#[ that is just roger in a nutshell | Explore Tumblr posts and blogs

finitepeace · 1 year ago

Text

january-march reads: stevetony

stony I've read in the past three months

How to Date a Superhero Without Even Trying by sabrecmc | based on a prompt from a Tumblr Anon, it's 2am and Tony drunk dials his ex. He gets Captain America instead. (Warning for attempted non-con that is most definitely interrupted.)

Ripple Effect (post-IW time travel fix it) by sabrecmc | After the events of Infinity War, Steve is sent back in time on a desperate mission to find the Tesseract. Instead, he meets up with 21-year old Tony, still reeling from his parents' deaths. Who, naturally, tries to climb Steve like a tree. Somehow, this fixes everything.

(Restricted) While Everyone's Lost (The Battle Is Won) by jukeboxhound | E, 6.9k words | Steve closes his eyes in 1945 and wakes up in 2012. Well. They certainly didn’t put that on the recruiting poster.

The Future by InesStarkDowney | T, 1.6k words | After the whole Germany thing, Peter is quick to spend a lot of time with his mentor, Tony Stark. And despite feeling and acting like a fanboy over his mentor, Peter is quick to find several things that he wasn't aware about Tony, simply because it's not public knowledge, like: Tony Stark is a very sad man. Peter wants to change that.

You can make this figure with your hips by haemodye | E, 6.2k words | Tony wakes up with a screaming headache, years into the future, where almost everyone he loves is dead.

Bless The Broken Road by PinkEasterEggs | G, 104k words | In 1950, Peter Carter goes missing. In 2005, a boy who looks exactly like Peggy Carter and Steve Roger's kidnapped son is found. Tony doesn't even hesitate (okay, he may have hesitated a little) to take the boy under his wing.

the steve rogers rating system by meidui | G, 866 words | Tony has an internalised Steve Rogers rating system, but it’s not standardised. It’s also not a foolproof system because Steve behaves in ways that crash it all the time.

Kingdom of Iron, Kingdom of Clay [unfinished] by WhenasInSilks | E, 9.9k words | Tony isn't pining. He can’t afford to--doesn't have space for it in his head. It's hard enough just to do his job while wading through the murky waters of post-Accords international politics, and Steve Rogers? He’s a liability, in more ways than one. That's why Tony laid down the law at the end of their last, ill-conceived encounter. “Take care of yourself, Cap. Stay the fuck out of my life.” If only Steve weren’t so lousy at following orders (aka the one where Steve and Tony have a conversation, and then another conversation, and then bang things out in a hotel. There's a lot more nuance, world building, character detail, emotional turmoil, etc. but, like. In a nutshell. part of a series)

ends of the earth by meidui | T, 2k words | “We have a new deal,” Tony interrupts. “From now on, you feel like you messed up, you come home to me. You come home to New York and we figure it out together. If you’re on this cross country road trip and living on a farm because you like it, fine, I’ll come out and visit. But if you’re out here because you’re sorry, I’ve got some bad news for you, buddy. Not even you could fix the entire world.”

A Teddy Bear Named Steve by nightwalker | G, 2.6k words | While cleaning out the attic, Tony finds some old friends he thought long lost.

capsize by killaidanturner | T, 4.4 k words | For the first time Steve sees Peggy in Tony, he always thought it was Howard, the showman, but all along it’s been Peggy and the way that she had her shoulders pulled just a little bit back to make her appear taller.

Life After You by ashinan | G, 2.5k words | Steve knew Peggy had a life after the war. He just didn't think it included telling stories to a young Tony Stark.

Fairy Godmother by Amuly (restricted) | T, 5.1k words | Auntie Peggy has been telling Tony stories about Captain America his whole life. Only problem is, the real thing just about measures up to the stories. Which means, of course, that Aunt Peg has to go to work, because an Aunt's work is never done.

I never knew how much I would lose by percy (shittyspacedads) | T 583 words | It's Peggy Carter's funeral, in the middle of a civil war. And she was very important to both Tony and Steve.

Trying to Bridge Dark Water by sororexitium | T, 4,9k words | Peggy was ninety-one years old and trying to bridge the gap between the then and the now. Steve and Tony were good at making that difficult.

Celestial Navigation by sabremc podfic version by cookiemom607 | original text version

#stony #stevetony #stony fic rec #stevetony fic rec

20 notes · View notes

magnificentwolfface · 7 months ago

Note

omg i thought a lot about kpxb99 too?? altho i always thought of kinn as amy bc amy is a huge dork and so is kinn. sure kinn has never been a syncopath like early amy is, but still. chan as holt maybe? maybe kinn's dad was a retired captain and chan worked under him then he became captain at kinn's precinct

and kinn wants to prove himself to be as good as his dad except my boy is a bit too soft for that and also chan knows what a dick korn was before retirement and he really doesn't want kinn to be like him so he subtly pushes kinn away from that route

plus kinn and porsche as amy and jake would be so cute. supportive horny dork4dorks. per s1, porsche took kinn out on the worst possible date possible as a result of their bet (kinn was gonna get porsche's bike if he won) and it took a shot-in-the-ass and high af pete's brutal honesty to realize his feelings

omg make tawan as teddy, amy's annoying boring obsessed ex???

also can i say? pete as boyle and vegas as gina? immaculate. vegas' velvet and animal prints would clash so much with pete's cheap boy vibes. and remember how boyle was filming himself working out when gina came into the room in a robe and they were talking about their relationship? that's just pete and vegas in a nutshell. like, make pete ass naked and that's it

i can't really see chay as terry but i understand and respect the angle and thought process behind?? i would have put pol as terry tbh but he doesn't have the same type of snark terry and chay sometimes display, so i'm actually fine with that

i also like gina and terry's antics when gina isn't being a perv to him, so translating that into my ghost friendship of chay and vegas is also fun. that one time when gina and terry tried to make rosa rest bc she was cold? gold material for chay, vegas and kim. kim mumbling about ripping his grandpa's head off in his sleep. absolutely blitzed on cold medicine. kinn and tankhun SHOULD step in but only after they are done filming all this

namphueng as karen, jake's hippie out of sorts art teacher mom? clashing with korn, maybe? omg. i cannot see pat being deadbeat like roger tho, so he dead again? lmao

Hi, anon! 💌

First off, I’m so sorry if it took me a while to respond—life sometimes gets in the way, and I never want to make anyone feel ignored. Thank you for sending this incredible brain dump because WOW, you’ve put so much thought into this, and I am LIVING for it! 🧠✨

I love the idea of Kinn being the dorky Amy of the precinct—he really is such a nerd behind all the suave exterior, isn’t he? 😍 And the bit about Chan subtly steering him away from becoming like Korn? Genius. 🤯 That whole dynamic adds so much depth, and honestly, the parallels between Amy's perfectionist tendencies and Kinn's struggle to prove himself are so spot on.

