written) rules temporarily replace the informal (social) rules of interactions. "You get to lie and stop your friends from doing what they want. Which you wouldn't do outside of the game." This is what some consider to be the magic circle.

Which I obviously don't agree with. My position is that the magic circle adds rather than subtracts from the norms and social conventions that make up our interaction. That it's the tension between formal game rules and informal social rules that we experience as escapism.

More importantly, resolving this tension by choosing which rules to follow gives us a sense of agency. So when Nguyen argues in 'Agency as Art' that players enter a kind of double consciousness of both caring and not caring about winning the game; he's missing the forest for the trees.

We don't do both. We choose which rules to prioritize from moment to moment. And it's this flexibility that makes games work on more than one layer.