In post 127, KittyTacky wrote:
Porkens' gimmick is mildly anti-town imo, particularly if they keep it up after earlygame. I think it's only slightly scummy at this point tho.
bianco is town imo, on vibes.
Something noteworthy about intertwining ChatGPT with your play (and something that made me feel like it was going to be useful for me to use in this game). The way that you ask ChatGPT about the game is more important than what ChatGPT says back.
I.e., rather than thinking of ChatGPT as a neutral party that you are "sterilizing" your posts with in order to make them obfuscated and unreadable, maybe the questions themselves that you are posing to ChatGPT can be thought of as the content.
Interacting with ChatGPT is fun. The results ChatGPT gives back are kind of mid and unimportant. But Porkens posts aren't exactly the results ChatGPT gives back, he is posting what he is asking of the AI
Spoiler: Porkens post
In post 143, Porkens wrote:
Porkens:
I’m in a mafia game on a forum and I have been using chatgpt to make my posts, half as a gimmick and half as an experiment. Unfortunately, some players are pushing me and saying that using chatgpt is somehow detrimental to the town. I don’t believe this is true but it feel I need to respond to this accusation. Write a post explaining these main points. Then express that Titus is suspicious for wagon going, but also that I don’t hate the wagon in bloodb0t. Write in a grumpy, just-woke-up-no-coffee style.
Porkens:
Cut the paragraph about how wagon hopping can be good and say that I don’t hate the wagon on bloodb0t but that’s mostly “gut” reading
Porkens:
Ok good summarize into one paragraph
Porkens:
One more time but don’t say about finding concrete evidence
Some players are accusing me of using ChatGPT to make my posts in the mafia game and saying it's detrimental to the town, but I disagree. I've got my eye on Titus, who has been wagon hopping without any clear reasoning, and my gut feeling is that something isn't right. I don't hate the wagon on bloodb0t, but let's focus on catching the scum. Those still giving me grief about using ChatGPT should focus on their own gameplay.
I cut out all the drafts to make it more easily digestible
You can see that Porkens is engaging in a dialogue with the bot which might be similar to an internal dialogue that someone might write out in a normal mafia post
So really porkens isnt playing in a game shatteringly different way IMO. Especially with regards to Mena saying he's the obvious correct move -- I think that this is something that can be read just like any other player.
When I was playing with ChatGPT this was all but a requirement (like, you had to prod it a bit) in order to get it to say things that sound right. Those things are controlled by a player in this game, therefore, we can read them
Yeah but thing is they only started including the prompts when they came under fire.
In post 137, Menalque wrote:
Porkens is basically RPing only more extremely bc he’s not thinking about the game he’s just parroting what the bot says to him
I'm not sure that it's true, it's quite plausible that he's trying to get the bot to say what he means it to by manipulating the prompt. In which case the bot would say things in line with Porkens' opinions.
Of course that's why I want Porkens to also post the prompts he's using, instead of just the bot's response
imo if he was doing this in good faith he's have included the prompts in the first place.
The choice to include / not include the prompts seems pretty arbitrary? What makes you think so?
-Morta
ChatGPT can be manipulated into dispensing certain takes with correct prompting. This doesn't just apply to Mafia, my hobby is convincing it to give e.g illegal advice (for entertainment purposes ofc).
Note that with the correct responses, a prompt can be spliced in to essentially fake ChatGPT's opinion.
When i open this game, and starting read, i immediatly close it because i starting strongly consider self last townie alive and others are just jesters.
In post 53, biancospino wrote:
It probably can't, but it's funny to ask it and see what it conjures up.
You can just dump the whole thread as a prompt and ask it to pass a judgment, it may need some poking to get it to say anything other than "not enough information" thou.
IMPORTANT: it has a "memory" of 4000-ish tokns (i.e words). So it will see only the last few pages at best.
theoretically speaking there should be no such hard limit on the memory of GPT, since it uses a transformer model blah blah blah. the ~4000 token limit is simply the maximum prompt length for computational/speed processes.
When i open this game, and starting read, i immediatly close it because i starting strongly consider self last townie alive and others are just jesters.
In post 136, Defective Agency wrote:
1) imaginality's response to us is +++scum.
2) Menalque's talk of 'correct' play being voting Porkens and 'boring' is quite strange, and possibly +scum.
3) KittyTacky seems towny so far.
4) Morning Tweet, why did you vote Bloodbot recently? It didn't look related to the post.
-Mr. Glance
Kitty? Towny? He hasn't said anything new, his only reads were basically parroting thread consensus.
I need to look more into imaginality but for now I'm actually going to VOTE: KittyTacky.
When i open this game, and starting read, i immediatly close it because i starting strongly consider self last townie alive and others are just jesters.
It's not lazy hunting, sure you COULD read Porkens but it's harder. Thus it's anti-town to obfuscate your play in this manner. What if in another game I posted only in limericks?
When i open this game, and starting read, i immediatly close it because i starting strongly consider self last townie alive and others are just jesters.
When i open this game, and starting read, i immediatly close it because i starting strongly consider self last townie alive and others are just jesters.