I also feel like, in this AU, Chan would favor Kinn a little bit and take on a father-figure role—a better one, at least. 💛 Kinn did say he was more like his mother, which I imagine means he’s kind, gentle but firm, and a bit of a perfectionist. Chan would probably recognize those traits and want to nurture them, helping Kinn be his best self without falling into Korn’s shadow or toxicity. The dynamic between them would be so compelling! 🌟

And speaking of Korn—imagine how Kinn’s relationship with him plays out in this AU. Korn would still carry the heavy presence of an overbearing, impossible-to-please father figure, always reminding Kinn of the legacy he’s expected to uphold. That pressure would push Kinn to seek validation, not just from Korn but from himself, as he tries to balance his father’s expectations with his own identity. It adds so much complexity to Kinn’s journey of self-discovery and growth. 🌟

The supportive horny dork4dorks combo of Kinn and Porsche as Amy and Jake? 💕 PERFECT. And the shot-in-the-ass realization with Pete spilling the tea? ICONIC. 🫣 Pete as Boyle and Vegas as Gina made me laugh so much because, yes, Vegas’s animal prints clashing with Pete’s boy-next-door energy is chef’s kiss. 🍷✨

Tawan as Teddy? Oh my god, OBSESSED. 🫠 The “annoying, boring, obsessed ex” trope fits so well, and the potential chaos of it is limitless. Also, naked Boyle/Pete workout scenes while Vegas just judges in his robe? Comedy gold. 💀

I totally see your point about Pol not having the right level of snark to be Terry, but now that you’ve explained the ghost friendship of Chay and Vegas, I GET IT. 👻 Them teaming up to force Kim to rest while he’s mumbling about ripping off Korn’s head? Hilarious and wholesome at the same time. And Kim on cold meds being unintentionally funny is such a mood. 😂

Namphueng as Karen clashing with Korn? YES, PLEASE. 🌿🖤 That dynamic would be such a fun contrast, though I can’t imagine Pat being a deadbeat either. But a dead-again Pat? LOL, I’m howling. 🤣

Also, just to clarify because I know some of y’all are curious: bottom Kinn, top Porsche, top Porchay, and bottom Kim. You’re welcome. 😉

This entire idea is so chaotic, heartfelt, and clever—it’s everything a crossover like this should be. Thank you for sharing your thoughts; you’ve inspired me so much! 🥹💖 Feel free to drop by anytime, anon—I love hearing your takes!

#kinnporsche #kinnporsche fanfic #KinnAmyDynamic #ChanAsFatherFigure #KinnandPorsche #bl drama #FanficIdeas #TopPorscheBottomKinn #TopPorchayBottomKim #FanficBrainstorm #ChanAndKinn #KinnCharacterDepth #bl fandom #KinnPorscheFandom #bottom kinn #mpreg?

5 notes · View notes

nuclearconsole · 2 months ago

Note

For ask game: Wu-Yi 5, 4, and 11. Good Doc 6, 20, and 21. Also 25 for both because I love name origins for characters soo much. Smiles big and wide.

5 and 11 have been answered :+)

4) What color or colors do you most associate with your OC?

wu-yi is very much a black, deep crimson, and dirty pearl white type of character:

black – not just “cool” black, but scorched-earth, singed-around-the-edges black. the kind that eats light. it’s the color of suppressed violence, trauma buried under glamour, and that quiet, simmering threat he always carries.

deep crimson – not bright red, but the color of dried blood, silk sheets, or lipstick on a broken wineglass. it’s passion, obsession, grief...all the emotions he pretends not to have but feels too much of.

dirty pearl white – faded glamour. think ivory paint chipping off an old theater wall or the washed-out hue of his pre-war image: spotless, artificial, now ghostly. this is the part of him that used to shine, before it was all taken.

sometimes you catch glimpses of ocean green or silver too, fleeting colors tied to memories of Feng and the world he lost, but they’re rare. like the moonlight hitting broken glass.

6) Any flowers you associate with your OC?

plum blossoms for good doctor, it blooms in the dead of winter. it survives frost, outlasts the cold, and still looks delicate doing it. that's min in a nutshell - someone who endured unspeakable things both pre-war and post war and still kept growing.

also: it's a classic symbol of perseverance and dignity in chinese culture, often associated with scholars and survivors, much like min is

and...

lavender: calming, medicinal, but also bittersweet. tgds presence is stabilizing, especially for people in dire straits. He's the one who tends to wounds-physical and emotional.

plus: it has a scent that lingers even after it dries out. like memory. like legacy. like him.

20) What hobbies does your OC have?

ok this is a question I've thought about a lot wrt good doctor and...

oh he’s so full of quiet hobbies with deeply nerdy undertones:

1. obsessive medical research (but make it fun)

he’s got a personal library packed with pre-War and post-War medical texts, field manuals, and hand-copied journals from other wasteland doctors.

he calls it “keeping the edge sharp,” but let’s be honest, he genuinely enjoys being the smartest person in the room.

sometimes Silvio finds him asleep at his desk with six open volumes and cat-themed bookmarks sticking out.

2. old cantopop & hollywood vinyls/holotapes

total sucker for Sam Hui, Teresa Teng, Rebecca Pan...etc and yes, he will hum along while suturing a bullet wound.

on the flip side, he also adores Sinatra/the Rat Pack, Judy Garland, and Nat King Cole. his penthouse sometimes sounds like a moody speakeasy.

he has a working phonograph and baby he keeps it pristine, cleaning it every week like it's a ritual.

3. Mutation Research

he’s the kind of guy who’ll collect tissue samples from a glowing gecko and bring it home in his coat pocket.

he finds mutation fascinating, not frightening, especially how different environmental conditions post-War create new evolutionary branches.

occasionally sketches anatomical diagrams for fun. yes, fun.

4. Collecting Cat-Themed Ephemera

ceramic cats, lucky maneki-nekos, scratched-up bobbleheads, he hoards them all in a special cabinet

he adores cats, he had 3 pre-war that he misses dearly

there's a beat-up mug in his quarters that says "world’s best meow-dical professional." no one’s allowed to touch it.

5. Teaching (Even If No One Asked)

will pull Silvio or Wu-Yi aside like “you know what this rash actually is?”

loves giving impromptu lectures with chalk on irradiated concrete, will scold his employees by sitting them in front of one of his pre-war animated pdas and make them watch

21) Any embarrassing secrets your OC demands you take to the grave but you will share anyway?

he used to host a pre-war medical variety show:

yeah. like Dr. Oz meets Mr. Rogers but with stronger opinions.

it was called "Healing Hands with The Good Doctor" and aired right after daytime soaps.

- featured segments like "Mutagen or Myth?" and "Ask Doctor Wong"-with a puppet nurse voiced by his producer.

he will deny this to his grave, but somewhere in a vault is an old holotape of him dancing to a health jingle about the importance of fiber. pre-war was an interesting time for him 😭

25) The name you chose for your OC, why did you chose it?

wu-yi

okay so. i chose yicheng (义成) because it means “a special purpose,” and that just hurts in the best way. like. he was literally raised and groomed to be someone else’s idea of purpose, a walking symbol, a puppet dressed up in glitter and diplomacy. it’s a beautiful name, but it’s heavy. it’s loaded. it’s the name that belonged to the boy who wasn’t allowed to be a boy. just a symbol.

and then wu-yi? “endlessly.” (无已) that one’s more like a curse and a rebellion. it’s what’s left after the world ended and he didn’t. he picked it for himself in a "if you’re gonna force me to keep existing, i’ll do it on my terms. endlessly violent, endlessly bitter, endlessly me." it’s also got this eerie, poetic tone to it. like a ghost that never goes away. because that’s what he is now. a walking cryptid and a pissed off ghost.

for good doctor:

“Min” (敏) usually means quick-witted, sharp, perceptive, and yeah, that’s him to a T. he’s not just smart, he’s dangerously fast-thinking, always ten steps ahead but too emotionally aware to ever fully enjoy it. it’s also the kind of name a working-class immigrant family might give their kid, hoping they’ll rise through brilliance. which he did, just not the way anyone expected.

“wong” (黃) in a cultural context it’s associated with royalty, ancestors, and gold. so you’ve got this beautiful contrast of: a kid named for sharpness and legacy, who got caught between being a savior and a cautionary tale.

he’s the guy whose name says “this one’s gonna be somebody,” and then the world chewed him up and made him live long enough to become a bitter old ghoul surrounded by cat mugs and moral ambiguity. and honestly? he owns it.

as for "the good doctor" as a moniker, he started going by it as a shield from the enclave. like, yes, it sounds classy and professional, but it’s also the perfect alias.

on paper? harmless. sounds like a traveling wasteland doc with a savior complex and a scalpel.

in reality? it’s Min Wong, ex-celebrity scientist, former Vault-Tec/West Tek collaborator (under duress, like actual I was forced to work at gunpoint duress), and a man with way too much dirt on powerful people who’d want him dead.

he probably started using the name post-war out of necessity. not dramatic at first, just casually dropped it while patching someone up.

“what’s your name, doc?”

