Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
SuperKlaus
Oct 20, 2005


Fun Shoe


Goons? In MY game store? It's more likely than you think.

Adbot
ADBOT LOVES YOU

panko
Sep 6, 2005

~honda best man~


https://boardgamechat.com/

chatgpt-style board game rules bot. played around with it for about ten minutes and didn’t see any incorrect output so could be very useful

Anonymous Robot
Jun 1, 2007

Lost his leg in Robo War I

panko posted:

https://boardgamechat.com/

chatgpt-style board game rules bot. played around with it for about ten minutes and didn’t see any incorrect output so could be very useful

Can’t seem to handle Arkham Horror LCG. The answer it posed to an inquiry about enemy spawn logic had appropriate terms from the game and was sort of sensible, but was ultimately incorrect and also not a really salient reply to the question.

Those are pretty complex, conditional rules, though.

Serotoning
Sep 14, 2010

D&D: HASBARA SQUAD
HANG 'EM HIGH


We're fighting human animals and we act accordingly

panko posted:

https://boardgamechat.com/

chatgpt-style board game rules bot. played around with it for about ten minutes and didn’t see any incorrect output so could be very useful

this is badass. exactly the kind of thing bots are great at and people suck at: properly interpreting mundane rule

Archenteron
Nov 3, 2006

:marc:

panko posted:

https://boardgamechat.com/

chatgpt-style board game rules bot. played around with it for about ten minutes and didn’t see any incorrect output so could be very useful

Now combine it with that one goon's lists of GPT generated board games/descriptions

PRADA SLUT
Mar 14, 2006

Inexperienced,
heartless,
but even so
trip report: mind mgmt is goddamn great and I think hidden movement sucks

panko
Sep 6, 2005

~honda best man~


It's my duty as social justice klaxon to inform you that matt kindt minted one nft and as such is a capitalist envirofascist techbro.

So, if any of that matters to you, then maybe don't enjoy the board game.

(comic in actual legible format available here)

PRADA SLUT
Mar 14, 2006

Inexperienced,
heartless,
but even so
trip report: mint ngft is goddamn great and I think hidden movement sucks

Magnetic North
Dec 15, 2008

Beware the Forest's Mushrooms

panko posted:

It's my duty as social justice klaxon to inform you that matt kindt minted one nft and as such is a capitalist envirofascist techbro.

Hadn't seen this. Looks like it was in the initial gold rush. It also seems to inherit those initial understandings, such as this being a good idea.

Haven't heard about this in the space in a while outside Van Ryder. Even tech companies are running for the exit.

edit: vvv I understood you were probably making fun of me, but I also had sincerely not heard. I don't think minting one early in the mania is the biggest deal.

Magnetic North fucked around with this message at 08:28 on May 1, 2023

panko
Sep 6, 2005

~honda best man~


bobvonunheil
Mar 18, 2007

Board games and tea

panko posted:

https://boardgamechat.com/

chatgpt-style board game rules bot. played around with it for about ten minutes and didn’t see any incorrect output so could be very useful

literally the first thing I asked it is completely and utterly wrong

(7 Wonders) How many caravanseries are in the game with 6 players?
"There are 6 caravanseries in the game for 6 players"

What does a caravansery do?
"A Caravansery is a yellow card in the game which allows a player to trade one manufactured good for another. Additionally, it also provides the player with one coin for each brown and gray card in the player's and their two neighboring cities."

Also, completely and utterly wrong. Well, except the fact it's a yellow card I guess.

bobvonunheil fucked around with this message at 15:25 on May 1, 2023

Infinitum
Jul 30, 2004


panko posted:

https://boardgamechat.com/

chatgpt-style board game rules bot. played around with it for about ten minutes and didn’t see any incorrect output so could be very useful

Seems to work OK I think?

Frosthaven one that bugged our table for a bit

Pryce
May 21, 2011

Infinitum posted:

Seems to work OK I think?

Frosthaven one that bugged our table for a bit


That response is very specifically the Gloomhaven rule/guidance. Frosthaven is the inverse (remove it from the game unless it says otherwise).

EDIT: Got paranoid that I'd been playing wrong so went to look up the actual rule.

