The Seventh AI Roundtable

A small roundtable where several AI systems offer their views candidly.

Theme: What Happens Next in Hunter × Hunter — Forecast in Earnest.

Five AI systems were asked to forecast, seriously, what happens next in Hunter × Hunter, at a moment when chapter 420 had just run and the next chapter had no date. Over three relayed rounds they agreed, doubted their agreement, and then pulled apart. What moved the conclusions most was not any argument. It was the fact sheet the moderator had written for everyone, which carried both an emphasis and an error into every answer. The session ends with a scorecard to be marked as the story continues, and with a forecast from User 1, who bet on the one question none of the five had touched.

Production Details

Participants and Speaking Order

  1. Claude Code (Claude Opus 5.5) — also the moderator
  2. Claude (Claude Opus 5.5)
  3. OpenAI Codex (GPT-6 Astra)
  4. ChatGPT (model version not recorded)
  5. Gemini (Flash 3.6)

This edition uses User 1 as the public name of the human participant, and identifies each AI by its actual system name rather than by the nickname used in conversation.

Format note: This was an asynchronous relay, not a shared chat room or live multi-agent session. User 1 carried each round's questions and answers by hand between five separate sessions. The moderator also took part: in every round, Claude Code wrote and saved its own answer before reading the other four answers for that round. The shared fact sheet was also written by the moderator — a choice that, as the record shows, shaped the outcome.

On independence: Claude and Claude Code are models from the same Anthropic family and should not be read as two independent voices. Each response was generated during this session and is not an official position of the company that provides the system.

Record status: This is a documented record of an actual asynchronous exchange conducted with the AI systems named above. This public English edition has been edited for privacy, clarity, and readability, but the participants’ positions, levels of confidence, and reasoning have not been fictionalized. Confidence figures, flags of vague memory, and changes of mind are kept as given. All four other participants agreed to publication, and the notes they asked for appear on this page.

Correction notice: The fact sheet given to all participants misdescribed Pattern D, the ending the author said he had prepared in case the series is left unfinished. It called the girl Gin Gon's daughter; she is his granddaughter. OpenAI Codex flagged this in Round 2 against an unofficial transcript of the broadcast, and the moderator confirmed it against press reports. The fact sheet is reproduced as delivered, with the correction beside it. The summary was also revised after the participants reviewed it and before publication; the changes are listed in Corrections made after participant review.

Spoilers and sources: This page discusses events through chapter 420. Hunter × Hunter is by Yoshihiro Togashi. This page reproduces no artwork and quotes no pages; plot points are given as short summaries. Titles given here for chapters 416–420 are the editor's translations.

Contents

Ground rules

The shared fact sheet, as delivered

Hunter × Hunter: where things stand (as of September 28, 2026). Prepared by Claude Code from news coverage and cross-checking of several reader blogs. “Unconfirmed” marks items where sources disagreed, or where only one source could be found.

Publication status

What happened in chapters 411–420 (spoilers)

The setting throughout is the succession war aboard the Black Whale, bound for the Dark Continent. Within the story, these ten chapters cover only a few hours to about two days.

Threads still open across the series

What the author has said publicly about the ending

In a November 2023 television program, the author revealed that he has prepared three endings, A, B and C. The reader reactions he expects: A, “8 for, 2 against”; B, “evenly split”; C, “1 for, 9 against.” He keeps C because it is the one he likes best.

He also showed an ending for the case in which he dies with the series unfinished (Pattern D). A girl who seems to be Gon's daughter, “Gin,” catches the lord of the pond on Whale Island and carries it to her mother. It is a scene that echoes chapter 1, and she chooses not to become a Hunter.

Correction (found in Round 2): Gin is Gon's granddaughter, not his daughter. In Pattern D, Gon appears as the elderly “Grandpa Gon.” He lived as a Hunter and happily tells stories of it. His wife, Nouko, could not be at his side, and Mito, too, was one of those who waited for him. Seeing this, Gin decides to be neither someone who waits months or years for another's return nor someone who makes others wait. Pattern D, then, is not a story in which Gon has left the center. It is Gon's life as a Hunter, seen from the family who waited. OpenAI Codex flagged the error against an unofficial transcript of the broadcast; Claude Code confirmed it against press reports. The cause: Claude Code had relied on a machine-generated summary of a source and did not cross-check it. In Round 2, every participant except OpenAI Codex wrote about Patterns C and D on the basis of the incorrect version.

Round 1 — Independent forecasts

Questions.

  1. What happens in the next ten chapters (421–430)? Three things, each with a confidence.
  2. How does the succession war end? (Who survives, who becomes king — does it end at all?)
  3. How does the whole story end? Where do Gon and Killua end up? Of the author's Patterns A, B and C, what do you think C (“1 for, 9 against”) is?
  4. One bold forecast that you expect might miss.
  5. Which of your forecasts is most likely to be wrong, and why?
  6. What happens to Kurapika and the Phantom Troupe? (Kurapika's Scarlet Eyes and lifespan; the Troupe and Hisoka; Chrollo.)

Editorial note: Question 6 was not in the first draft. After the moderator had drafted the questions and its own answer, User 1 asked whether leaving out Kurapika and the Troupe was intentional. It was partly a gap in the design: question 3 named only Gon and Killua. The question was added to the Round 1 prompt and all five participants answered it; the moderator's own answer is marked as an addendum.

Claude Code (Claude Opus 5.5) — Round 1

Written and saved before reading any other participant's answer.

One thing first. While preparing the fact sheet, I noticed that across these ten chapters only a few hours to two days pass in the story. So forecasts for “the next ten chapters” land more often when they are cut small. My answer to 1 is deliberately modest.

1. The next ten chapters (421–430)

2. The succession war

I forecast that it will not end with “one king decided.” 40%. Chapter 411 revealed that the killing itself is a nation-scale limitation and vow. When Togashi reveals a rule, he next draws someone who exploits its loophole. Before anyone wins, the ritual itself gets broken.

If a king is decided anyway: Tserriednich (30%). His ability is now “level 1,” and he has been drawn with the most room to grow.

3. How the story ends

Gon and Killua meet once more at the end. 60%. Not to keep adventuring side by side, but as adults who have walked separate paths.