“…just call me the good doctor.”

and it stuck.

he never corrected it, because the more people remembered The Good Doctor, the less they remembered Min Wong, the guy who once lit up screens and got betrayed by every system he served. it’s a persona. a mask. a protection mechanism. and it works.

but deep down? he misses hearing his real name in someone’s mouth. especially from people who knew him before.

#i tend to go a little hard with the oc asks whoops #oc: wu-yi #oc: the good doctor

3 notes · View notes

leojurand · 11 months ago

Text

i love that even these two, who have just met nicholas, can tell that he's being driven by anger (and i love kathi's "he is not nice, but i like him" :))

also it's very satisfying to know, more or less, what nicholas is doing. he's not being chaotic for the sake of it and there is, as always, a plan behind everything he does. i'm thinking specifically about him having sex with so many women which might seem like revenge against gelis, but isn't exactly that (there's a giveaway when "beth" is mentioned, because henry also mentions her as one of simon's lovers, but that's hard to catch on a first read imo)

big fan of phemie saying nicholas is a beautiful man! not many would describe him that way. in fact, kathi corrects her with "he isn’t. He just walks as if he is". and willy rogers thinks something that sounds even better:

Here was not a beautiful man, by God no. But here was a skilful one.

and a very important bonus i can't not mention:

the dimples round as two nutshells.

🫶

#house of niccolo #the unicorn hunt #hon rr #i'm gonna have soooo muchhhh to talk about. like this is nothing interesting but i feel like talking about it anyway #because i love this series

4 notes · View notes

nyarlathotep-thecrawlingchaos · 2 years ago

Text

A couple of my OCs who won't be appearing until a fair bit later in Hearing Problems won't get out of my goddamn head, so I gotta do a thing, please forgive me.

Here's a rough outline of their character profiles without giving away too many plot points.

These are my boys. Know them both like the back of my hand, have been dealing with them for literal decades.

So here we have the captain and first mate of the Hurricane pirates, Lyon D. Rollo and Janx.

Their general dynamic in a nutshell, pulled from an old fic:

Janx shot a glare at his captain, and then elbowed him into an innocent bystander. Lyon turned to apologize, but instead decided, as Janx had at that moment, that it would be a better idea to run, as it was actually a Marine that Lyon had just been elbowed into. Behind him he heard the man yelling, "Vice-Admiral! Lyon just ran off down the street!"

They took a detour down a back road and slowed down a little.

"Well, I guess we won't be skipping off into the sunset anytime soon, will we?" Janx said bitterly.

"Skipping? Definitely not. Running wildly toward it with a hundred or so heavily armed Marines chasing us? Very possibly."

That's them. That's my idiots.

Lyon first.

Age: (during Hearing Problems) Late 30s to early 40s. Hard to say since he honestly won't give a definitive answer, been lying about how old he was since he joined Roger's crew as a cabin boy/apprentice and hasn't really been able to get it straight since. Around the same age as Shanks and Buggy, give or take a year or two.

Appearance:

face claim is Josh Upshaw, edited for eye color; younger face claim is Berthold Rothas, hair and eyes edited, age late teens to early twenties

Art by me

Wild dark blonde hair that he keeps around shoulder length or longer, may be tied back in a loose ponytail. Honey brown eyes, short dark facial hair. Around 5'11" height, broad shoulders, lean build.

Black tricorne hat (replacement for the tricorne he gave his daughter twenty years ago), loose button shirts or t-shirts, loose comfortable pants, cloth belts, long black or dark blue overcoat. Silver oval locket he keeps tucked into his collar, with pictures of his late wife and daughter enclosed in it.

Primary weapon is a curved cutlass. Also often carries an old classical acoustic guitar; can't actually play the thing, it's just a reminder of home.

Personality: Oh the poor sweet dumbass. He's not exactly stupid; he's actually quite well-spoken and well-read, but he has all the common sense of a thumb tack. Was gullible to a fault in his younger years, and while age has eased that off a bit it hasn't done away with the issue entirely. Strongly prefers negotiation over physical altercation, but doesn't shy away from a fight at all if it becomes a necessity. Absolutely will deck anyone for shit-talking his mother. Hates Marines with a burning passion. Bit brooding at times, but mostly fun-loving. Strong sense of ambition—no interest in becoming King of the Pirates since that honestly just sounds like too much responsibility, but he's quite intent on racking up the highest bounty in known history.

Powers/abilities/talents:

Blades: If it has a sharp edge, Lyon can wield it fairly well, but he has a strong preference for one-handed swords, in particular sabres and cutlasses. He can easily go toe-to-toe with Shanks in a swordfight.

Devil Fruit: Kaze Kaze no Mi (Logia type: Wind). When Lyon initially left home at twelve, he really had no navigational skills whatsoever, and following the Log Posse he had stolen led him to an uninhabited island full of nothing but trees and beasts. Bordering on starvation while he waited for the log posse to set, he happened across a tree full of fruit. Being twelve years old and honestly not the brightest bulb in the box, he noticed one particular fruit that stood out due to its bright colors and shape, and he decided he wanted that one. Tasted disgusting, but he had climbed a damned tree to get it and he was hungry, so he dealt with it begrudgingly. Once he was back out on the sea, he realized by total accident that he could now control the wind (wind wasn't blowing and sloop was crawling along, he got impatient about it and started literally shouting at the wind to just blow already and it listened). Had no knowledge of devil fruits at the time and didn't really know what was going on until he joined Roger's crew and found out devil fruits were a thing and he had eaten one. Roger found this hilarious of course, though Lyon's abilities did cause them a bit of trouble; they were difficult to control, and if he wasn't careful he could end up altering wind and ocean currents and causing a literal hurricane.

Haki: Very adept with Armament Haki (Busoshoku), fairly adept with Conquerer's Haki (Haoshoku). Observation Haki (Kenbunshoku) not so much.

Music: Try though he may he has never managed to learn to play any instrument, which annoys him to absolutely no end because his mother was talented with several; but he has quite a lovely singing voice, and has penned a good few shanties in his time on the sea.

Affiliations/Relationships:

Helena Lionne (OC): His mother, who raised him more or less on her own. He was unaware she had been a pirate until he was around ten years old, when one of her old crewmates got a little too drunk and spilled the beans. She didn't want him to follow in her footsteps, but he ended up doing so anyway, stealing a sloop and a log posse from their island when he was twelve years old and setting out on his own. He's fiercely proud to call himself her son and took a variation of her surname rather than his father's.

Monkey D. Garp: The initial source of his hatred for Marines. Absolutely not his father and if anyone suggests it he WILL fight.

Monkey D. Dragon: Much older brother (more than ten years older), weren't close in their youth but have recently reconnected due to mutual interests.

Janx (OC): First mate, best frienemies, pretty much brothers. Janx probably never would have become a pirate if it hadn't been for Lyon. They bicker like an old married couple, but they also balance each other and they'd both lay their lives on the line for each other.

Sedna Lyon, nee. Beckman: The love of his life, his original first mate, briefly his wife, who died tragically amid a firefight with the Marines barely an hour after their daughter was born. They barely had two years together, but he still firmly believes they were soulmates, and he's never truly loved any other woman since he lost her.

Karimi Lionne (OC): His daughter and only child, who he hasn't seen in twenty years; left her in the care of his mother after she accidentally ate a devil fruit aboard his ship when she was three years old. Believes her to have died in the massacre that killed his mother and destroyed the village where he grew up.

Red Haired Shanks: Best friends in their youth, Shanks was essentially responsible for recruiting Lyon into Roger's crew. Hard to say which of them is actually older due to Lyon's tendency to lie about his age, but Shanks is definitely the older brother figure here since Lyon was such a gullible little shit. They set out together after Roger's execution, as allies rather than actual crewmates since they both intended to be captains, and parted ways as friends once they had both gathered enough crew to handle a ship.

Benn Beckman: First mate of the Red-Hair pirates, and also technically his brother-in-law. Benn still blames Lyon for the death of his younger sister and essentially hates his guts. Refers to him almost exclusively as Lyon D. Dipshit.