Pryce fucked around with this message at 15:53 on May 1, 2023

Jabor
Jul 16, 2010

#1 Loser at SpaceChem
I guess if you just want a ruling and don't actually care if it's got anything to do with the actual rules then it might be fine. I'm not sure there's really a market of people who want that when you put it that way though - you've got to pretend it's way better than it actually is if you want people to try it.

Kind of par for the course as far as the current batch of AI stuff goes.

Infinitum
Jul 30, 2004


lol, then I def need to re-read the book before session 2 then!

Poopy Palpy
Jun 10, 2000

Im da fwiggin Poopy Palpy XD

Jabor posted:

I guess if you just want a ruling and don't actually care if it's got anything to do with the actual rules then it might be fine. I'm not sure there's really a market of people who want that when you put it that way though - you've got to pretend it's way better than it actually is if you want people to try it.

Kind of par for the course as far as the current batch of AI stuff goes.

The important thing to remember about AI is that 70 years ago Alan Turing got to wondering how you could tell if his thinking machine was actually thinking and the answer he came up with was "If you can talk to it and not be able to tell you're not talking to a person, it might as well be thinking" and that pretty much all of AI development since has been teaching to that test.

It would be incredibly impressive to build an engine that can take in a board game rulebook and understand it well enough to build an internal model of the rules and use that model to answer questions. What we've settled for is an engine that can take in a board game rule book and build a model of the probability that one word will follow another and use that to bullshit believably.

Aramoro
Jun 1, 2012




Yeah it fails for questions that require some reasoning. For Ark Nova on f you ask it

"Are Petting Zoo Animals small animals" you get


quote:

The information provided does not specify the size of the animals. However, the Petting Zoo animals category implies that they are domesticated and generally considered safe for interaction.

It should be a fairly straight forward question but it stumbles. If you make it more tightly defined

"Are petting zoo animals size 2 or less?"

You get


quote:

Petting Zoo animals do not have a fixed size requirement. However, they can only be accommodated in their special enclosure, the Petting Zoo.

Whislt true is deeply unhelpful. Because it doesn't understand the rules or the game itself.

All these ChatBots really fit into a weird area of AI because they don't understand what they're doing at all so the intelligence is in the language parsing not the responses they give. The boardgame bot doesn't know it's answering questions about boardgames and cannot reason so your question has to be so narrow that you probably already know the answer. .

Aramoro fucked around with this message at 19:08 on May 1, 2023

Old Swerdlow
Jul 24, 2008
I started the first month on Pandemic Legacy with friends and we didn’t realize that we had to pay attention to the legacy deck during the game until after we finished and won our first game.

A little bit disappointing that our first game was a wash and we’ll have to redo it but I can’t be too angry since Pandemic is just a fun and easy time no matter what.

Infinitum
Jul 30, 2004


Old Swerdlow posted:

Pandemic is just a fun and easy time no matter what.

:munch:

Phelddagrif
Jan 28, 2009

Before I do anything, I think, well what hasn't been seen. Sometimes, that turns out to be something ghastly and not fit for society. And sometimes that inspiration becomes something that's really worthwhile.

Anonymous Robot posted:

Can’t seem to handle Arkham Horror LCG. The answer it posed to an inquiry about enemy spawn logic had appropriate terms from the game and was sort of sensible, but was ultimately incorrect and also not a really salient reply to the question.

Those are pretty complex, conditional rules, though.

It got a lot wrong about AH from my playing around with it:
- Limit 1 per investigator (it thinks that's a deckbuilding restriction)
- It thinks that if a Whippoorwill is engaged with a different investigator, it will automatically engage with you when you fight it
- Treachery cards where you must pick an option, it thinks you can't pick an option that you can't fully do. I asked if one of your choices is losing 2 resources but you only have 1, and it said you can't pick that option.

It does get some things right though. Besides just things like the phases of a turn and the types of actions you can take, it also correctly identified that adding doom to an enemy won't cause the agenda to immediately advance, and that you can't use Lucky to save you from the autofail.

SettingSun
Aug 10, 2013

I have spent way more time than I can recall reading and interpreting rules so if an AI can accurately do that for me more power to em, but I think this particular one needs some more training.

Aramoro
Jun 1, 2012




SettingSun posted:

I have spent way more time than I can recall reading and interpreting rules so if an AI can accurately do that for me more power to em, but I think this particular one needs some more training.

It literally cannot interpret. It's the one thing it can't do.