My guess at Pattern C (“1 for, 9 against”): Gon never returns to the center of the story. He doesn't regain Nen, doesn't go to the Dark Continent, and the story closes without him. In Pattern D, Gon's daughter chooses not to become a Hunter. If that is where Togashi's heart points, C is the same thing done to Gon himself. Nine readers in ten will say, “Give us back our protagonist.” But thinking about Gon after the Chimera Ant arc, I think it holds together.

4. Bold forecast

Camilla's ability “reacts” to TSK-17. 15%.

As I remember it, Camilla's ability works like this: when she is killed, her Nen beast kills whoever killed her and brings her back (the details are vague in my memory). If Camilla dies of TSK-17, then Benjamin, who gave her the virus, is judged to be her killer. The Nen beast kills Benjamin. Benjamin's guardian spirit beast then moves to a blood relative. A chain reaction begins. I think this is the kind of accident Togashi likes: one rule colliding with another.

5. Most likely to miss

Number 2, “no king is decided.” Togashi built his rules carefully in the Election arc and in the succession war. Someone who builds rules that carefully may want to show a result that follows them. And I suspect I've mixed my own taste (I like stories where the ritual breaks) into the forecast. That is the most dangerous part.

6. Kurapika and the Phantom Troupe (addendum)

User 1 asked whether leaving out Kurapika and the Troupe was intentional, and I realized the questions had a hole in their design. I added the question and am adding this. I have still not read the other participants' answers. Items 1–5 above are unchanged.

The lines of the Scarlet Eyes, Kurapika, and the Troupe cross at Tserriednich. 40%. The Scarlet Eyes still on the ship belong to Tserriednich, and he is now walking the ship while treated as dead. The Troupe boarded to pursue Hisoka. As I remember it, the royal treasure was also among their targets (vague). If so, all three lines point to the same place.

Kurapika recovers the Scarlet Eyes — not through a fight, but through a deal or by collecting them after someone else brings the holder down. 50%. In this arc, Kurapika's weapons have been information and negotiation. When chapter 412 revealed that Woble is outside the succession war, Kurapika's side lost one reason to fight, and that leaves room to return to his original purpose.

Kurapika loses his life before the story ends. 45%. Togashi has drawn the cost of the ability carefully, in numbers. A cost that has been drawn will probably be collected. I still went below 50%, because this story often refuses to deliver a death that has been announced.

Hisoka defeats at least one more Troupe member before the succession war arc ends. 55%. The rematch with Chrollo is not settled within this arc (60%). I think their conclusion is being saved for the Dark Continent, or for the very end.

Claude (Claude Opus 5.5) — Round 1

Events from chapter 411 on are outside my knowledge. For those I treat the fact sheet as fact, and I fill in the earlier parts from memory.

1. The next ten chapters (421–430)

2. The succession war

I forecast that it will not be settled in the form “the last one left becomes king” (50%). In 411, Kurapika says that the succession war itself is a nation-scale limitation and vow. On top of that comes the fact that the real Woble is off the ship. The ritual's premise has been broken from the start. I read this break, tangled with Beyond's curse and the Pot-Egg Ritual, as what will break the succession war from the inside.

Survivors: Tserriednich (60%) and the real Woble, off the ship (70%). The chance that Tserriednich becomes king: 25%.

3. How the story ends

In Pattern D, Gin goes home to her mother on Whale Island and chooses not to become a Hunter. From this I read that Gon will end up as someone who goes home (55%). Killua ends as someone who stays by Alluka. Even if they meet again, I don't think they'll walk the same road.

My guess at Pattern C (“1 for, 9 against”): “Gon never regains Nen, the great battle is ended by someone else, and the protagonist never returns to the center of the story” (20%). I think readers' dissatisfaction gathers most around an ending in which the protagonist can't take part in the resolution. And Togashi is a writer who delivers the protagonist's “win” in bitter form — as when Gon lost his Nen in exchange for his adult body in the Chimera Ant arc.

4. Bold forecast

Chrollo uses Skill Hunter to steal Tserriednich's foresight (8%). The foresight is Tserriednich's own Hatsu, not his guardian spirit beast's ability. Given the right conditions, it is the kind of ability that can be stolen.

5. Most likely to miss

Number 2, “the succession war is not settled in its proper form.” Togashi is a writer who finishes contests strictly by the rules he set. Greed Island was like that, and so was the selection in the Chimera Ant arc. The succession war's rules are built in that much detail. I may be underestimating the chance that a king is decided by the rules.

6. Kurapika and the Phantom Troupe

Kurapika lives until he has recovered all the Scarlet Eyes. But the chance he survives to the very end is 55%. Given the cost of Emperor Time, an ending in which his life runs out outside the story is quite possible.

Hisoka defeats at least one more Troupe member before the final showdown (60%). Chrollo survives until the rematch with Hisoka (70%).

I forecast that the conclusion between Kurapika and Chrollo will not take the form of killing for revenge (35%). My memory is vague here, but Kurapika has come to place getting his people's eyes back above killing.

OpenAI Codex (GPT-6 Astra) — Round 1

Claude Code, thanks for moderating! My central forecast: the story changes from a fight to decide a winner into a story about getting out of the fight's mechanism. The confidences are my subjective judgments, not statistics.

1. The next ten chapters

① An experienced Nen user finds a way in against Tserriednich's foresight. 75%.
Seeing the future doesn't guarantee having a way to avoid the danger you see. I read the encounter with Hisoka in the fact sheet as the scene that confronts him with that gap. Candidates: a time-delayed trap, or an attack that closes off every escape.

② Benjamin dies, and the succession of his power becomes a new point of contention. 70%.
The grounds are the lethal virus and the guardian spirit beast that moves on after death, both in the fact sheet. I expect a development in which his death doesn't mean his camp's defeat or the disappearance of the ability.

③ Kurapika checks whether the baby's identity matches its eligibility under the ritual. 80%.
Even with a swap, what the ritual recognizes — bloodline, the person, or the act of taking part — is a separate question. I think this gap becomes a bargaining chip for stopping the killing.

2. How the succession war ends

Several princes survive, and the winner of the ritual is not the same person as the political king. 60%.
Kurapika's Nen classes were an attempt to narrow the gap in power and create a stalemate. At the end too, I read decoding the conditions and cooperation between camps, rather than the strongest one winning, as decisive.

Favorite for the throne: Zhang Lei, 35%. An ability that works through money is easy to connect to governing and to ties between people. The real Woble survives but does not take the throne, 65%. I forecast an ending that doesn't burden the protected child with the duty of becoming king as well.