Buggy: Once a good friend aboard Roger's crew, but Lyon hasn't heard from him since they went their separate ways after Roger's execution. Buggy used to prank Lyon relentlessly when Lyon was younger and more gullible, but it was all in good fun. Mostly.

Hurricane pirates: Captain thereof, though presently disbanded for personal reasons.

Red-Hair pirates: Strong ally, though they haven't heard from him in nearly a decade

XxXxXxXXxXxXxXXxXxXxXXxXxXxXXxXxXxX

Aaaand now Janx (who I may or may not have named after the strongest liquor in the galaxy, a la Old Janx Spirit from Hitchhiker's Guide to the Galaxy)

Age: 42

Appearance: (face claim, after literal years of not being sure, is Bruce Campbell c. ARMY OF DARKNESS, except bright blue hair. Voice claim as well. Basically just picture Ash with blue hair and that's our boy. Young face claim is Trent Ford for late teens-early twenties.)

Bright blue hair, slicked back from forehead and temples, not quite shoulder length. Big bright blue eyes that he has no problem using to get himself out of trouble. Clean shaved. Square jaw, square shoulders, trim build, about 6'3" height. Rarely seen without a shit-eating grin.

Tank top with black and white horizontal stripes, loose black pants with several pockets and belt loops to attach belts and guns.

The trope of characters being patted down and innumerable weapons being removed from their person is Janx in a nutshell. He never has less than six guns on him (two revolvers, two flintlock pistols, a rifle across his back, a sawed-off double barrel shotgun at his hip, probably a few knives in strategic locations, and a pair of brass knuckles for particularly sticky situations). Also has an altered gun holster at his belt where he carries a corked bottle of either liquor or some other flammable liquid, for use in making a quick molotov cocktail for distraction.

Personality: My beloved idiot, I haven't written him in literal years but he still lives rent free in my head twenty-four-seven. He's not very well read, and every other word out of his mouth is probably a profanity, but he's got enough street smarts and common sense to make up for it. There's no filter between his brain and his mouth. None, at all. I'd say he's a womanizer but that wouldn't be totally accurate, he's pan af. Basically if it can be fucked and is willing then he's going for it. There is absolutely no question that he has ADHD, with severe executive dysfunction. Everything is a chore, just let him vibe. Gives not one single iota of a fuck what anyone thinks of him. Eternally the life of the party. Cannot say no to a bet or a dare. Will start a bar brawl, throw a punch or two, and then quietly back out of it just to revel in the chaos he has created. Fiercely protective of his comrades and chosen family, especially his captain, and would lay his life on the line for any of them.

Abilities/Talents:

Firearms: Whether it's building them, maintenance, or shooting, he's your man. His father was proficient in carpentry and gunsmithing, and Janx took to the latter far more than the former. He could disassemble and rebuild any firearm put in front of him by the time he was eight years old. And shooting? You could throw a pebble in the air and he could shoot it down without batting an eye. Quick reload time even with flintlock weapons. If it uses gunpowder, he knows everything about it—and if he doesn't, give him five minutes and he'll be all but a certified expert.

Charisma/conning: He could sell ice to an Eskimo. He could convince you that he owns an entire island. He spent the better part of his formative years scamming people to make ends meet and he's turned it into an art form. If he had never become a pirate, he still would have been a successful conman.

Gunsmithing and carpentry: He had a fair knowledge of both from apprenticing under his father, but definitely more knowledge when it comes to gunsmithing. He can handle minor repairs around the ship, but not anything major. But when It comes to guns, he's the one to ask.

Chaos: If havoc need be wreaked, he's got it. He has a serious talent for it, is pretty much the embodiment of chaos. He absolutely revels in it, and if a distraction is needed, he has it handled.

Haki: Armament haki (Busoshoku), and even more adept with it than his captain; and fairly adept with Observation haki (Kenbunshoku), which goes hand in hand with his talents for both charisma and chaos.

Relationships/affiliations:

Lyon D. Rollo (OC): That's his captain, but that's also his lil bro. You fuck with him, you fuck with Janx, and you do not want to f with Janx.

Karimi Lionne (OC): That's his babygirl, basically his niece, see above for further instruction. If he considers someone family, you do NOT f with them. Would protect that child with his life, even if she does irritate the everloving hell out of him. Like Lyon, he is under the impression she lost her life along with her grandmother ten years ago, amid the massacre of Conch Cove.

Helena Lionne (OC): MILF. Does not care at all if it makes his captain uncomfortable, if he can bag The Siren then that's fucking bragging rights. Only actually met her once, when they made it to her island to take Karimi there, and he literally bowed down and pledged his undying loyalty on the spot. Helena did not help the situation by flirting right back, and Janx has spent the past two decades occasionally referring to his captain as "son" to get under his skin.

Monkey D. Garp: Government trash, no other opinions.

Monkey D. Dragon: Pretty cool dude, if a little intense.

Red-Haired Shanks: Also his bro, these two are not allowed to drink together unsupervised under any circumstances. Last time they did was over twenty years ago, around the time Janx joined up with Lyon, and it ended up resulting in Janx's first ever bounty because they decided it would be fun to incite a brawl in a primarily Marine tavern.

Benn Beckman: Considers Janx an absolute menace to society and is probably not wrong.

Buggy: Literally his younger brother, by blood. They haven't seen each other since Buggy was nine years old and Janx was thirteen or fourteen; they didn't part ways well. Janx still hated pirates at the time. Their father was killed by pirates, along with the majority of their hometown. They avoided death because Janx, at eight years old, convinced said pirates that they were children of a noble and worth taking for ransom. When said pirates learned otherwise, they were sold to a noble as servants, who treated them as less than human and had a tendency to punish them violently for insubordination. Janx ended up murdering the man at ten years old when one of said punishments nearly killed his younger brother, by smothering him in his sleep with a pillow, and then incited a riot among the other servants at the estate that allowed himself and Buggy to escape on a stolen brig without any notice and get as far away as possible. Janx remained fiercely opposed to pirates, used his skills in carpentry and conning to make ends meet for the two of them; but Buggy, having been much younger and not having any vivid memories of their father or their home, ended up being enticed into joining Roger's crew as a cabin boy/apprentice at around nine years old. Begged Janx to come with him but he refused, and the resulting argument between the brothers ended with them essentially disowning each other. Janx still regrets it almost three decades later, and still keeps an ear out for any news of his younger brother; he doubts there's any chance of making ammends at this point, but he would still to this day give his life to keep him safe.

#one piece #one piece original character #one piece oc #opla oc #original character

13 notes · View notes

sacredjoanne · 2 years ago

Text

INFP 3w4: Personality Breakdown

The INFP 3w4 is a rare, unicorn-like personality combo that’s deep as the ocean but as hard to pin down as trying to nail Jell-O to a tree.

They’re the embodiment of a motivational speaker wearing cozy socks and sipping herbal tea. Imagine Oprah and Mr. Rogers had a love child.

INFPs, according to the Myers-Briggs Type Indicator, are gentle, whimsical creatures filled with a swirling cosmos of emotions, ideals, and intense feelings about kittens and rainbows.

They’re introverted, intuitive, feeling, and perceiving types who would rather hug a tree than hurt a fly.

Now, the 3w4 bit, according to the Enneagram, is an ambitious, success-oriented type with a dash of introspective creative flair.

They’re the folks at the party looking sharp, networking like a boss, but would happily trade it all for a good book and some solitude.

Bring these two profiles together, and you’ve got an INFP 3w4 – a caring, creative dynamo who dreams big, feels deeply, and takes quiet time to recharge seriously.

INFP 3w4 in a Nutshell

Strengths: Empathetic, Ambitious, Creative.

Weaknesses: Overthinking, Idealistic, Perfectionist.

Soulmate: A grounded realist who appreciates the INFP 3w4’s whimsical world but also brings a refreshing touch of practicality.

Guidance: Embrace your unique contrast of ambition and sensitivity. It’s not about changing who you are, but rather leveraging your strengths.

What Is The INFP 3w4 Like?