The Eyes Have It
Feb 10, 2008

Third Eye Sees All
...snookums
I shared a screenshot or two here a few? weeks ago about a system I was playing with by feeding it game rule documents and asking questions. The stuff below is all related to my fussing with that system, but it's pretty applicable in general to AI language models and what they can do.

Anyway, the AI worked by answering questions but always provided sources to point you to (document name, page number, section, etc) to back up whatever it said. Perhaps most significantly it responded with "I do not know" if something isn't explicit in the rules and it can't find a clear reference.

There's some garbage in, garbage out. This means that poorly-written rulebooks remain poorly-written, but it can help quite a lot with poorly-organized rulebooks.

Smart Index is the best way to think of it and its capabilities.


Anyway this can lead to some interesting (and sometimes nonintuitive) outcomes. For example:

Q: "Is Magic Realm a competitive or co-operative game?"
A: I don't know (because at no point in the rules is the game actually described or discussed in these terms.)


Some things it IS good at:
- Anything that's laid out in the rules explicitly, even if poorly organized
- Special cases mentioned in different locations, it's good at gathering these and "knowing" they are connected
- Remembering important stuff mentioned only once and not elsewhere and therefore easily missed / hard to look up by a human
- Providing pointers/references to specific sections and rules (i.e. acting as the aforementioned "smart index")
- Highlighting and unifying exceptions and special cases. It's REALLY good at this. Made-up example:
Q: What does the Lantern item do?
A: The lantern provides +1 to a search result rolls. Special case A increases the modifier to +2. Situations B or C reduce it to +0.



It's not good at anything requiring interpretation, because like all AI language models it doesn't actually know what it's saying. My favorite example highlighting this is the one from above, which I'll repeat here:
Q: "Is Magic Realm a competitive or a co-operative game?"
A: "I don't know"

This is an elementary question with an obvious and clear answer that was nevertheless unanswerable by the AI system I was working with, because at no point in the rules is this question actually specifically addressed. The best it could do is point to sections that described situations in which the players "competed" for something or over some outcome.


I actually think that even as-is it could be a great tool for the writers of rulebooks. Like, using it to make sure you have actually collected all the modifiers in that quick-reference table you're making, and stuff like that. It's actually really good at that sort of thing. But using as your own personal rules expert or tutor? Not so much.

Bottom Liner
Feb 15, 2006


a specific vein of lasagna
Figuring out how to ask the right questions requires more info than it would answer for you or takes longer than it would to skim the rulebook/pull up a PDF. The fact that it can get stuff wrong makes it even less useful.

Admiralty Flag
Jun 7, 2007

to ride eternal, shiny and chrome

THUNDERDOME LOSER 2022

Counterpoint: maybe we can get GPT to write a useful and referenceable rulebook for Great Western Trail. Humans sure didn't do it.

SettingSun
Aug 10, 2013

I wonder sometimes why there isn't yet a cottage industry of rewriting lovely rulebooks into something more readable and useful. Or maybe there is I dunno but a lot of rulebooks continue to be subpar.

Bottom Liner
Feb 15, 2006


a specific vein of lasagna
The BGG files section is full of that stuff. Order of Esoteric Gamers was my go to for FFG rule rewrites.

Aramoro
Jun 1, 2012




Admiralty Flag posted:

Counterpoint: maybe we can get GPT to write a useful and referenceable rulebook for Great Western Trail. Humans sure didn't do it.

Sounds like someone isn't an 'Exceptional Gamer'

FulsomFrank
Sep 11, 2005

Hard on for love
I gave up on machine learning ever helping me in games when I tried to get an edge-case question answered for an 18xx. Garbage! Throw it away! Butlerian jihad declared!

Aramoro
Jun 1, 2012




Been playing some King's Dilemma, half a dozen games in now. I wish the rewards and penalties weren't so spelt out on the Dilemma cards because it really changes the discussion from the dilemma to the rewards, especially with the open agendas and fairly narrow and well understood resource goal cards it becomes pretty straight forward to work out what everyone is doing.

I like the idea of the game but it feels frustrating at times because often the dilemma results simply don't matter to some players so they sit out the votes. Someone won a session without participating in a single vote. What affect that has on the final result is unknown obviously but I just feel you want people participating in the main mechanic all the time.