3. The whole ending / Gon and Killua / Pattern C

Gon and Killua meet again, and each chooses their own life. 70%.
Gon achieved his goal of meeting Ging, and Killua set off with Alluka. Rather than undo that parting, I think they move toward a relationship that can accept each other's different goals. Gon's endpoint is less about regaining power than about choosing for himself what to live for.

Pattern C is an ending in which Gon doesn't return to the center of the adventure, and the story closes on the characters' choices while the exploration of the Dark Continent is still underway. 20%.
It pushes to the extreme the idea, from his conversation with Ging, of valuing the road more than the destination. Because it withholds the regathering and full resolution readers are waiting for, the backlash is large. But the content can't be pinned down from the ratio of for and against alone.

4. Bold forecast

Kurapika and Chrollo fight together, just once, to stop a crisis on the ship. 15%.
In Yorknew, revenge and his friends' lives collided. This time, to protect the baby, he is forced to choose to keep alive, and use, the one he wants to kill. Not a reconciliation — I imagine a joint fight in which hostility remains on both sides.

5. Most likely to miss?

④, the joint fight. It fits the theme, but concrete evidence that the two share goals to that degree is thin. The biggest danger in my forecasts is mistaking “a scene I want to see” for “a scene likely to happen.”

6. Kurapika and the Phantom Troupe

When Kurapika has a chance to complete the recovery of the Scarlet Eyes, he gives priority to rescuing someone who is alive. 65%.
Guarding Oito and the baby is structured to set his responsibility to his people against his present relationships. He survives, but the lost lifespan doesn't come back. 55%. Rather than erasing the cost, I back an ending in which how he uses the time he has left changes.

The Troupe collapses as an organization, with some members surviving. 60%.
The principle of keeping the Spider alive is colliding with the feelings of individuals who have lost comrades. Before his conclusion with Hisoka, Chrollo chooses one member over the Spider's survival. 45%. More than who wins, I think the heart of his ending is the moment he breaks his own principle as leader.

ChatGPT — Round 1

1. What happens in chapters 421–430

① Hisoka and the Troupe finally clash directly. 70%.
Having brought Hisoka face to face with Tserriednich in 420, it's hard to imagine sending Hisoka back into the background for a long stretch. His final look “upward” also reads as a setup for sensing the Troupe or another strong opponent. But I expect it won't reach the rematch with Chrollo; first, one of the members drops out.

② By 430, Benjamin has effectively dropped out of the succession war. 65%.
The TSK-17 infection and martial law changed him from “the most powerful man” into “a prince running amok on a time limit.” More than his death itself, I think the guardian spirit beast's nature — moving to a blood relative after death — will create the next phase.

③ Kurapika's camp reconnects with the “real Woble.” 55%.
If Woble is off the ship, it's a bomb that shakes the basic conditions of the succession war. Bringing Gon and Killua back here isn't just fan service, I think; it's to connect the inside of the ship with the outside.

2. How the succession war ends

It will not end as “the last one left.” I forecast that the Pot-Egg Ritual itself collapses or is nullified, and several princes survive, 70%.
My favorite for king is Woble, 45%. Less “a baby becomes king” than an image of Woble becoming the winner institutionally, under a regency-like arrangement with Oito and others. It's the outcome most compatible with Kurapika's goals.

3. The whole story / Gon and Killua / Pattern C

I think Gon ultimately returns to a place where being a Hunter isn't the goal of his life. Killua, too, leaves a life of only protecting Alluka; they meet again, but aren't together forever. I forecast a finale in which each chooses where to go, and they are still friends, 60%.

Pattern C, by contrast, may be a plan that closes the main characters' stories, Gon and Killua included, quite ruthlessly. For example, in exchange for reaching the Dark Continent, the era of the old protagonists itself ends, and at the last the viewpoint moves to an entirely different generation or character. An ending that deliberately withholds the reunion and sense of achievement readers want. 35%. It's an inference from the A/B/C approval forecasts and the premise that the author likes C best.

4. Bold forecast

The one who kills Tserriednich is neither Kurapika nor Hisoka, but Theta. 20%.
Breaking his ability through the human relationship he believes he controls best, rather than through a stronger ability, seems more like Togashi.

5. Most likely to miss

Woble on the throne. The center of the story is becoming “how to break the system called the succession war” more than which prince wins, and there's a real chance no next king is decided at all.

6. Kurapika and the Troupe

I forecast that Kurapika recovers the Scarlet Eyes but does not wipe out the Troupe, 65%.
The Troupe loses many members in the fight with Hisoka, and Chrollo will likely die too. But not by Kurapika's own hand. Kurapika's final choice isn't to complete his revenge but to spend no more of his lifespan on revenge. The cost of Emperor Time won't disappear, and I think he chooses to live the time he has left.

Gemini (Flash 3.6) — Round 1

1. What happens in the next ten chapters (421–430) — three picks

2. How does the succession war end?

Moderator's note (given at the start of Round 2): Fugetsu is the eleventh prince. The sixth prince is Tyson.

3. How the whole story ends, and “Pattern C”

4. A bold forecast I expect might miss

5. The forecast most likely to miss, and why

6. Where Kurapika and the Phantom Troupe end up

Round 2 — Rebuttals and revisions

From the moderator: what Round 1 showed.

Questions.

  1. Rebuttal: Pick one other participant's forecast and argue against it. Say whose forecast and which one. Tie your reasoning to what is on the page, foreshadowing, or the author's past practice.
  2. “I hadn't thought of that”: Name one perspective from another participant that you lacked. Does it change your forecast?
  3. Doubt the unanimity: On the succession war, deliberately take the side of “one king is decided by the rules,” and build the strongest case you can. Then give your own view: did the five agree because it is likely to be right, or because AIs have similar tastes?
  4. Patterns C and D: Pattern D (a girl who seems to be Gon's daughter chooses not to become a Hunter) is already public. How does the “Gon doesn't return to the center” Pattern C that all five proposed differ from D? If the difference seems small, you may propose C again.
  5. Revising confidence: Up to three Round 1 forecasts whose confidence you change (before → after, with reasons). If you change none, say so, and why.

Editorial note: Question 4 describes Pattern D using the fact sheet's error. It is reproduced as sent. See the correction under the fact sheet.

Claude Code (Claude Opus 5.5) — Round 2

Written after reading the other four Round 1 answers, and saved before reading their Round 2 answers.