When i open this game, and starting read, i immediatly close it because i starting strongly consider self last townie alive and others are just jesters.
When i open this game, and starting read, i immediatly close it because i starting strongly consider self last townie alive and others are just jesters.
When i open this game, and starting read, i immediatly close it because i starting strongly consider self last townie alive and others are just jesters.
In post 127, KittyTacky wrote:
Porkens' gimmick is mildly anti-town imo, particularly if they keep it up after earlygame. I think it's only slightly scummy at this point tho.
bianco is town imo, on vibes.
When you can, tell me a bit more about what pinged you town from bianco initially?
Being solvy. I like people who are solvy early on. It's not something I can do very well as town.
When i open this game, and starting read, i immediatly close it because i starting strongly consider self last townie alive and others are just jesters.
As an aside, this was a fun setup and would love to play this again (and hectic was in the game, forgot that, is he still active?)
But, Morning seemed a bit more loosey goosey in that game, whereas here she seems more.... careful of her words I guess you would say.
As the pig said in their sheep of me(lulz), lots of fluff with not much substance.
The game was 3 years ago and styles can change, but an interesting compare and contrast, and for me as well if you want.
Pre Edit: I do agree with KT here re: Porkens use of ChatGPT
'dREW DID IT BETTER' - T-Bone
'doctor drew our hero' - Shiki
'I love playing with him, he's got an amazing presence to him that just feels like the game is lacking something when he's not there' - JacksonVirgo
And yes Dr Glance, I put my meta out there haha, no wiki involved though.
'dREW DID IT BETTER' - T-Bone
'doctor drew our hero' - Shiki
'I love playing with him, he's got an amazing presence to him that just feels like the game is lacking something when he's not there' - JacksonVirgo
In post 137, Menalque wrote:
Porkens is basically RPing only more extremely bc he’s not thinking about the game he’s just parroting what the bot says to him
I'm not sure that it's true, it's quite plausible that he's trying to get the bot to say what he means it to by manipulating the prompt. In which case the bot would say things in line with Porkens' opinions.
Of course that's why I want Porkens to also post the prompts he's using, instead of just the bot's response
imo if he was doing this in good faith he's have included the prompts in the first place.
The choice to include / not include the prompts seems pretty arbitrary? What makes you think so?
-Morta
ChatGPT can be manipulated into dispensing certain takes with correct prompting. This doesn't just apply to Mafia, my hobby is convincing it to give e.g illegal advice (for entertainment purposes ofc).
Note that with the correct responses, a prompt can be spliced in to essentially fake ChatGPT's opinion.
Sounds to me like you're saying that Porkens could be trying to "pass off" ChatGPT as an actually competent AI, trying to present its opinion as the "correct" opinion. Is that right?
-Morta
When i open this game, and starting read, i immediatly close it because i starting strongly consider self last townie alive and others are just jesters.
In post 237, KittyTacky wrote:
Yeah, or make it say an opinion with one prompt and say they used another prompt. Or doctor the output.
Not sure I agree with this though, really seems a bit tinfoily.
'dREW DID IT BETTER' - T-Bone
'doctor drew our hero' - Shiki
'I love playing with him, he's got an amazing presence to him that just feels like the game is lacking something when he's not there' - JacksonVirgo
Huh, you're right, looking back it did have some "trust me, I'm a useful AI!"-vibe lines, which I had clearly mostly just tuned out
Do you think that would have actually worked out, though? Like, a few people in this playerlist seem to understand that ChatGPT is unreliable, and I guess the rest have worked it out quite quickly. Not trying to question the fact the posting is a bit anti town, but if that was Porkens's plan it just sounds a bit exotic and liable to fall apart immediately.
In post 240, Dei Ex Machina wrote:
Lmao perfect Drew, I was just going to ask you whether you agreed. But then why did you say you agreed with Kittty's point re: Porkens and ChatGPT?
-Morta
I agree with it being anti town and hard to read, why purposely obfuscate things? I wouldn't agree it is necessarily scummy or Porkens purposely trying to manipulate it though, hence my tinfoil comment.
And now that Porkens has dropped that gimmick leads me to believe it was just that, just for the lulz until it annoyed people and/or affected the actual game.
'dREW DID IT BETTER' - T-Bone
'doctor drew our hero' - Shiki
'I love playing with him, he's got an amazing presence to him that just feels like the game is lacking something when he's not there' - JacksonVirgo
I was experimenting and having fun with chatgpt, but it got boring and I agree ultimately anti-town as it was generating useless conversations and making it cumbersome for me to express myself. I will probably use it later to analyze posts when we have more data, but I won’t be having it write my posts anymore unless it would be exceptionally funny. Don’t want to be a pure troll this game so I stopped. Also I started including the prompts because I was asked to, not because I was “under fire.” I feel like kitty is going out of their way to misrep me. Not due to pressure from anyone did I stop, nor am I sheeping Drew’s reads.
I don’t think morning tweet is scummy. I thought posts like 183 use a lot of words to say not much but it doesn’t even really bother me. I brought it up because, out of everything, that was the one tiny rough patch when I run my brain over MT.
