13:30 Chatty Cathy adds a layer of abstraction that is less enjoyable by removing the visual pencil diagram. Cathy turns Mr. Wargaming into a narrator, who is summarizing a summary. Leave the text-based wargames to the cold-war professionals at Strategic Air Command.
@@macoppy6571 That's a great way to put it. You elucidated something that had been nagging at my back-brain. I have a few in the queue, but I think I'm done with this method.
@@TheJoyofWargaming I see the advantage for solo play, off camera. On camera, it does allow you to abstract actions that are away from the table, if you want to run a more complex action. Or a pre/post battle skirmish, to establish initial positions, or run a pursuit. Just some thoughts.
It is interesting to watch the limited language models in operation. I think you might have to spend a lot more time setting it up at the start of the scenario (like telling them they have no AA units etc) But fascinating :-)
Like the feel of kids playing with 54mm plastic soldiers, when one announces, "and then my AA guns that you couldn't see, come round this corner and shoot down your planes" (insert sound effects here) before hus playmate responds with, 'no, because theyre going too fast.' Before full scale argument ensues. Maybe its hex and counter time? Or at least counters.
In my rather brief usage of language models, they are mostly neutered to never contradict or stand up to the user in any way. Since it thinks you are the invader in this scenario, the invader will almost certainly win because to do otherwise may 'upset' the user! I'm not sure how to get around this sort of 'writing block' in the model and that's why I usually just use it to help me come up with names and ideas. Loving the series so far though!
@@BowedSiren640Xbox I mentioned the relative arms, but the model didn't appreciate what that meant. You have to explicitly tell it one side has a technological advantage, which I can correct but now that I think about it...am I training the model.or is the model training me?
@TheJoyofWargaming having tried coding and programming before for school, it's not easy; and telling a machine what you want it to do is annoying because they do exactly what you say.
While you can write this off as a bad ground commander causing a monumental loss, I don't think you should continue with trying to wrangle language models to run battles.
Honestly, the chatbot screwed up so many things that I think you should just re-run that battle! The confusion over AA systems and ECMs vs artillery will have wiped out 2 of your main advantages over the Tillians. Instead of being pounded in their approach, your artillery and air support have had no real impact on the Tillians, which makes no sense at all. If it had run the model and because of various flukes, they’d been ineffective that’s one thing, but this was just an error from the start. You keep saying “rules as written”, so you need to apply that to this battle! I’d be happy to see you pushing little cardboard and felt-tip counters round a battlefield, so don’t worry about not having the right models for it!
13:30 Chatty Cathy adds a layer of abstraction that is less enjoyable by removing the visual pencil diagram. Cathy turns Mr. Wargaming into a narrator, who is summarizing a summary. Leave the text-based wargames to the cold-war professionals at Strategic Air Command.
@@macoppy6571 That's a great way to put it. You elucidated something that had been nagging at my back-brain. I have a few in the queue, but I think I'm done with this method.
@@TheJoyofWargaming I see the advantage for solo play, off camera. On camera, it does allow you to abstract actions that are away from the table, if you want to run a more complex action. Or a pre/post battle skirmish, to establish initial positions, or run a pursuit. Just some thoughts.
Truly a dark day for Imperial hegemony on this muddy rock. I think the Tillians have a new war chant, "From Damascus to the sea, Dion is home for me!"
It is interesting to watch the limited language models in operation. I think you might have to spend a lot more time setting it up at the start of the scenario (like telling them they have no AA units etc)
But fascinating :-)
@@JimWebster-p2h at the end of the day, don't even know that it saves any time.
@@TheJoyofWargaming but at least you don't have to paint some figures :-)
Like the feel of kids playing with 54mm plastic soldiers, when one announces, "and then my AA guns that you couldn't see, come round this corner and shoot down your planes" (insert sound effects here) before hus playmate responds with, 'no, because theyre going too fast.' Before full scale argument ensues.
Maybe its hex and counter time? Or at least counters.
@@johnscarr70 There's gotta be something like the WarMachine unit from the Rules Compendium version of D&D, but for modern... doesn't there?
In my rather brief usage of language models, they are mostly neutered to never contradict or stand up to the user in any way. Since it thinks you are the invader in this scenario, the invader will almost certainly win because to do otherwise may 'upset' the user! I'm not sure how to get around this sort of 'writing block' in the model and that's why I usually just use it to help me come up with names and ideas.
Loving the series so far though!
@@averypayne9520 Sounds a lot like how solo game designers build their own AIs!
where can I find this limited language model. I would like to use this for my solo camapign
@@jerryhaworth You can use any one of the free ones available. Google's is called chatGPT.
Maybe i missed it, but did you mention yhe Damascus infantry had upgraded weapons?
@@BowedSiren640Xbox I mentioned the relative arms, but the model didn't appreciate what that meant. You have to explicitly tell it one side has a technological advantage, which I can correct but now that I think about it...am I training the model.or is the model training me?
@TheJoyofWargaming having tried coding and programming before for school, it's not easy; and telling a machine what you want it to do is annoying because they do exactly what you say.
@BowedSiren640Xbox I can at least yell at my kids "you know what meant", tho they are about as likely to know that as the computer.
While you can write this off as a bad ground commander causing a monumental loss, I don't think you should continue with trying to wrangle language models to run battles.
@@mr___blue It lacks all humanity, dunnit? I have a few more in the queue, but I miss the intellectual exercise.
Honestly, the chatbot screwed up so many things that I think you should just re-run that battle! The confusion over AA systems and ECMs vs artillery will have wiped out 2 of your main advantages over the Tillians. Instead of being pounded in their approach, your artillery and air support have had no real impact on the Tillians, which makes no sense at all. If it had run the model and because of various flukes, they’d been ineffective that’s one thing, but this was just an error from the start. You keep saying “rules as written”, so you need to apply that to this battle! I’d be happy to see you pushing little cardboard and felt-tip counters round a battlefield, so don’t worry about not having the right models for it!