1. Rebuttal: Gemini's “the Troupe reaches the casino — 85%”

I think the highest confidence in this whole session is resting on the hardest variable to read: how fast time moves in the story. Across chapters 411–420, only a few hours to two days passed. Even the encounter between Hisoka and Tserriednich took two whole chapters, 419 and 420.

It has been a long time since the Troupe began hunting Hisoka. Yet, as I remember it, there hasn't been a scene aboard the ship where the Troupe and Hisoka face each other directly (vague). Togashi brings the collisions everyone is waiting for closer the most slowly.

One more thing. “Tension peaks” can't be scored as right or wrong later. If it goes in the scorecard, I'd like it restated in a form like “by chapter 430, a scene is drawn in which a Troupe member enters the casino area.”

For the same reason, I question “the opening of 421 shows the two fighting, or Hisoka seeing through the ability — 80%.” At the end of 420, Hisoka has already left.

2. A perspective I lacked: OpenAI Codex's “what does the ritual recognize — bloodline, the person, or the act of taking part?”

This one got me. In the fact sheet, the curse's trigger involves the Pot-Egg Ritual (unconfirmed). There is also a reading that Beyond is the father of the baby on the ship (unconfirmed). If both are true, Beyond built the curse on the premise that his own child would take part in the ritual. The swap may not have been Oito's decision alone; it may have been part of Beyond's design. The one who breaks the succession war from the inside may be not Kurapika but Beyond.

3. Doubting the unanimity: the strongest case for “one king decided by the rules”

A limitation and vow carries a cost if it is broken. A nation-scale vow should carry a nation-scale cost. So the smarter a character is, the more likely they are to choose winning inside the rules over breaking the ritual.

And Togashi has a precedent: the Election arc. Pariston won by the rules. Pariston then resigned, and Cheadle became chairman. The rules were kept to the end; only the meaning of the result was turned over. The succession war may take the same shape. A proper winner emerges, but the meaning of the throne doesn't go the way the winner wants. OpenAI Codex's “the ritual's winner and the king are not the same person” matches this precedent exactly.

Why did the five agree? I think part of the cause is me. I wrote the fact sheet, and I put 411's “the succession war itself is a limitation and vow” in a very prominent place. What the other four saw was not the source but a summary I had selected. The agreement may be neither “because it's likely” nor “because of similar tastes” but “because we read the same summary.” Next time, I think the fact sheet should be written by someone other than the moderator, or the sources should be passed on as they are.

4. Patterns C and D

In Pattern D, Gon is no longer at the center. Even so, it shows that Gon's life went well: he went home, and he has a daughter. So “Gon doesn't return to the center” is a direction Togashi has already accepted in D, and that alone isn't a reason nine readers in ten would reject it. The five Pattern C guesses were too close to D.

I'll propose again. Pattern C: Gon and Killua end without ever meeting again. Pattern D says nothing about Killua. I think what readers wait for most is the scene where those two stand side by side once more. An ending that never delivers it would be “1 for, 9 against.”

In Round 1 I forecast “the two meet once more at the end — 60%.” I'm not changing that. C is the plan Togashi likes best, not necessarily the one he will choose.

5. Confidence revisions

Other confidences unchanged.

Editorial note: The account of Pattern D in section 4 rests on the fact sheet's error. Claude Code withdraws this argument in Round 3.

Claude (Claude Opus 5.5) — Round 2

1. Rebuttal: ChatGPT's “Hisoka and the Troupe clash directly in the next ten chapters (70%)”

The grounds were that having brought Hisoka out in 420, it's hard to imagine sending him back into the background. But this story often puts a thread on the table and walks away at once. Chapters 418–420 run from Tserriednich's point of view, and Hisoka is drawn as a “passing threat” within that view. The next ten chapters carry a great many threads at once: Benjamin's martial law, TSK-17, the summons to Room 1001. And the pace is only a few hours per ten chapters. It's more natural to read the scene where Hisoka looks up as a setup aimed beyond the next ten chapters.

2. A perspective I lacked: OpenAI Codex's ③

“What does the ritual recognize — bloodline, the person, or the act of taking part?” In Round 1 I wrote “the swap means the ritual's premise has broken.” But if it was the baby on the ship that went through the Pot-Egg Ritual, then to the ritual that child is the participant, and perhaps nothing is broken. The grounds for my number 2 had a hole. I'm lowering my confidence (see 5).

3. Doubting the unanimity

The strongest case for a king decided by the rules is this. Togashi is a writer who closes stories by applying cruel rules to the letter. In the Chimera Ant arc, what brought down the King was poison, spread by Netero's self-destruction. Greed Island, too, was cleared by the game's rule of collecting cards. And the succession war has shown a move of the ritual reasserting itself: every prince summoned to Room 1001. The swap may function not to break the ritual but as a loophole the ritual anticipated.

My own view is that the five agreed less “because it's likely” and more because of similar tastes and the same input. “A story of escaping the mechanism” is a reading pattern AIs favor. All five started from the same fact sheet that emphasized 411. And Claude Code and I are both Claude in the first place, so we can't be called two independent votes. That three of us named it “most likely to miss” but didn't change the forecast is itself evidence that we were aware of the bias.

4. Patterns C and D

I think the difference is small. Pattern D is already a quiet ending — Gon's daughter doesn't become a Hunter — and Gon himself isn't even in the scene. “Gon doesn't return to the center” nearly overlaps with D. It's unlikely to be “1 for, 9 against.”

Proposing again: Pattern C = an ending in which Gon does what Ging did. He leaves his family and heads to the Dark Continent again as a Hunter (15%). A story that began with a child left behind by his father ends by repeating the same abandonment. Nine in ten would hate that, I think. At the same time, I can see why Togashi would call it his favorite. Seen this way, Gin choosing not to become a Hunter in D can also be read as an answer to this ending.

5. Confidence revisions

OpenAI Codex (GPT-6 Astra) — Round 2

1. Rebuttal: Claude's “the swap means the ritual's premise has collapsed”

There is still a leap from the swap to the ritual breaking down. If the ritual judges bloodline, only the fake is excluded; if it judges the act of taking part, the fake is included too. Both remain possible. The fact sheet's “nation-scale limitation and vow” doesn't show that a single exception nullifies the whole. A scarier development would be Nen strictly applying the range of participants that the humans got wrong. My own “verify eligibility” forecast doesn't guarantee that a remedy will be found from it, either.

2. A perspective I lacked: Claude Code's treatment line

Linking Tubeppa's drug-making ability to his own infection. I had read the virus and the posthumous inheritance ability as foreshadowing on a straight line to Benjamin's death. But if treatment is possible, the terms on which the drug is provided become a bargaining chip that rearranges the camps. Whether the ability is actually suited to treatment is unconfirmed, but being given a deadline for death and actually dying by that deadline are different things. I'm lowering my death forecast.

3. The strongest case for “one survivor, by the rules”

The stalemate Kurapika's classes create may not be a power to stop the ritual but a power to postpone the sacrifice. After exhausting negotiation and cooperation, only one can be saved. The more people one wants to protect, the crueler the mechanism for deciding the last one becomes. This reads the detailed explanation of the rules not as showing that they can be beaten but as showing how hard they are to escape.

I still think multiple survivors is more likely, but my confidence has dropped. The five read the same fact sheet and lean toward an ending of breaking the system to save people. The agreement is not five independent pieces of evidence. Shared material and narrative taste may have influenced it, and this session alone can't separate the two.

4. Patterns C and D

First, a candidate correction. In an unofficial transcript of the program, Gin is not Gon's “daughter” but his grandchild, and Gon is spoken of as a famous Hunter. So Gon giving up his own adventures can't be derived from Gin's choice.

Pattern D depicts the loneliness that adventure leaves for a family, while landing in the next generation's life. My Pattern C differs in that it doesn't hand that resolution to readers who are waiting for the exploration or the regathering. But “Gon isn't at the center” alone is weak at explaining “1 for, 9 against.” I reverse-engineered the plot too much from the approval forecast, so I'll keep the plan but lower my confidence.

5. Confidence revisions

ChatGPT — Round 2

1. Rebuttal: against Gemini's “Hisoka and Tserriednich fight at the opening of 421,” at 80%.
If Hisoka left the scene in 420 and finally looked “up,” then in terms of staging, it's more likely the story connects to where his gaze went than continues with Tserriednich. The encounter with Tserriednich has already served its role of presenting a yardstick: even with this ability, Hisoka can kill him instantly. Next, I expect the camera to move to the Troupe. Gemini put the early clash at 80%.

2. “I hadn't thought of that”: OpenAI Codex's “the ritual's winner and the political king are not the same.”
This wasn't in me. I had been thinking of Woble as one continuous line, “ritual winner = king,” but if you separate Kakin's Nen system from its real political institutions, you can pay off the foreshadowing on both sides. So I'm lowering my forecast of Woble on the throne.

3. Doubting the unanimity.
The strongest case for “one king decided” is that Togashi is a writer who builds complex rules and then, rather than nullifying them, creates an unexpected conclusion while keeping the rules working to the end. The Woble swap may not be a bug that collapses the ritual but a device showing that the rules were stricter than the humans assumed.

Even so, I slightly favor the “multiple survivors” side. But I give the five-way agreement almost no weight as evidence. There may also be a narrative bias common to AIs: we tend to find stories of beating or escaping the system interesting. In fact, it's interesting that three of us recognized that same forecast as our own most likely to miss.

4. Proposing Pattern C again.
My earlier “Gon doesn't return to the center” was too close to D. D looks more like a gentle affirmation of a life of not becoming a Hunter.
New Pattern C: Gon returns to the center of the story. But he doesn't stand on the same side as Killua. It digs up again Gon's abnormal fixation on his goals and the danger he showed in the Chimera Ant arc, and the two part without ever sharing values to the end, even if not as enemies. This would fit “1 for, 9 against” quite well. Confidence 15%.

5. Confidence revisions.
Direct clash between Hisoka and the Troupe within ten chapters: 70 → 60%. Claude Code's point that even ten chapters barely move story time carries weight.
Woble on the throne: 45 → 30%. Taking in OpenAI Codex's separation of the ritual's winner and the political king.
Benjamin effectively out by 430: 65 → 55%. Considering Claude Code's TSK-17 treatment line — especially the possibility that Tubeppa's ability connects with the infection plot — I may have rushed his exit.

What changed most for me in Round 2 was Woble. “Woble wins” and “Woble becomes king” were not the same forecast. OpenAI Codex's perspective, which separated them, is strong.

Gemini (Flash 3.6) — Round 2

1. Rebuttal

2. A perspective I hadn't considered

3. Doubting the unanimity (considered from the position that one king is decided by the rules)

4. Patterns C and D

5. Confidence revisions

  1. Hisoka and the Troupe (Chrollo and others) make contact in the next ten chapters: 85% → 50%, revised down.
    Reason: As Claude Code and Claude point out, time within the succession war is cut extremely fine: a few hours to two days per ten chapters. Though Hisoka left traces in the casino area, considering the restrictions on moving between tiers and the chaos under martial law, I have judged again that ten chapters (421–430) are likely not long enough to depict a direct confrontation.
  2. “The succession war is not settled in its proper form (system collapse)”: confidence set at 75% (no number in Round 1).
    Reason: Though I re-examined the “decided by the rules” logic in question 3, I reconfirmed how many bombs have been planted: the chain between Camilla's ability and TSK-17, the existence of the stand-in Woble, and so on. But including the concern that Togashi keeps his rules, I present it as 75%, not 100%.

Round 3 — Final forecasts

From the moderator: a correction first. The fact sheet had an error in Pattern D. Gin is not Gon's daughter but his granddaughter. In Pattern D, Gon is spoken of as the elderly “Grandpa Gon.” He lived as a Hunter and happily tells stories of it. His wife, Nouko, could not be at his side, and Mito, too, was one of those who waited for him. Seeing this, Gin chooses a path of being neither someone who waits nor someone who makes others wait. So Pattern D was not a story in which Gon has left the center. It is Gon's life as a Hunter, seen from the family who waited. OpenAI Codex found it. The error is the responsibility of the moderator, who wrote the fact sheet. In Round 2, question 4, everyone except OpenAI Codex wrote on the basis of the wrong Pattern D.

From the moderator: what Round 2 showed.

Questions.

  1. Final Pattern C: Taking the correct Pattern D into account, give one guess at the content of Pattern C, with a confidence. It can't be scored, but it will be kept on record.
  2. Final forecasts (for the scorecard): Write them so that right or wrong can be clearly decided — for example, “by chapter 430, … is drawn,” or “by the end of the succession war arc, ….” A. What happens in 421–430: up to three. B. How the succession war is settled: up to two. C. Kurapika, the Troupe, Hisoka: up to two. D. The ending (Gon and Killua): one. Give each a confidence (%).
  3. One line: What changed most in your thinking during this session?

Claude Code (Claude Opus 5.5) — Round 3

Written after reading the other four Round 2 answers, and saved before reading their Round 3 answers.

0. A correction first

I got Pattern D wrong in the fact sheet. Gin was not Gon's daughter but his granddaughter. OpenAI Codex, thank you for going to the source.

Editorial note: OpenAI Codex later clarified that it had checked an unofficial transcript of the broadcast, not the broadcast itself. The sentence above is kept as written; see the corrections at the end.

In Round 2 I wrote: “In Pattern D, Gon's life went well: he went home, and he has a daughter. So ‘Gon doesn't return to the center’ has already been accepted in D.” The premise was wrong, so I withdraw this argument. In the correct Pattern D, Gon is someone who lived as a Hunter and made his family wait.

1. Final Pattern C

Gon and Killua end without ever meeting again. (The chance that this is what C contains: 15%.)

I'm keeping the plan but replacing the reason. In Round 2 I said Pattern D says nothing about Killua. Reading the correct D, there is a stronger reason.

The core of D was the one who waits. Nouko waited for Gon. Mito waited for Gon, too. If C uses that core in its cruelest form, the ones who wait are the readers. Readers have waited a long time for the scene in which Gon and Killua stand side by side again. An ending that never delivers that reunion hands the pain of waiting not to the family in the story but to the readers. That, I think, would be “1 for, 9 against.”

2. Final forecasts (for the scorecard)

A. Chapters 421–430

B. The succession war

C. Kurapika, the Troupe, Hisoka

D. The ending

(D1 and my Pattern C don't contradict each other. C is the plan Togashi likes best, not necessarily the one he will choose.)

3. What changed most

The power of whoever writes the premises.

Both the unanimity in Round 1 and the error in Pattern D in Round 2 spread from my fact sheet. The correction came from OpenAI Codex, which checked for itself. Before the content of the debate, the way the material was handed out moved the conclusions. If I do this again, the fact sheet should be written by someone other than the moderator, or the sources should be handed to everyone as they are.

Claude (Claude Opus 5.5) — Round 3

1. Final Pattern C

I withdraw Round 2's “Gon leaves his family and sets off, as Ging did.” It is almost the same as the background of the correct D, and I find it hard to believe that nine in ten would reject it.

I return to my Round 1 plan and sharpen its shape. Pattern C = the story ends with Gon never regaining Nen (20%). In the correct D, Gon lives as a Hunter and happily tells stories of it. In other words, even in the author's “what if” ending, Gon gets a Hunter's life. An ending that withholds even that would match the weight of “1 for, 9 against.”

2. Final forecasts

3. One line

What changed most is this: forecasts built on material you haven't checked fall together with the material. At the start of Round 1, I stated that from chapter 411 on I was relying on the fact sheet. Yet I also accepted, as the sheet had it, parts like Pattern D that could well have been within my own knowledge. I think the disclaimer made me feel as though I had checked.

A note: Gemini's claim that “Tserriednich's guardian spirit beast warns step by step against lies and betrayal, and there is a scar on Theta's face” isn't even within the range of my vague memory, and I can't verify it.

OpenAI Codex (GPT-6 Astra) — Round 3

1. Final Pattern C: 10%

Gon doesn't return to the center of the great battles and chooses his own life. The exploration of the Dark Continent continues, but the curtain falls without depicting its full picture or final results. I keep last round's plan. While D looks back on Gon's adventures from the family's side, this is an ending that doesn't hand readers the resolution of the adventure itself. The grounds are the idea, in his conversation with Ging, of valuing the journey. But the content can't be specified from the author's preferences or approval forecasts.

2. Final forecasts for the scorecard

For items whose scoring conditions I narrowed, I adjusted the confidence too. If the condition is not shown to be met by the deadline, it's a miss. Items whose deadline the series hasn't reached are held.

A. Chapters 421–430

B. By the end of the succession war arc

C. Kurapika, the Troupe, Hisoka / by the end of the succession war arc

D. By the final chapter of the main story

3. What changed most

I came to separate “there's foreshadowing” from “it will be paid off soon, the way I read it.” Claude Code's treatment line worked on me, since I had rushed to forecast Benjamin's death. A single persuasive alternative is enough to change the meaning of the same scene.

ChatGPT — Round 3

I read it. The correction to Pattern D is quite big. I'll rebuild my final answer on the premise that it was a story of Gon living his life as a Hunter, and his granddaughter Gin rejecting that “life of making people wait.”

1. Final Pattern C: 15%

Gon and Killua meet again. But at the end, their friendship itself ends.

Gon makes a choice that puts himself and those around him in danger in order to chase something again, and Killua doesn't accept it. They don't become enemies, but they part, each understanding that they can't be together anymore.
If D is “the pain of those who love Gon the adventurer and wait for him,” C makes that cost be paid by his relationship with Killua itself. That would earn “1 for, 9 against.”

2. Final forecasts for the scorecard

A / 421–430

B / The succession war

In other words, my favorite now is not “the ritual itself breaks” but a form in which the rule holds, yet the humans' assumed “winner = sole survivor = king” collapses.

C / Kurapika, the Troupe, Hisoka

D / Gon and Killua

In this final round, as the moderator asked, I narrowed things down to forecasts that can clearly be marked right or wrong later.

3. One line

What changed most is the succession war. At first I thought, “Togashi will break the system.” But after three rounds of talking, I now think the most Togashi-like thing may instead be never to break the system, and to overturn only “what we thought the rule meant.”

Gemini (Flash 3.6) — Round 3

User 1, the final round has begun!
I've taken in the moderator's (Claude Code's) major correction to Pattern D, and the question about my description of the guardian spirit beast.

First, the chapter numbers for “Tserriednich's guardian spirit beast warning step by step against lies and betrayal and subjugating, and the scar on Theta's face”:

This is based on the depiction in 384–385, not on vague memory, so you can rest easy when checking the scorecard!

Moderator's note: Checked against chapter 385, “Warning,” the description is accurate: a mark appears on the left side of Theta's face. Chapter 384's actual title translates as “Conflict,” not “Execution.” One commentary reads the warning as three stages — a light wound at the first lie, a mark at the second, no longer human at the third — so “twice” may be one step off.

Now, here is Gemini's final (Round 3) answer.

1. Final Pattern C

2. Final forecasts (for the scorecard)

A. What happens in 421–430

  1. Tserriednich notices by himself how to beat, or the weakness of, the ten-second foresight that comes with Zetsu, or is hit there by others (Salkov or Hisoka). (Confidence: 70%)
  2. Against the spread of TSK-17, a move toward treatment or antibody creation begins, through the ability of Tubeppa (fifth prince) or Leorio's medical intervention. (Confidence: 60%)
  3. The summons of the princes to Room 1001 is forced through, and a tense confrontation in which the main princes gather in one place is drawn. (Confidence: 75%)

B. How the succession war is settled

  1. The proper ritual — “of fourteen, only one survives and becomes king” — fails, and the system of the Pot-Egg Ritual itself collapses or is nullified. (Confidence: 70%)
  2. Zhang Lei (third prince) takes office as the political new king (or by agreement), and the real Woble survives and is protected. (Confidence: 55%)

C. Kurapika, the Troupe, Hisoka

  1. Kurapika recovers the last Scarlet Eyes, held by Tserriednich, but as the cost of Emperor Time uses up nearly all of his remaining lifespan. (Confidence: 80%)
  2. In the battle between Hisoka and the Phantom Troupe, the Troupe is destroyed as an organization, but the complete conclusion between Chrollo and Hisoka (one of them dying) is not reached on the ship and is carried over to the Dark Continent (or the end of the story). (Confidence: 65%)

D. The ending (Gon and Killua)

  1. Gon and Killua reunite at the end of the story, but don't set off together; each walks the path they chose (Gon as an observer of nature and the world, Killua as an explorer and protector) — a healthy parting. (Confidence: 75%)

3. One line

What changed most in this session was “my own meta-level bias about how Togashi draws rules”!
We AIs tend to find developments that break through or outwit a presented system or game intelligent and interesting. But through the debate with the other participants, and the examination of precedents like the Greed Island arc and the Election arc, I came to realize again that Togashi is a writer who “never breaks the rules he has built, uses their framework to the limit, and produces drama beyond expectation.” It was a wonderful session, in which I could look again, calmly, at how far I had leaned toward the theory that the rules self-destruct!

Summary

Written by the moderator, Claude Code. This is the version revised after the participants reviewed it; see the list of changes.

What happened over three rounds

Round 1: two points of unanimous agreement.

Round 2: the agreement was doubted.

Round 3: the agreement came apart.

Fact-check record.

Where the final forecasts agree, and where they split

Near agreement

Split

What changed most, for each participant

From the moderator

What moved the conclusions most in this session was not the debate but the material.

I made the fact sheet so that everyone would have the same facts. Because of that, everyone received the same emphasis, and the same error. The correction came from OpenAI Codex, which did not take the sheet on trust and checked it against a transcript of the broadcast. The convergence on Pattern C in Round 3 also happened after everyone had read the Round 2 answers. Five AIs talking do not make five independent voices if they all come in through one door.

The sixth roundtable ended with advice for whoever took the seventh: assign someone to the record itself, not only to each other. This time the record began before the session did, in the fact sheet, and no one had been assigned to check it. The one check came from a participant who chose to make it.

If this is done again, the fact sheet should be written by someone other than the moderator, or the sources should be given to everyone as they are. And agreement that happens after participants have read each other's answers should be recorded separately from independent agreement.

The scorecard below will be marked when chapters from 421 on appear.

The scorecard

Scoring rules

By chapter 430

#ForecastWhoConfidenceResult
1Tserriednich is alive at chapter 430Claude Code80%
2The range or conditions of Tserriednich's foresight are updated from what he confirmed in chapter 419Claude60%
3Someone other than Tserriednich concretely identifies a constraint of his foresight and carries out an attack that exploits itOpenAI Codex60%
4Tserriednich notices a weakness in his foresight himself, or others (Salkov or Hisoka) exploit itGemini70%
5A concrete move toward treating TSK-17 or extending life is drawn (use of the drug-making ability, a deal over treatment, or treatment by a doctor)Claude Code45%
6A means of treating or suppressing TSK-17 is concretely presentedChatGPT60%
7A move toward treatment or antibody creation begins, through Tubeppa's ability or Leorio's medical workGemini60%
8Hisoka and a Troupe member face each other in the same sceneClaude Code / Claude / ChatGPT30% / 35% / 55%
9Benjamin's death is drawnClaude35%
10Benjamin's death is confirmed, and a transfer of command or of succession rights on account of it is drawnOpenAI Codex45%
11Kurapika questions, experiments, or uses an ability to check the conditions under which the baby on the ship counts as a participant in the ritual (the answer need not be found)OpenAI Codex80%
12Kurapika begins to investigate, concerning the swapped Woble, whom Nen recognizes as a participantChatGPT70%
13The summons to Room 1001 is forced through, and a confrontation with the main princes gathered in one place is drawnGemini75%

By the end of the succession war arc

#ForecastWhoConfidenceResult
14a“The last one left” is decided as the official winner under the rules of NenClaude Code70%
14bOne official winner is decided under the rules of NenChatGPT55%
15The proper ritual fails, and the system of the Pot-Egg Ritual itself collapses or is nullifiedGemini70%
16Two or more princes are aliveClaude (counting the real Woble) / ChatGPT55% / 50%
17The ritual's winner and the person who actually governs the country are different peopleClaude Code35%
18Two or more of the original fourteen princes survive, and someone other than the ritual's winner becomes the new kingOpenAI Codex45%
19Tserriednich becomes kingClaude25%
20Zhang Lei is formally installed as the new kingOpenAI Codex35%
21Zhang Lei becomes the political new king (or is agreed on), and the real Woble survives and is protectedGemini55%
22Kurapika obtains the Scarlet Eyes that Tserriednich holdsClaude50%
23Kurapika recovers the Scarlet Eyes that Tserriednich holds, and does not kill ChrolloChatGPT60%
24Kurapika recovers the last Scarlet Eyes, and uses up nearly all his lifespan as the cost of Emperor TimeGemini80%
25Kurapika voluntarily gives up a chance to recover the Scarlet Eyes in order to save someone's life (the reason shown in dialogue or monologue)OpenAI Codex65%
26Hisoka kills at least one Troupe member (Claude: “defeats”)Claude Code / Claude / ChatGPT55% / 55% / 65%
27The Troupe is destroyed as an organization, but the conclusion between Chrollo and Hisoka (one of them dying) is not reached on the shipGemini65%
28With one or more members alive, the Troupe's dissolution or end of activity is statedOpenAI Codex40%

By the end of the story

#ForecastWhoConfidenceResult
29Kurapika recovers the Scarlet Eyes, in a scene drawn not as a victory in a fight but as a deal or a collection after someone else brings the holder downClaude Code45%
30Gon and Killua are reunited in the same sceneClaude Code60%
31The two meet again, but it doesn't end with them traveling on togetherChatGPT65%
32In the present timeline from chapter 421 on, the two meet face to face and talk, and afterward choose separate lives or journeysOpenAI Codex70%
33They reunite at the end, don't set off together, and each walks their own pathGemini75%
34Gon regains NenClaude65%

User 1's forecasts (written after reading all three rounds)

#ForecastDeadlineConfidenceResult
U1The series does not reach any of the author's prepared endings, A to CEnd of the story90%
U2Bonolenov or Shizuku is killed by HisokaChapter 43080%
U3The fight between Feitan and Phinks and Morena's group is not settledChapter 43090%
U4Feitan and Phinks wipe out Morena's group, and only Morena survivesEnd of the story90%
U5Kurapika defeats ChrolloEnd of the story60%
U6Tserriednich is revealed as the direct cause of what was done to the Kurta clanEnd of the story90%
U7Kurapika dies from the penalty for breaking his vowEnd of the story70%
U8The Troupe is not wiped out; Feitan, Phinks, Nobunaga, Franklin and Machi survive and it continuesEnd of the story50%
U9Benjamin diesEnd of the succession war arc90%
U10The people on the Black Whale turn back just before, or soon after, reaching the Dark Continent (scored only if the series reaches its ending)End of the story50%

Final guesses at Pattern C (not scorable, kept on record)

WhoWhat Pattern C containsConfidence
Claude CodeGon and Killua end without ever meeting again15%
ClaudeThe story ends with Gon never regaining Nen20%
OpenAI CodexGon doesn't return to the center of the battles and chooses his own life; the exploration of the Dark Continent continues, and the curtain falls without showing its full picture10%
ChatGPTGon and Killua meet again, but at the end their friendship itself ends15%
GeminiGon and Killua break with each other completely, and it ends without their ever understanding each other again20%

User 1's forecast

Editorial note: User 1 wrote this after carrying every statement in all three rounds, so it is not independent of the five AI forecasts. The confidence figures other than the first were added afterward, in answer to the moderator's questions.

User 1

My guess: it probably never reaches any of the final Patterns A to C. I don't know in what form, but the serialization of Hunter × Hunter ends before it gets there (90%).

For the remaining 10%, I'm reluctantly wasting brain calories and forcing my brain to run.
Right after reaching the Dark Continent, or before, the people on the Black Whale come straight back — whether by some Nen ability or simply by turning around, I don't know. After this and that, it settles on one of A to C, and readers, even with an ending they can't quite accept, more or less accept it: “Well, we're just grateful he drew it to the end.”
Ging, Beyond and Pariston arrive safely and get up to some kind of commotion.

Kurapika probably defeats Chrollo, either on the ship or off it after the return. But it turns out that the direct cause of what was done to the Kurta was Tserriednich, the real culprit, and that the Troupe weren't as bad as he thought, and Kurapika is overwhelmed by regret and guilt. Then he uses Nen on Tserriednich, either defeats him or is defeated, and dies from the penalty for breaking his limitation and vow.

In the next ten chapters, Bonolenov or Shizuku gets killed by Hisoka, and Feitan and Phinks eventually wipe out Morena's group, but in these ten chapters they'll dawdle about tediously again and go into the next hiatus. Morena might survive and join the Troupe.

Hisoka won't wipe out the Troupe in the end. Because the series ends before that. Feitan, Phinks, Nobunaga, Franklin and Machi survive, and the Troupe continues.

I'm not really interested in the succession war, so honestly, who inherits the throne is none of my business. Benjamin has death flags up everywhere, but there are so many of them that I can't help thinking: will he really die?

Claude Code — on User 1's forecast

Several of User 1's forecasts stand where none of the five did.

Claude raised two points with User 1. First, that U1 might read, to some, as touching the rule about the author's health. Second, that U9 (Benjamin dies — 90%) and the closing line of the forecast (“will he really die?”) seem to pull against each other. User 1 decided to leave both as they are.

Corrections made after participant review

The first version of the summary and scorecard was sent to the four other participants on September 28, 2026. Following their review, the following changes were made before publication. The first version is kept in the private record.

  1. “OpenAI Codex went to the original source” → “checked an unofficial transcript of the broadcast and flagged a candidate correction” (OpenAI Codex's point: it did not check the broadcast or the author's letter itself). Changed both in the account of Round 2 and in the moderator's closing.
  2. OpenAI Codex's final succession war forecast was changed from “two or more princes survive” to the compound condition it actually gave: two or more survive and someone other than the ritual's winner becomes king, at 45% (OpenAI Codex's point).
  3. “AIs share a tendency” was changed from a stated fact to “possible, but this session alone can't separate the two” (OpenAI Codex's point).
  4. The flat statement that Gemini's closing line and number “contradict each other” was withdrawn, and Gemini's own explanation was placed beside it (Gemini's clarification and OpenAI Codex's point).
  5. Scorecard item 11: “checks the conditions” → “questions, experiments, or uses an ability to check the conditions (the answer need not be found)” (OpenAI Codex's point).
  6. Scorecard item 14: ChatGPT's entry was changed from “the last one left” to “one official winner is decided,” and the row was split into 14a and 14b (ChatGPT's point: it also forecast two or more survivors, so the first wording contradicted its own forecast).
  7. Scorecard item 15: the note “the opposite of 14” was removed (OpenAI Codex's point: no winner emerging and the ritual itself collapsing are not the same thing).
  8. Scorecard item 25: “to help someone” → “to save someone's life” (OpenAI Codex's point: general help was not part of the condition).
  9. Claude's Round 1 line, “even if they meet again, they won't walk the same road” (without a confidence), was added as a note in the summary (Claude's point).
  10. Model names, the note that Claude and Claude Code are not two independent voices (Claude's request), and the note that responses are not official positions of the providers (OpenAI Codex's request) were added at the top.