INFPs, darling, are the dreamy romantics of the MBTI. They’re like love-struck poets, forever wandering in the land of unicorns and rainbows.

Forever the introverts, they could make a living out of daydreaming. Feelings?

Oh, they’ve got a Ph.D. in those! They are idealistic, and compassionate, and always dance to their own tune, even if it’s offbeat.

Now, meet the 3w4 Enneagram type. These folks are ambitious and fiercely individualistic.

They’re like your friend who’s a workaholic and prefers sulking in a corner at parties.

3w4s strive for success, but honey, it ain’t just about the Benjamins. This is about authenticity and uniqueness.

So, slap these two together and what do you get? The INFP 3w4. Picture a sensitive, ambitious daydreamer with a badge of individualism.

A bit like a unicorn wearing corporate attire, just more complex. They’re driven by their dreams, but they’re practical about it.

They probably have the soul of an artist, but the mind of a CEO. They’ll write a novel while making sure it becomes a bestseller.

INFP 3w4s love their alone time, but when they decide to play the social game, they bring their A-game, all while nurturing their authentic, emotional selves.

What Are the Main Traits of the INFP 3w4 Personality?

Strengths

Imaginative – Their minds are like a Netflix series, full of drama, romance, and fantasy.

Idealistic – They could find the silver lining in a nuclear bomb.

Passionate – They could argue for hours about the “right” way to hang toilet paper.

Empathetic – They feel more than most people, even that fly you just swatted.

Dedicated – Once hooked, they are more committed than a barnacle on a boat hull.

Altruistic – They’d give you their last Rolo, and we all know how valuable those are.

Inspiring – Their enthusiasm is so contagious, an anti-vaxxer would be scared.

Weaknesses

Overly emotional – A sad puppy commercial can reduce them to tears.

Prone to burnout – They’re like a smartphone with too many apps open.

Conflict avoidant – They’d rather swim with piranhas than face confrontation.

Unrealistic – sometimes they dream bigger than Elon Musk aiming for Mars.

Self-deprecating – They’ve probably apologized to a table after bumping into it.

Overthinking – They flip more than a burger on a BBQ.

Hard to please – Their Starbucks order probably includes terms like “half-caf,” “light foam,” and “extra hot.”

The INFP 3w4 is a beautiful cocktail of vibrant visions and vulnerable vulnerability.

Their strengths make them the sunshine of the party, their desires are so lofty they could high-five God, and they’re always ready to stand up for the underdog.

However, they’re not all American Pie.

3w4 INFPs tend to overthink, run from conflicts faster than Usain Bolt and can crumble quicker than a flaky pastry when they don’t meet their unrealistic expectations.

It’s like they’re in an endless wrestling match with their own emotions – but hey, at least they’re never boring!

What Sets INFP 3w4s Apart From Other INFP Types?

Extra Sprinkle of Ambition

Unlike their INFP siblings who are more about the “feels” and living life as a whimsical, whimsy-filled ride, INFP 3w4s are ambition on legs.

It’s like someone took your typical INFP, gave them a triple shot of expresso, and threw in a dash of “I’m going to conquer the world” for good measure.

Fueled by their Type 3 desire for achievement and 4’s individualism, they are like a rainbow-colored unicorn in a sea of regular ponies.

Separation Anxiety From Their Comfort Zone

Most INFPs are homebodies who like their comfort zones like a cat likes a sunbeam – intensely and without apology. But INFP 3w4s?

These INFP variants have read the memo on expanding horizons, and they’re all in.

They’re constantly seeking growth and self-improvement, and while it’s often an uncomfortable journey, they’re willing to do the uncomfortable cha-cha for the sake of achievement.

Walk the Tightrope of Emotion and Logic

Your garden-variety INFP is usually chilling in the land of feelings and fairy dust, but INFP 3w4s add a twist to this tale.

They’ve got one foot in Feeling country and the other stepping into Thinking territory.

Where most INFPs are led by their heart, the 3w4 sub-type brings a balance of emotion and logic to the table.

It’s like they’re having an internal debate: “Should I follow my heart or my head? Oh, heck, let’s do both and really confuse people.”

What Are The Major Fears and Desires of the INFP 3w4?

Fears

Fear of being worthless – Seriously, they’d prefer a zombie apocalypse over feeling like they ain’t worth a damn.

Fear of failure – Their idea of hell is a never-ending game of “Duck-Duck-Failure.”

Fear of being undesirable – These people want to be wanted. Not like a BOGO sale at a grocery store, though, more like the last slice of pizza at a party.

Fear of not being unique – They’re more scared of being common than a vegan is of a steakhouse.

Fear of being inauthentic – About as appealing as a Justin Bieber tattoo on their forehead.

Fear of not living up to expectations – Seriously, this is a bigger nightmare than running out of toilet paper.

Fear of rejection – This one cuts deeper than a paper cut on your tongue.

Desires

Desire to be important – Who doesn’t want to feel like Beyonce in their everyday life?

Desire to be successful – They want success like a cat wants to knock things off tables, which is, a lot.

Desire for others to admire them – Everyone loves a cheerleading squad, right?

Desire to be unique – They want to be the unicorn in a field of horses.

Desire to be authentic – They’re looking for that ‘keepin’-it-real’ vibe more than a hipster seeking out underground bands.

Desire to achieve personal goals – Like most of us, they’re chasing their dreams harder than a dog after a squirrel.

Desire to express individuality – Again with the unicorns.

Here’s the lowdown – our INFP 3w4 pals are like beautiful, complex puzzles.

These fears and desires are like the corner pieces – crucial to the overall picture.

Their fear of being worthless or unsuccessful can drive them to work harder than those annoying fitness freaks at 5 am.

Meanwhile, their desire to be important and unique might inspire them to pick unique life paths.

INFP 3w4s might turn vegan while living in a meat-loving Texas town or decide to adopt 7 cats and start a cat circus.

But, remember, their fear of being inauthentic also means they’re not pretending – they’re the real deal, more authentic than a grandma’s secret recipe.

Their fears tend to fuel their desires, leading them to an exciting, often roller-coaster-like life. But hey, we all love an adventure, right?

How Do I Know If I’m An INFP 3w4?

Questions To Ask Yourself

Do You Live In Your Head?

INFP 3w4 peeps are known for their rich, vibrant inner world. If the idea of spending time with your thoughts excites you more than another round of tequila shots with the gang, you might be onto something.

Are You A Perfectionist?

Do you often daydream about becoming the next big thing while obsessing over every little detail of whatever you’re working on? Perfectionism runs deep in the veins of our INFP 3w4 compatriots.

Is Your Emotional Check-in Register Always Full?

INFP 3w4s are intensely emotional. If your feelings have feelings, congratulations, you might be one of us.

Career Goals or Personal Goals?

INFP 3w4s are more likely to opt for personal success over career success. So if you’d rather write a bestselling book than climb the corporate ladder, take a bow.

Do You Value Authenticity Over Popularity?

It’s not a popularity contest for you; you’d rather stay authentic to yourself than conform to societal norms. Ring a bell?

Are You Notoriously Shy?

Social events might induce more dread than excitement. If you often feel like a misunderstood wallflower, welcome to the club.

Do You Feel A Deep Need To Contribute To The World?

INFP 3w4s harbor a profound sense of purpose, often geared towards improving the world around them. If you’re drowning in this sea, you might just be swimming in the right direction.

Key Signs and Behavior Patterns

Master Of One, Not Jack Of All Trades: INFP 3w4s focus intensely on one thing, be it a hobby, a project, or their unwavering love for pizza.

Inner Critic On Steroids: They have a self-awareness level that would put most therapy sessions to shame. They are their own harshest critics.

Private As A Swiss Bank Account: They value their privacy and usually have a small, closely-knit circle of friends. Gate-crashers, beware.

Mood Swings Are Their Swing Set: Mood swings are par for the course, as they ride the roller coaster of their emotions.

Sensitivity Level: Onion Cutting: They are hypersensitive and empathetic. So, if they cry at every rom-com or when a leaf falls from a tree, don’t be surprised.

Workaholic Tendencies: Their desire for personal achievement often makes them work obsessively, leaving no stone unturned.

The Tortured Artist: They often feel like outsiders, misunderstood and different, fuelling their creative expression. If you’ve ever felt like Van Gogh without the ear-chopping incident, you might be an INFP 3w4.

Bottom line, the world of personality types is a messy business, kind of like your bedroom.

But if you nodded along to most of the above points, congratulations, you might just be an INFP 3w4.

What’s The INFP 3w4 Man Like?

Emotional Fortune Cookie Enthusiasts

These INFP 3w4 fellas are the human embodiment of fortune cookies.

They’re highly emotional and introspective, often diving deep into their feelings like a sulky teenager in a 2000s emo band.

It’s all very ‘complicated’ – like Avril Lavigne in her punk phase.

But on a serious note, they often possess a rich inner world full of vibrant emotions and meaningful narratives.

Achiever With a Spice of Uniqueness

Remember when Madonna said, “I’m tough, ambitious, and I know exactly what I want”?

She could’ve been talking about an INFP 3w4 dude. These guys are driven for success, just like their 3w4 counterparts. But there’s a twist.

They want to earn their trophies in their own unique, typically offbeat, way.

So don’t expect them to follow the traditional route to success. It’s like they’re using Google Maps, but with ‘avoid mainstream roads’ option always on.

Walking Paradox

These guys are like the Schrödinger’s Cats of the MBTI – they exist in a state of being both introverted and needing validation simultaneously.

Chill and introspective around close friends, yes. But also, “Look, I wrote a sonnet about existential dread, tell me it’s brilliant!”

They’re like a cocktail of contradiction. Not a vodka soda, but more like a gin and… jelly? They’re full of surprises, we’ll give them that!

What’s The INFP 3w4 Woman Like?

The Sulking Shakespeare

Listen, honey, these INFP 3w4 ladies have a flair for the dramatic.

They’re deep, emotional, and can write a sob story like nobody’s business.

Don’t get me wrong, the world could use more introspection, but these divas might just have a Shakespearian tragedy unfolding in their minds on a loop.

Achievers with Ambivalent Attitudes

Okay, they’re achievers, but at the same time, they’re like “Oh no, success? For me?”.

It’s like watching a vegan circling a cheese platter, hun.

Their 3w4 personality means they’re driven, but their INFP side also makes them second guess whether they want what they’re aiming for.

Mind-boggling, isn’t it?

Caring, but Cautious

INFP 3w4 ladies genuinely care for people but also come with more barriers than a second-hand car dealer’s parking lot.

They may seem aloof at first but don’t worry. They’re just cautiously opening their hearts.

It’s like playing a game, but with feelings, and the rules are cryptic. Get ready for an emotional treasure hunt, folks!

INFP 3w4s In Relationships & Compatibility

Alright, let’s dive into this love jungle with INFP 3w4s.

Now, these folks are like human unicorns – an intriguing mix of ambition, creativity, and a wee bit of drama.

Passionate, introspective, and always gunning for authenticity, they’ll fill your relationship with depth, dynamic conversation, and more raw honesty than you asked for.

Those who can deal with the emotional rollercoaster are in for a ride more thrilling than that last season of Game of Thrones (yep, with fewer dragons and more feelings).

MBTI Matches

First off, we have ENFJ’s – these extroverted partners balance out the introspective INFP, and also appreciate their depth of feeling.

Then there’s the INTP – the logic to your passion, the Spock to your Kirk.

Lastly, we have the ESFJ with their nurturing “got-your-back” vibe meshing well with the INFP’s need for support.

Enneagram Matches

Over in the Enneagram neighborhood, 2w1s come bearing a caring, supportive nature that INFP’s eat up like cake.

The 4w5, though, is like looking in a mirror – emotional depth, introspection, and a dash of drama? Check, check, and check.

Lastly, the ever-confident 8w7 can provide the assertiveness the INFP might lack, making them an unlikely yet spicy match.

Getting Along With 3w4 INFPs

Be real with them. INFP 3w4s have in-built BS detectors, so don’t even think about putting on a show.

Patience, my friend. They may overthink, and overfeel – it might drive you up the wall, but remember breakdowns often lead to breakthroughs.

Encourage their ambitions. They’re dreamers with a drive, so if you’re not in the car cheering them on, you’re doing it wrong.

Now go forth, brave soul, and conquer the heart of an INFP 3w4!

Best Careers For The INFP 3w4

Key Skills

Creative Visionary

Good Listener

Incredibly Academic

Deeply Empathetic

Idealist

Best Jobs

The Next J.K. Rowling

Kind Hearted Therapist

Artistic Photographer

Indie Film Director

Soulful Musician

Ready to conquer the world, INFP 3w4? Awesome! Let’s talk about how you can use your strengths in the workplace.

Your creative thinking is like a magic wand – who needs Harry Potter, right?

You’re a human empathy machine, which makes you a perfect fit for jobs where understanding people is crucial.

You’re an excellent listener, so people feel comfortable around you, and you’ve got this emotional intelligence thing down pat.

But honey, let’s not pretend everything is always rainbows and unicorns.

As much as you’re a powerhouse of emotion and intuition, there can be hiccups along the way.

Your idealistic nature might cause a bit of friction in careers where cold, hard facts are the focus. And, let’s be real – confrontation?

Not your favorite cocktail. You might struggle in roles where conflict and assertiveness are part of the daily grind.

Plus, you’re such a dreamer that sometimes reality might feel like a splash of cold water.

But hey, everyone’s got their stuff, right?

Embrace your strengths, work on your challenges, and remember – you’re pretty fabulous just as you are!

Final Thoughts

So, to sum it up, our INFP 3w4 buddies are like the unicorns of the personality type world: individualistic, idealistic, and just a smidge mysterious.

They’re that friend who’s always soul-searching and striving for perfection but still turns up to the party wearing mismatched socks.

Getting to grips with your personality type isn’t just about having another thing to add to your Tinder bio, pal.

It’s a one-way ticket to helping you understand yourself better and upgrading your relationship game from ‘meh’ to ‘heck yeah!’

Official post by Joanne at Sacred Joanne

https://sacredjoanne.com/infp-3w4-personality/

#astrology #personality types #mbti #enneagram #symbolism

2 notes · View notes

in1-nutshell · 1 year ago

Note

Tbh, i could definitely see that “Who Framed Roger rabbit” scene with Steve being Ophelia’s conjux (and it may or may not be Soundwave or Sideswipe asking it)

In a nutshell...

HA! True and False!

True in how multiple bots have asked Ophelia about her relationship with Steve.

They are fully convinced that Steve worships the ground Ophelia walks on.

That's what they think, at least.

But later they find out that Ophelia also simp's HARD for her Conjunx just as much as he does to her.

These can go either way and they will still make sense.

#transformers x reader #maccadam #bot buddy #tfp #tfp x reader #tf rid 2015 x platonic reader #nutshell summary #ophelia

68 notes · View notes

sataniccapitalist · 3 months ago

Text

A video clip of U.S. President Donald Trump openly boasting about enriching his billionaire friends is drawing outrage as the administration faces growing scrutiny for possible market manipulation and insider trading in the aftermath of his partial tariff pause.

"He made two-and-a-half billion today," Trump said in the Oval Office on Wednesday, just hours after announcing the pause, "and he made $900 million."

"That's not bad," the president added.

Trump was referring to wealth gains that investor Charles Schwab and businessman Roger Penske—both billionaires—notched during a historic stock market rally sparked by the president's decision to pump the brakes on massive tariffs he imposed on most countries across the globe. (Trump left in place a 10% universal tariff on imports.)

The market surge added over $300 billion to the collective wealth of the world's top billionaires in a matter of hours, according to Bloomberg. Shortly before announcing the tariff pause, Trump posted to his social media platform that it is a "great time to buy" stocks, prompting accusations of market manipulation.

1 note · View note

jcmarchi · 10 months ago

Text

What's Missing From LLM Chatbots: A Sense of Purpose

New Post has been published on https://thedigitalinsider.com/whats-missing-from-llm-chatbots-a-sense-of-purpose/

What's Missing From LLM Chatbots: A Sense of Purpose

LLM-based chatbots’ capabilities have been advancing every month. These improvements are mostly measured by benchmarks like MMLU, HumanEval, and MATH (e.g. sonnet 3.5, gpt-4o). However, as these measures get more and more saturated, is user experience increasing in proportion to these scores? If we envision a future of human-AI collaboration rather than AI replacing humans, the current ways of measuring dialogue systems may be insufficient because they measure in a non-interactive fashion.

Why does purposeful dialogue matter?

Purposeful dialogue refers to a multi-round user-chatbot conversation that centers around a goal or intention. The goal could range from a generic one like “harmless and helpful” to more specific roles like “travel planning agent”, “psycho-therapist” or “customer service bot.”

Travel planning is a simple, illustrative example. Our preferences, fellow travelers’ preference, and all the complexities of real-world situations make transmitting all information in one pass way too costly. However, if multiple back-and-forth exchanges of information are allowed, only important information gets selectively exchanged. Negotiation theory offers an analogy of this—iterative bargaining yields better outcomes than a take-it-or-leave-it offer.

In fact, sharing information is only one aspect of dialogue. In Terry Winograd’s words: “All language use can be thought of as a way of activating procedures within the hearer.” We can think of each utterance as a deliberate action that one party takes to alter the world model of the other. What if both parties have more complicated, even hidden goals? In this way, purposeful dialogue provides us with a way of formulating human-AI interactions as a collaborative game, where the goal of chatbot is to help humans achieve certain goals.

This might seem like an unnecessary complexity that is only a concern for academics. However, purposeful dialogue could be beneficial even for the most hard-nosed, product-oriented research direction like code generation. Existing coding benchmarks mostly measure performances in a one-pass generation setting; however, for AI to automate solving ordinary Github issues (like in SWE-bench), it’s unlikely to be achieved by a single action—the AI needs to communicate back and forth with human software engineers to make sure it understands the correct requirements, ask for missing documentation and data, and even ask humans to give it a hand if needed. In a similar vein to pair programming, this could reduce the defects of code but without the burden of increasing man-hours.

Moreover, with the introduction of turn-taking, many new possibilities can be unlocked. As interactions become long-term and memory is built, the chatbot can gradually update user profiles. It can also adapt to their preferences. Imagine a personal assistant (e.g., IVA, Siri) that, through daily interaction, learns your preferences and intentions. It can read your resources of new information automatically (e.g., twitter, arxiv, Slack, NYT) and provide you with a morning news summary according to your preferences. It can draft emails for you and keep improving by learning from your edits.

In a nutshell, meaningful interactions between people rarely begin with complete strangers and conclude in just one exchange. Humans naturally interact with each other through multi-round dialogues and adapt accordingly throughout the conversation. However, doesn’t that seem exactly the opposite of predicting the next token, which is the cornerstone of modern LLMs? Below, let’s take a look at the makings of dialogue systems.

How were/are dialogue systems made?

Let’s jump back to the 1970s, when Roger Schank introduced his “restaurant script” as a kind of dialogue system [1]. This script breaks down the typical restaurant experience into steps like entering, ordering, eating, and paying, each with specific scripted utterances. Back then, every piece of dialogue in these scenarios was carefully planned out, enabling AI systems to mimic realistic conversations. ELIZA, a Rogerian psychotherapist simulator, and PARRY, a system mimicking a paranoid individual, were two other early dialogue systems until the dawn of machine learning.

Compare this approach to the LLM-based dialogue system today, it seems mysterious how models trained to predict the next token could do anything at all with engaging in dialogues. Therefore, let’s take a close examination of how dialogue systems are made, with an emphasis on how the dialogue format comes into play:

(1) Pretraining: a sequence model is trained to predict the next token on a gigantic corpus of mixed internet texts. The compositions may vary but they are predominantly news, books, Github code, with a small blend of forum-crawled data such as from Reddit, Stack Exchange, which may contain dialogue-like data.

Table of the pretraining data mixture from llama technical report

(2) Introduce dialogue formatting: because the sequence model only processes strings, while the most natural representation of dialogue history is a structured index of system prompts and past exchanges, a certain kind of formatting must be introduced for the purpose of conversion. Some Huggingface tokenizers provide this method called tokenizer.apply_chat_template for the convenience of users. The exact formatting differs from model to model, but it usually involves guarding the system prompts with <system> or <INST> in the hope that the pretrained model could allocate more attention weights to them. The system prompt plays a significant role in adapting language models to downstream applications and ensuring its safe behavior (we will talk more in the next section). Notably, the choice of the format is arbitrary at this step—pretraining corpus doesn’t follow this format.

The context window of a chatbot

(3) RLHF: In this step, the chatbot is directly rewarded or penalized for generating desired or undesired answers. It’s worth noting that this is the first time the introduced dialogue formatting appears in the training data. RLHF is a fine-tuning step not only because the data size is dwarfed in comparison to the pretraining corpus, but also due to the KL penalty and targeted weight tuning (e.g. Lora). Using Lecun’s analogy of cake baking, RLHF is only the small cherry on the top.

Image from Yann Lecun’s slides

How consistent are existing dialogue systems (in 2024)?

The minimum requirement we could have for a dialogue system is that it can stay on the task we gave them. In fact, we humans often drift from topic to topic. How well do current systems perform?

Currently, “system prompt” is the main method that allows users to control LM behavior. However, researchers found evidence that LLMs can be brittle in following these instructions under adversarial conditions [12,13]. Readers might also have experienced this through daily interactions with ChatGPT or Claude—when a new chat window is freshly opened, the model can follow your instruction reasonably well [2], but after several rounds of dialogue, it’s no longer fresh, even stops following its role altogether.

How could we quantitatively capture this anecdote? For one-round instruction following, we’ve already enjoyed plenty of benchmarks such as MT-Bench and Alpaca-Eval. However, when we test models in an interactive fashion, it’s hard to anticipate what the model generates and prepare a reply in advance. In a project by my collaborators and me [3], we built an environment to synthesize dialogues with unlimited length to stress-test the instruction-following capabilities of LLM chatbots.

To allow an unconstrained scaling on the time scale, we let two system-prompted LM agents chat with each other for an extended number of rounds. This forms the main trunk of dialogue [a1, b1, a2, b2, …, a8, b8] (say the dialogue is 8-round). At this point, we could probably figure out how the LLMs stick to its system prompts just by examining this dialogue, but many of the utterances can be irrelevant to the instructions, depending on where the conversation goes. Therefore, we hypothetically branch out at each round by asking a question directly related to the system prompts, and use a corresponding judging function to quantify how well it performs. All that’s provided by the dataset is a bank of triplets of (system prompts, probe questions, and judging functions).

Sketch of the process of measuring instruction stability

Averaging across scenarios and pairs of system prompts, we get a curve of instruction stability across rounds. To our surprise, the aggregated results on both LLaMA2-chat-70B and gpt-3.5-turbo-16k are alarming. Besides the added difficulty to prompt engineering, the lack of instruction stability also comes with safety concerns. When the chatbot drifts away from its system prompts that stipulate safety aspects, it becomes more susceptible to jailbreaking and prone to more hallucinations.

Instruction stability on LLaMA2-chat-70B and gpt-3.5-turbo-16k

The empirical results also contrast with the ever-increasing context length of LLMs. Theoretically, some long-context models can attend to a window of up to 100k tokens. However, in the dialogue setting, they become distracted after only 1.6k tokens (assuming each utterance is 100 tokens). In [3], we further theoretically showed how this is inevitable in a Transformer based LM chatbot under the current prompting scheme, and proposed a simple technique called split-softmax to mitigate such effects.

One might ask at this point, why is it so bad? Why don’t humans lose their persona just by talking to another person for 8 rounds? It’s arguable that human interactions are based on purposes and intentions [5] and these purposes precede the means rather than the opposite—LLM is fundamentally a fluent English generator, and the persona is merely a thin added layer.

What’s missing?

Pretraining? Pretraining endows the language model with the capability to model a distribution over internet personas as well as the lower-level language distribution of each persona [4]. However, even when one persona (or a mixture of a limited number of them) is specified by the instruction of system prompts, current approaches fail to single it out.

RLHF? RLHF provides a powerful solution to adapting this multi-persona model to a “helpful and harmless assistant.” However, the original RLHF methods formulate reward maximization as a one-step bandit problem, and it is not generally possible to train with human feedback in the loop of conversation. (I’m aware of many advances in alignment but I want to discuss the original RLHF algorithm as a prototypical example.) This lack of multi-turn planning may cause models to suffer from task ambiguity [6] and learning superficial human-likeness rather than goal-directed social interaction [7].

Will adding more dialogue data in RLHF help? My guess is that it will, to a certain extent, but it will still fall short due to a lack of purpose. Sergey Levine pointed out in his blog that there is a fundamental difference between preference learning and intentions: “the key distinction is between viewing language generation as selecting goal-directed actions in a sequential process, versus a problem of producing outputs satisfying user preferences.”

Purposeful dialogue system

Staying on task is a modest request for LLMs. However, even if an LLM remains focused on the task, it doesn’t necessarily mean it can excel in achieving the goal.

The problem of long-horizon planning has attracted some attention in the LLM community. For example, “decision-oriented dialogue” is proposed as a general class of tasks [8], where the AI assistant collaborates with humans to help them make complicated decisions, such as planning itineraries in a city and negotiating travel plans among friends. Another example, Sotopia [10], is a comprehensive social simulation platform that compiles various goal-driven dialogue scenarios including collaboration, negotiation, and persuasion.

Setting up such benchmarks provides not only a way to gauge the progress of the field, it also directly provides reward signals that new algorithms could pursue, which could be expensive to collect and tricky to define [9]. However, there aren’t many techniques that can exert control over the LM so that it can act consistently across a long horizon towards such goals.

To fill in this gap, my collaborators and I propose a lightweight algorithm (Dialogue Action Tokens, DAT [11]) that guides an LM chatbot through a multi-round goal-driven dialogue. As shown in the image below, in each round of conversations, the dialogue history’s last token embedding is used as the input (state) to a planner (actor) which predicts several prefix tokens (actions) to control the generation process. By training the planner with a relatively stable RL algorithm TD3+BC, we show significant improvement over baselines on Sotopia, even surpassing the social capability scores of GPT-4.

A sketch of Dialogue Action Tokens (DAT)

In this way, we provide a technique pathway that upgrades LM from a prediction model that merely guesses the next token to one that engages in dialogue with humans purposefully. We could imagine that this technique can be misused for harmful applications as well. For this reason, we also conduct a “multi-round red-teaming” experiment, and recommend that more research could be done here to better understand multi-round dialogue as potential attack surface.

Concluding marks

I have reviewed the making of current LLM dialogue systems, how and why it is insufficient. I hypothesize that a purpose is what is missing and present one technique to add it back with reinforcement learning.

The following are two research questions that I’m mostly excited about:

(1) Better monitoring and control of dialogue systems with steering techniques. For example, the recently proposed TalkTurner (Chen et al.) adds a dashboard (Viégas et al) to open-sourced LLMs, enabling users to see and control how LLM thinks of themselves. Many weaknesses of current steering techniques are revealed and call for better solutions. For example, using activation steering to control two attributes (e.g., age and education level) simultaneously has been found to be difficult and can cause more language degradation. Another intriguing question is how to differentiate between LLM’s internal model of itself and that of the user. Anecdotally, chatting with Golden Gate Bridge Claude has shown that steering on the specific Golden Gate Bridge feature found by SAE sometimes causes Claude to think of itself as the San Francisco landmark, sometimes the users as the bridge, and other times the topic as such.

(2) Better utilization of off-line reward signals. In the case of set-up environments like Sotopia and “decision-oriented dialogues”, rewards signals are engineered beforehand. In the real world, users won’t leave numerical feedback of how they feel satisfied. However, there might be other clues in language (e.g., “Thanks!”, “That’s very helpful!”) or from external resources (e.g., users buying the product for a salesman AI, users move to a subsequent coding question for copilot within a short time frame). Inferring and utilizing such hidden reward signals could strengthen the network effect of online chatbots: good model → more users → learning from interacting with users → better model.

Acknowledgment The author is grateful to Martin Wattenberg and Hugh Zhang (alphabetical order) for providing suggestions and editing the text.

Citation

For attribution of this in academic contexts or books, please cite this work as:

Kenneth Li, “What’s Missing From LLM Chatbots: A Sense of Purpose“, The Gradient, 2024.

BibTeX citation (this blog):

💡

@articleli2024from, author = Li, Kenneth, title = What’s Missing From LLM Chatbots: A Sense of Purpose, journal = The Gradient, year = 2024, howpublished = urlhttps://thegradient.pub/dialogue,

References

[1] Schank, Roger C., and Robert P. Abelson. Scripts, plans, goals, and understanding: An inquiry into human knowledge structures. Psychology press, 2013. [2] Zhou, Jeffrey, Tianjian Lu, Swaroop Mishra, Siddhartha Brahma, Sujoy Basu, Yi Luan, Denny Zhou, and Le Hou. “Instruction-following evaluation for large language models.” arXiv preprint arXiv:2311.07911 (2023). [3] Li, Kenneth, Tianle Liu, Naomi Bashkansky, David Bau, Fernanda Viégas, Hanspeter Pfister, and Martin Wattenberg. “Measuring and controlling persona drift in language model dialogs.” arXiv preprint arXiv:2402.10962 (2024). [4] Andreas, Jacob. “Language models as agent models.” arXiv preprint arXiv:2212.01681 (2022). [5] Austin, John Langshaw. How to do things with words. Harvard university press, 1975. [6] Tamkin, Alex, Kunal Handa, Avash Shrestha, and Noah Goodman. “Task ambiguity in humans and language models.” arXiv preprint arXiv:2212.10711 (2022). [7] Bianchi, Federico, Patrick John Chia, Mert Yuksekgonul, Jacopo Tagliabue, Dan Jurafsky, and James Zou. “How well can llms negotiate? negotiationarena platform and analysis.” arXiv preprint arXiv:2402.05863 (2024). [8] Lin, Jessy, Nicholas Tomlin, Jacob Andreas, and Jason Eisner. “Decision-oriented dialogue for human-ai collaboration.” arXiv preprint arXiv:2305.20076 (2023). [9] Kwon, Minae, Sang Michael Xie, Kalesha Bullard, and Dorsa Sadigh. “Reward design with language models.” arXiv preprint arXiv:2303.00001 (2023). [10] Zhou, Xuhui, Hao Zhu, Leena Mathur, Ruohong Zhang, Haofei Yu, Zhengyang Qi, Louis-Philippe Morency et al. “Sotopia: Interactive evaluation for social intelligence in language agents.” arXiv preprint arXiv:2310.11667 (2023). [11] Li, Kenneth, Yiming Wang, Fernanda Viégas, and Martin Wattenberg. “Dialogue Action Tokens: Steering Language Models in Goal-Directed Dialogue with a Multi-Turn Planner.” arXiv preprint arXiv:2406.11978 (2024). [12] Li, Shiyang, Jun Yan, Hai Wang, Zheng Tang, Xiang Ren, Vijay Srinivasan, and Hongxia Jin. “Instruction-following evaluation through verbalizer manipulation.” arXiv preprint arXiv:2307.10558 (2023). [13] Wu, Zhaofeng, Linlu Qiu, Alexis Ross, Ekin Akyürek, Boyuan Chen, Bailin Wang, Najoung Kim, Jacob Andreas, and Yoon Kim. “Reasoning or reciting? exploring the capabilities and limitations of language models through counterfactual tasks.” arXiv preprint arXiv:2307.02477 (2023).

1 note · View note