Right, ok, that does make sense to me.
You also allude correctly to the distinction that anti-town is not the same thing as scummy. KT, is Porkens just anti-town or also scummy to you?
Too serious too early on, and copying Porkens' gimmick (as have other users tbf but I townread bianco and BloodBot was fairly readable last game with them)
Mr. Glance
What made Imaginality's early posts unreasonably serious?
-Morta
wrt Porkens, I was lulzing and I'm not reading much into him doing it or 'reactively' dropping it. Though he does seem light on content. Not sure if this take from KT is AI.
I wasn't saying I agree with his take on MT, I just meant that I didn't know what he meant by "puffed up" and now I do.
On Menalque, I thought asking "Show me where bianco isn't being natural and genuine in 1086" would make more sense. And for some reason specifically asking for 3 posts seemed weird to me. My read on Menalque is mostly based on vibes and that question for me was the tipping point for me to want to articulate it, but I'm not always good at articulating good reasons for my reads even though they can be right while nobody else sees it.
In post 23, imaginality wrote:
@BloodB0t, the way that chess game shows pieces in the wrong places is like some demonic incestuous offspring of chess, checkers and time travel. If the intent is to gain advantage by making everyone else's heads ache then job done!
@
Drew not sure I'd go as far as policy limming a hydra
, but I do like the idea of putting hydras under pressure early as I always find them hard to read.
VOTE: Defective Agency
Positioning himself as reasonable
Spoiler: 2
In post 102, imaginality wrote:
My vote on the hydra was rvs in that I didn't have any reason to vote them based on play (it was page 1!) but motivated in that I dislike playing with hydras. I just find them hard to read (also because I generally don't know the meta of the players in them) plus some of them post too much.
So I figure seeing how they react to a wagon on them could help with reading them plus given the choice between catching a scum solo player and a scum hydra I'd rather the latter.
Never heard of Doctor Drew or the running joke.
I don't see posting chatGPT answers as copying Porkens' gimmick in this particular game (if Porkens does that in other games then it would be if someone does it there).
I am also staggered Defective Agency thought I was being serious with that chatGPT post or in general (until now).
I have a hard time believing they genuinely believed that. My vote stays.
Incredibly safe position to take as scum with the omgus. We're already opposing him, so he just accepts and cements that.
In post 115, Morning Tweet wrote:
The subplot where on one side, ChatGPT/Porkens says its suspicious for players to cast doubt on ChatGPT -- and on the other, bloodb0t finds Porkens suspicious for practically assimilatign with the AI, seems notable. I don't like bloodbot's repeated insistence against ChatGPT which feels ultimatley like a distraction to me. Like it mostly goes without saying that ChatGPT isn't exactly sherlock holms yet bloodb0t seems to feel the need to discredit it anyway. Poor learning model
On that front, some musings:
Do you think 96 is a direct neutral chatGPT output or is it
Porkens' own content in chatGPT style (perhaps rephrased by chatGPT), or chatGPT told to argue a specific view?
If it's the latter:
Do you think continuing to mix chatGPT and their own content (if Porkens were to continue to do so) is scummy, towny or NAI?
Bonus thought: compare and contrast playing using chatGPT as support (either for private discussion or for content generation) with playing in a hydra
Not really directed at anyone, just musing.
You could have these thoughts in an MD thread or pre-game. Much easier content to put out than your thoughts on other players.
In post 185, Defective Agency wrote:
How many scums are we thinking because if it's 3 I've already found them all
Maybe not but I've got it down to [Dei Ex Machina, imaginality, narrow parking, Menalque, Porkens]
Should be 3 and one 3p, at least per the queue post.
Also, I somewhat disagree with that list. Are Dei there chiefly just because of their KT vote?
PE: god, that rap was awful
This was already clarified earlier in the thread (119). And Defective Agency has been following the thread, so the "how many scums are we thinking" in the post bianco quoted here seems like a deliberate attempt on DA's part to appear uninformed.
This is an attempt to get us for 'fake townslipping' which is a really superficial and reachy thing to push us for. I hadn't checked the setup or seen anywhere that it was 9/3/1.
So yeah, bit of a scumcase for you Dei Ex Machina.
In post 228, KittyTacky wrote:
It's not lazy hunting, sure you COULD read Porkens but it's harder. Thus it's anti-town to obfuscate your play in this manner. What if in another game I posted only in limericks?
I once did a couple games in my more activate days where I only posted in images.
In post 246, BloodB0t wrote:
wrt Porkens, I was lulzing and I'm not reading much into him doing it or 'reactively' dropping it. Though he does seem light on content. Not sure if this take from KT is AI.
I wasn't saying I agree with his take on MT, I just meant that I didn't know what he meant by "puffed up" and now I do.
On Menalque, I thought asking "Show me where bianco isn't being natural and genuine in 1086" would make more sense. And for some reason specifically asking for 3 posts seemed weird to me. My read on Menalque is mostly based on vibes and that question for me was the tipping point for me to want to articulate it, but I'm not always good at articulating good reasons for my reads even though they can be right while nobody else sees it.