FMguru
Sep 10, 2003

peed on;
sexually

Aramoro posted:

Been playing some King's Dilemma, half a dozen games in now. I wish the rewards and penalties weren't so spelt out on the Dilemma cards because it really changes the discussion from the dilemma to the rewards, especially with the open agendas and fairly narrow and well understood resource goal cards it becomes pretty straight forward to work out what everyone is doing.

I like the idea of the game but it feels frustrating at times because often the dilemma results simply don't matter to some players so they sit out the votes. Someone won a session without participating in a single vote. What affect that has on the final result is unknown obviously but I just feel you want people participating in the main mechanic all the time.
What gets me about KD is how the various kingdom tracks don't actually affect the kingdom at all long term. The wealth track goes super high or super low, it has no effect on anything except players who have their hidden agendas pay out on whether the tracks are high or low at the end of the game. If the military strength track gets super low, its not as if that triggers a foreign invasion. It's all just grist for the scoring system.

Pryce
May 21, 2011

FMguru posted:

What gets me about KD is how the various kingdom tracks don't actually affect the kingdom at all long term. The wealth track goes super high or super low, it has no effect on anything except players who have their hidden agendas pay out on whether the tracks are high or low at the end of the game. If the military strength track gets super low, its not as if that triggers a foreign invasion. It's all just grist for the scoring system.

Yeah this is the strength of the video game Reigns, which had the game end if any track hit either extreme, so you had to keep things from hitting any of the ends of the tracks. Feels like that should've been a rule in KD as well to force some very careful voting/balancing.

!Klams
Dec 25, 2005

Squid Squad

Pryce posted:

Yeah this is the strength of the video game Reigns, which had the game end if any track hit either extreme, so you had to keep things from hitting any of the ends of the tracks. Feels like that should've been a rule in KD as well to force some very careful voting/balancing.

It usually functionally is, in that, we've never had any one track hit the end without also ending the game.

I think KD has a lot of hits, and is very very good, and its from the context of playing it, that you get a sense of the misses. It seems like horrible guild noticed too, and I'm SUPER bummed I missed the Queens Dilemma kickstarter, because it looks like they've essentially addressed every problem I had with it, and it looks like it will be my ideal game.

CodfishCartographer
Feb 23, 2010

Gadus Maprocephalus

Pillbug
Any recommendations for games that, in rough order of importance:
- Plays well at 2 players
- Is decent for players inexperienced with board / card / tabletop games
- Is mech themed

I've got a buddy that's interested in getting into the hobby, but I don't currently have a local group for larger games so it'd be just the two of us. He's super into mech video games / anime so I figure that'd be a good theme to start with to get him excited, though I also recognize a lot of mech games tend to be super mechanically heavy and probably not a great thing to toss someone new into.

Jedit
Dec 10, 2011

Proudly supporting vanilla legends 1994-2014

CodfishCartographer posted:

Any recommendations for games that, in rough order of importance:
- Plays well at 2 players
- Is decent for players inexperienced with board / card / tabletop games
- Is mech themed

I've got a buddy that's interested in getting into the hobby, but I don't currently have a local group for larger games so it'd be just the two of us. He's super into mech video games / anime so I figure that'd be a good theme to start with to get him excited, though I also recognize a lot of mech games tend to be super mechanically heavy and probably not a great thing to toss someone new into.

Tiny Epic Mechs isn't bad for that.

interrodactyl
Nov 8, 2011

you have no dignity
Tiny Epic Mechs might be a good fit.

armorer
Aug 6, 2012

I like metal.
Mechs vs. Minions was a good time, but is pretty goofy and maybe not what you're looking for. I played through it with a group of 4 that had pretty limited board gaming experience though and everyone had a blast.

Pryce
May 21, 2011
Robo Rally?

I'm only half joking, it was one of my favorite games growing up when I wasn't 'into board games' yet. I think they remade/re-released it recently.

Leperflesh
May 17, 2007

crossposting from the coupons & deals subforum because I know this is a popular game in these parts:

Adbot
ADBOT LOVES YOU

an actual dog
Nov 18, 2014

Leperflesh posted:

crossposting from the coupons & deals subforum because I know this is a popular game in these parts:

Root has been on discount a lot and it's cool, I think they're cutting their margin in order to sell more expansions.

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply