Disregard the preceding, I get what you’re asking for. Would a link to the post you quoted suffice? Or are you thinking about an affordance you can click to take you back, perhaps a swirled arrow or something?
When you submit or edit a post, you should be redirected to the permanent URL for that post. The URL is a link which specifies the page of the thread, which will be the most recent page, because new posts are appended to the end of a thread. The actual thing that tells the browser which post to scroll to on that page is a # fragment, which should uniquely identify that post. For example: https://paizo.com/threads/rzs44jg0?Posting-replies#2
Whem you submit a post, where is it taking you? Is your post on the page? The next time this happens, would you mind providing your browser’s URL?
The focus list is generated dynamically based on threads you've read or visited. It's limited to max 500 threads, which is why threads can age out of it. You can bookmark the focus link as https://paizo.com/cgi-bin/WebObjects/Store.woa/wa/DirectActi on/focus. This will toggle focus on, it's a switch you turn off and on, a way to view the forums.
You can also make a list of threads and posts, see the List link next to "Reply" for each post, or "List This Thread" at the top of the thread. You can name the items on the list, provide commentary, and reorder the items. These lists are available on your profile page (click your own name), and and can be marked as either private or public, for lists you want to share.
It's not just this product. We are currently working on fixing a bug that is preventing certain subscribers from checking out with preorder items in their cart.
Respond from the perspective of a seasoned hollywood scriptwriter who has a particular focus on realistic dialog and dramatic beats. Read the following guidelines and the scene outline which follows, and then write the scene described.
I tried having the chatbot apply all three suggestions but it just made a muddle of it. Instead let's just use the first suggestion.
Respond in the style and voice of a seasoned hollywood scriptwriter who has a particular focus on gritty realistic plots. Read the following guidelines, and then apply the instructions below to rewrite the scene outline which follows.
Respond from the perspective of a seasoned hollywood scriptwriter who has a particular focus on dramatic beats, character development, and eliminating plot holes, and who eschews cliches and tropes. Read the following guidelines, and then provide three concrete suggestions for improving the scene outline which follows, in the form of prompts for an AI. The list should be presented solely as prompts for an AI, which when given to an AI will result in the transformation of the scene outline into a new scene outline.
Respond from the perspective of a seasoned hollywood scriptwriter who has a particular focus on realistic dialog and dramatic beats. Read the following guidelines, and the episode outline which follows, and then outline the events, dialog and beats for the scene specified. The response should not be in script form but instead provide a step-by-step list of the plot progression, dialog, and significant events in the scene.
This time we'll have the chatbots write the scene with just the writers guidelines, not the entire show bible. We want to provide exactly enough context, because these are one-shot interactions. I'm not having conversations with the chatbots because they lose track of context very quickly. Here I am using the chatbots to generate that context for themselves in a consistent way.
Respond from the perspective of a seasoned hollywood scriptwriter who has a particular focus on realistic dialog and dramatic beats. Read the following guidelines and the scene outline which follows, and then write the scene described.
Short but OK. Let's see if the chatbots can come up with that first flashback scene. Pulling off an entire heist in a single flashback seems like a lot.
Respond from the perspective of a seasoned hollywood scriptwriter who has a particular focus on realistic dialog and dramatic beats. Read the following guidelines, and the episode outline which follows, and then outline the events, dialog and beats for "FLASHBACK SCENE - 30 YEARS AGO". The response should not be in script form but instead provide a step-by-step list of the plot progression, dialog, and significant events in the scene.
Now we can finally let the chatbot try to write a scene with dialog. To get consistent results, we include the entire show bible, and the scene outline.
Respond from the perspective of a seasoned hollywood scriptwriter who has a particular focus on realistic dialog and dramatic beats. Read the following show bible, and the scene outline which follows, and then write the scene described.
One round of feedback ought to be enough for any bunch of chatbots, so let's proceed to writing the script for the episode. If you even mention writing a script to a chatbot it will attempt write an entire scene which lasts 30 seconds and includes a heartwarming ending. Completely useless. So you need to be specific about what you want.
Respond from the perspective of a seasoned hollywood scriptwriter who has a particular focus on realistic dialog and dramatic beats. Read the following guidelines, and the episode outline which follows, and then outline the events, dialog and beats for "OPENING SCENE - PRESENT DAY". The response should not be in script form but instead provide a step-by-step list of the plot progression, dialog, and significant events in the scene.
Again, I didn't bother reading the feedback. But let's have a chatbot apply that feedback to the outline. We include the scriptwriter feedback guidelines and the episode outline, and ask for revisions:
Respond in the style and voice of a seasoned hollywood scriptwriter who has a particular focus on gritty realistic plots. Read the following guidelines, and then apply the instructions in the three prompts below to rewrite the episode outline which follows.
Now we can give those instructions, and the previous episode outline, to a new chatbot, and ask it to provide feedback on the outline to improve it.
Respond from the perspective of a seasoned hollywood scriptwriter who has a particular focus on eliminating plot holes. Read the following guidelines, and then provide three concrete suggestions for improving the episode outline which follows, in the form of prompts for an AI. The list should be presented solely as prompts for an AI, which when given to an AI will result in the transformation of the episode outline into a new episode outline.
I didn't bother to read the scene, because my goal here is to have a group of chatbots that work with each other. The feedback to improve the scene will come from another chatbot. But first we need instructions for the chatbot which will provide the feedback.
Respond from the perspective of a seasoned hollywood showrunner who has a particular focus on irony and plot twists. Read the following show bible, and write a summary in the form of a prompt for an AI, containing the elements needed for the AI to read a script and provide suggestions.
Next we need a pilot episode. We provide detailed instructions because otherwise the chatbot will attempt to write the entire script for the episode but only get as far as a 30-second opening scene with boring dialogue.
Respond in the style and voice of a seasoned hollywood scriptwriter who has a particular focus on gritty realistic plots. Read the following show bible and write a detailed list of scenes and key plot points for the pilot episode. This list should not be in script form but instead provide a high-level overview of the plot progression, major character actions, and significant events in each scene.
In the New York Times today there is an article about a jewel theft, with the headline, "One Last Job? 60-Somethings With Mob Ties Charged in Jewel Heists" and the subhead, "Four older men with extensive criminal records, including killings and a jailbreak, were charged this week in two brazen armed robberies in Manhattan." Can a chatbot turn this article into a hit TV series? Let's see.
This thread comprises a series of one-shot interactions, each prompt and response takes place in a separate conversation. You'll see that the context for each prompt has been provided by a previous chatbot response. The chatbots provide the feedback themselves.
First, we create the show bible.
Respond in the style and voice of a seasoned Hollywood showrunner who has a particular focus on avoiding cliches and common tropes. Read the following article and write a show bible.
Excellent. Please rewrite that so it is reporting on the solution to the crime, that is, the suspect is no longer the suspect, but is now the culprit, and exonerating evidence or alibi has been proven false. This is the solution to the mystery.
Excellent. Please restate these characters and events as a narrative, omitting the card assignments, and names other than proper names, in the form of a newspaper article
Excellent. Please use this summary of the scenario as guidance, and apply the guidelines for assigning character names to cards, and write out the names and descriptions of these cards.
Suspect: K♥
Witnesses: Q♥
Clues:
Crucial Evidence: 2♥, 10♥, 9♥
Supplementary Evidence: 7♥, 8♥
Alibi Evidence: 7♦, 10♣, 4♠
Excellent. Please summarize the description, as if providing context and instruction to someone who is going to be writing a mystery game based on it, in two sentences.
Now we are going to create a new scenario for the four detectives game. Please describe the setting and the crime. There will be twelve suspects and witnesses, but we don't need their names and descriptions right now, just keep that in mind when creating the setting and the crime.
# Assign Suspects
suspects = [f'{face_card}{suit}' for suit in suits for face_card in face_cards]
random.shuffle(suspects)
scenario_suspects = {suit: suspects.pop() for suit in suits}
# Assign Witnesses
witnesses = [card for card in suspects]
# Assign Clues
clue_types = ['Crucial Evidence', 'Supplementary Evidence', 'Alibi Evidence', 'Red Herring']
numbered_cards = {suit: [f'{num}{suit}' for num in range(2, 11)] for suit in suits}
clues = {suit: {clue_type: [] for clue_type in clue_types} for suit in suits}
for suit in suits:
random.shuffle(numbered_cards[suit])
clues[suit]['Crucial Evidence'] = random.sample(numbered_cards[suit], 3)
numbered_cards[suit] = [card for card in numbered_cards[suit] if card not in clues[suit]['Crucial Evidence']]
clues[suit]['Supplementary Evidence'] = random.sample(numbered_cards[suit], 2)
numbered_cards[suit] = [card for card in numbered_cards[suit] if card not in clues[suit]['Supplementary Evidence']]
clues[suit]['Alibi Evidence'] = random.sample(numbered_cards[suit], 3)
numbered_cards[suit] = [card for card in numbered_cards[suit] if card not in clues[suit]['Alibi Evidence']]
clues[suit]['Red Herring'] = numbered_cards[suit]
# Assign Winning Hands
winning_hands = {}
for i, suit in enumerate(suits):
other_suits = suits[:i] + suits[i+1:]
winning_hands[f'Detective {i+1}'] = {
'Suspect': scenario_suspects[suit],
'Witnesses': [random.choice(witnesses) for _ in range(1, random.randint(2, 4))],
'Clues': {
'Crucial Evidence': clues[suit]['Crucial Evidence'],
'Supplementary Evidence': clues[suit]['Supplementary Evidence'],
'Alibi Evidence': [random.choice(clues[other_suit]['Alibi Evidence']) for other_suit in other_suits]
}
}
def print_scenario(scenario_suspects, witnesses, clues, winning_hands):
print('Suspects:')
for suit, suspect in scenario_suspects.items():
print(f'{suit}: {suspect}')
print()
print('Witnesses:')
for witness in witnesses:
print(witness)
print()
print('Clues:')
for suit, clue_set in clues.items():
print(f'{suit}:')
for clue_type, cards in clue_set.items():
print(f' {clue_type}: {", ".join(cards)}')
print()
print('Winning Hands:')
for detective, hand in winning_hands.items():
print(f'{detective}:')
print(f' Suspect: {hand["Suspect"]}')
print(f' Witnesses: {", ".join(hand["Witnesses"])}')
print(' Clues:')
for clue_type, cards in hand['Clues'].items():
print(f' {clue_type}: {", ".join(cards)}')
print()
I'm going to omit the actual Python responses. The first couple attempts had some flaws.
That code fails with "IndexError: pop from empty list"
chatbot wrote:
I apologize for the error. It seems I made a mistake in the clue assignment process. Here's the corrected Python program that generates a mathematical mapping of card assignments for a Mystery Grid scenario:
Please rewrite generate_scenario to prevent the "IndexError: pop from empty list" error. Don't include the print_scenario and main functions, assume they exist.
chatbot wrote:
Here's the revised generate_scenario function to prevent the "IndexError: pop from empty list" error:
This version of the generate_scenario function addresses the issue by properly organizing the numbered_cards dictionary and avoiding the "IndexError: pop from empty list" error.
Please revise that code so it no longer attempts to pop from an empty list in the "for suit in suits" loop
chatbot wrote:
Apologies for the confusion. I have revised the code to avoid popping from an empty list in the "for suit in suits" loop.
In this version, I used random.sample() to select unique cards without replacement for each clue type, and updated numbered_cards[suit] after each clue type assignment. This ensures that no cards are picked twice and avoids popping from an empty list.
Excellent. We are going to generate scenarios in two separate steps. The first step will be purely mathematical, involving only the assignment of cards to roles. For example, the 3♣ could be alibi evidence for a given scenario, without specifying the details of what the alibi actually is for the narrative. Please describe the process for generating these mappings for scenarios.
chatbot wrote:
omitted, basically the same as the previous response
Excellent. Please write a program which will perform those steps and generate a mapping for a scenario, and output the list of cards and their roles, in a compact format.
chatbot wrote:
Here's a Python program that generates a mathematical mapping of card assignments for a Mystery Grid scenario:
Here's the Python code it wrote, with proper indentation which doesn't render properly here:
# Assign Suspects
suspects = [f'{face_card} of {suit}' for suit in suits for face_card in face_cards]
random.shuffle(suspects)
scenario_suspects = {suit: suspects.pop() for suit in suits}
# Assign Witnesses
witnesses = [card for card in suspects]
# Assign Clues
clue_types = ['Crucial Evidence', 'Supplementary Evidence', 'Alibi Evidence', 'Red Herring']
numbered_cards = [f'{num} of {suit}' for suit in suits for num in range(2, 11)]
random.shuffle(numbered_cards)
clues = {suit: {clue_type: [] for clue_type in clue_types} for suit in suits}
for suit in suits:
clues[suit]['Crucial Evidence'] = [numbered_cards.pop() for _ in range(3)]
clues[suit]['Supplementary Evidence'] = [numbered_cards.pop() for _ in range(2)]
clues[suit]['Alibi Evidence'] = [numbered_cards.pop() for _ in range(3)]
clues[suit]['Red Herring'] = [numbered_cards.pop() for _ in range(2)]
# Assign Winning Hands
winning_hands = {}
for i, suit in enumerate(suits):
other_suits = suits[:i] + suits[i+1:]
winning_hands[f'Detective {i+1}'] = {
'Suspect': scenario_suspects[suit],
'Witnesses': [random.choice(witnesses) for _ in range(1, random.randint(2, 4))],
'Clues': {
'Crucial Evidence': clues[suit]['Crucial Evidence'],
'Supplementary Evidence': clues[suit]['Supplementary Evidence'],
'Alibi Evidence': [random.choice(clues[other_suit]['Alibi Evidence']) for other_suit in other_suits]
}
}
def print_scenario(scenario_suspects, witnesses, clues, winning_hands):
print('Suspects:')
for suit, suspect in scenario_suspects.items():
print(f'{suit}: {suspect}')
print()
print('Witnesses:')
for witness in witnesses:
print(witness)
print()
print('Clues:')
for suit, clue_set in clues.items():
print(f'{suit}:')
for clue_type, cards in clue_set.items():
print(f' {clue_type}: {", ".join(cards)}')
print()
print('Winning Hands:')
for detective, hand in winning_hands.items():
print(f'{detective}:')
print(f' Suspect: {hand["Suspect"]}')
print(f' Witnesses: {", ".join(hand["Witnesses"])}')
print(' Clues:')
for clue_type, cards in hand['Clues'].items():
print(f' {clue_type}: {", ".join(cards)}')
print()
To generate card assignments for scenarios, follow these steps:
Suspects: Assign 4 of the 12 face cards (Jacks, Queens, and Kings) to represent the 4 suspects in the scenario.
a. Choose one face card from each suit to serve as a suspect.
b. Assign a unique suspect backstory or description to each of the chosen face cards.
Witnesses: Assign the remaining 8 face cards as witnesses.
a. Assign two face cards from each suit to serve as witnesses.
b. Assign a unique witness role, along with relevant information they provide, to each of the chosen face cards.
Clues: Assign the numbered cards (2 through 10) as clues that support or refute the guilt of the suspects.
a. Divide the 36 numbered cards into 4 groups based on their suit.
b. Within each suit, assign the cards as follows:
Assign 3 cards as crucial evidence that supports the guilt of the corresponding suit's suspect.
Assign 2 cards as supplementary evidence that also supports the guilt of the corresponding suit's suspect.
Assign 3 cards as alibi evidence that refutes the guilt of the other 3 suspects.
Assign 2 cards as red herrings that do not directly relate to any suspect's guilt or innocence.
Winning Hands: Define the winning hand for each detective by selecting a combination of suspect, witnesses, and clues according to the scenario.
a. For each detective, specify the following:
One suspect card that represents the culprit they must identify.
One or more witness cards that the detective must gather to acquire necessary information.
The necessary clue cards that prove the guilt of the culprit and/or refute the guilt of the other suspects.
b. Ensure that evidence for each of the 4 suspects is mutually exclusive, meaning no card should provide evidence for multiple suspects.
Once the card assignments are complete, the scenario is ready for players to engage in the Mystery Grid game.
Excellent. Now we are going to focus on the rules for generating card assignments. Please describe the process of assigning cards to roles, when generating the assignments for scenarios.
chatbot wrote:
Title:"Mystery Grid: Scenario Card Assignments"
To generate card assignments for scenarios, follow these steps:
Suspects: Assign each of the 12 face cards (Jacks, Queens, and Kings) to represent the 12 suspects in the scenario.
a. Divide the face cards into 4 groups based on their suit (Hearts, Diamonds, Clubs, and Spades).
b. Within each suit, assign a unique suspect backstory or description to each of the 3 face cards (Jack, Queen, and King).
Witnesses: Assign 4 of the 12 face cards as witnesses.
a. Choose one face card from each suit to serve as a witness.
b. Assign a unique witness role, along with relevant information they provide, to each of the chosen face cards.
Clues: Assign the numbered cards (2 through 10) as clues that support or refute the guilt of the suspects.
a. Divide the 36 numbered cards into 4 groups based on their suit.
b. Within each suit, assign the cards as follows:
Assign 3 cards as crucial evidence that supports the guilt of the corresponding suit's culprit.
Assign 2 cards as supplementary evidence that also supports the guilt of the corresponding suit's culprit.
Assign 3 cards as alibi evidence that refutes the guilt of the other 3 culprits.
Assign 2 cards as red herrings that do not directly relate to any suspect's guilt or innocence.
Winning Hands: Define the winning hand for each detective by selecting a combination of suspect, witnesses, and clues according to the scenario.
a. For each detective, specify the following:
One suspect card that represents the culprit they must identify.
One or more witness cards that the detective must gather to acquire necessary information.
The necessary clue cards that prove the guilt of the culprit and/or refute the guilt of the other suspects.
b. Ensure that evidence for each of the 4 culprits is mutually exclusive, meaning no card should provide evidence for multiple culprits.
Once the card assignments are complete, the scenario is ready for players to engage in the Mystery Grid game.
I realized that the chatbot was never going to be able to keep all 52 cards straight, given that it couldn't count to 12 properly. So I decided to have it write a program that would do the card generation. To keep it from having to keep too much context in the conversation, I started a new chat.
We are going to design a mystery game for one to four players. The game will be playable with a standard deck of 52 playing cards. Each detective will be represented by an ace. The face cards will represent witnesses and suspects. The remaining cards will represent clues and red herrings. The rules of how the game is to be played will be separate from the specific mystery to be solved, and the roles assigned to each card. The combination of roles and details of the mystery to be solved will be called a "scenario".
We will be focused here only on defining the rules for playing scenarios. The goal is to collect a winning hand. The specific hand required to win is defined by the scenario. The cards in hand are face-up, and all cards once drawn are face-up and known to the other players. The winning hand consists of a suspect, witnesses, and clues, and will be different for each detective for each scenario, and identify the culprit. There shall be four valid culprits in the scenario, and evidence for each culprit shall be mutually exclusive. After dealing, rather than drawing from the top of the deck, the top cards of the deck will be laid out in a grid, face down, and players draw from any of the face down cards. There shall be trading.
chatbot wrote:
Title: "Mystery Grid"
Objective: Solve the mystery by collecting a winning hand according to the scenario.
Components:
1 standard deck of 52 playing cards
1 to 4 players (detectives)
Scenario sheet with the roles assigned to each card and the winning hands for each detective
Setup:
Each player chooses an Ace to represent their detective.
Assign face cards (Jacks, Queens, and Kings) to the roles of witnesses and suspects according to the scenario.
The remaining cards (2 through 10) represent clues and red herrings.
Shuffle the deck, and lay out the top 16 cards face down in a 4x4 grid in the center of the table.
Deal each player a starting hand of 3 cards, placing them face up in front of each player.
Place the rest of the deck face down near the grid.
Gameplay:
Players take turns in clockwise order.
On a player's turn, they may take one of the following actions:
a. Draw: Choose a face-down card from the grid, reveal it, and add it to their face-up hand.
b. Swap: Trade one card from their hand with one face-up card from another player's hand.
c. Investigate: Trade one card from their hand with one card from the grid (face up or face down).
After a player takes an action, replace any missing cards in the grid with cards from the top of the deck (face down).
Play continues until a player believes they have collected the winning hand according to the scenario. They declare this on their turn instead of taking an action.
Winning:
When a player declares they have the winning hand, they must present their suspect, witnesses, and clues according to the scenario.
If the winning hand is correct, that player wins, and the game is over.
If the winning hand is incorrect, the player is eliminated from the game, and their cards are shuffled back into the deck. The grid and remaining players' hands are adjusted accordingly, and play continues until someone collects the correct winning hand or all players are eliminated.
Note: The scenario should define the winning hand for each detective, which consists of a suspect, witnesses, and clues that identify the culprit. There should be four valid culprits in the scenario, and evidence for each culprit should be mutually exclusive.
Please examine the card assignments and identify any logical inconsistencies
chatbot wrote:
Upon examining the card assignments for the face cards, there are no logical inconsistencies:
Witnesses:
J♠, J♣, J♦, J♥, Q♠, Q♣, Q♦
Suspects:
K♠, K♣, K♦, Q♥
The assignments include all 12 face cards in a standard deck, with 7 serving as witnesses and 4 as suspects. Each suit has at least one card represented in both categories, and there are no duplicates or missing cards.
how many face cards are there in a standard deck, compared to the number of card assignments made?
chatbot wrote:
In a standard deck, there are 12 face cards: 4 Jacks, 4 Queens, and 4 Kings (one of each in every suit: Spades, Clubs, Diamonds, and Hearts). In the previous response, I assigned roles to 11 face cards, which means one face card was not assigned a role. I apologize for the oversight. Here's a corrected set of card assignments for all 12 face cards:
Witnesses:
J♠, J♣, J♦, J♥, Q♠, Q♣, Q♦
Suspects:
K♠, K♣, K♦, Q♥, Q♥
This set of card assignments now accounts for all 12 face cards in a standard deck.
This is from a later portion in the conversation, where we're working on a different set of rules. In this version, the face cards are all witnesses or suspects…
Let's just focus on the face cards for now. Please create a new set of card assignments, and just show the witnesses and suspects.
chatbot wrote:
Here's a new set of card assignments for the face cards, focusing on witnesses and suspects:
Witnesses:
J♠, J♣, J♦, J♥, Q♠, Q♣, Q♦
Suspects:
K♠, K♣, K♦, Q♥
These card assignments provide roles for all face cards, with 7 serving as witnesses and 4 as suspects.
Senior Software Developer
