Wold Newton
Villager
Archivist AI Data Busts the “Murder Hobo” Myth: Real Campaigns Are Overwhelmingly Good, and Paladins Are Suddenly Cool
New analysis of real tabletop campaigns finds Evil-aligned characters are nearly extinct, while Paladins and Bards surge past Fighters and Wizards under the D&D 2024 rules.
SAN LUIS OBISPO, CA, August 24, 2026 – Archivist AI, the campaign memory platform built for tabletop roleplaying game groups, today released new findings drawn from its dataset of real, in-progress tabletop campaigns, built from the actual characters players are bringing to the table. The headline: the hobby’s reputation for chaotic, morally grey antiheroes doesn’t hold up, and the newest edition of Dungeons & Dragons is reshaping which classes players reach for.
Archivist has spent the last year turning session content into structured campaign records for thousands of tables. That same record-keeping now offers something rarer in the TTRPG space: a real, aggregate look at who players actually choose to be, rather than who the internet assumes they are.
The murder hobo is basically extinct
Across the analyzed D&D 5e and D&D 2024 characters, only about 2% have an Evil alignment, while nearly two-thirds skew Good. The chaotic-neutral rogue who burns down the tavern for fun remains a meme rather than the majority – the data show most tables are populated by earnest, good-hearted characters, whatever their class or backstory. A thief with a heart of gold? Call it the Chris Pine Effect.
Paladins and Bards are having a moment
The shift is sharpest at the class level. Under the classic 5e ruleset, Paladins accounted for roughly 5.5% of characters. Under the newer D&D 2024 rules, that number has climbed to about 7% – enough to leapfrog both Fighter and Wizard into the top three classes. Bard has climbed alongside it, while the classic “lone adventurer” archetypes have quietly lost ground - less Strider, more Shakespeare. Something is pulling players toward oath-bound, party-first, support-driven characters instead of the classic damage-dealer.
“We didn’t set out looking for this – it just kept showing up in the data,” said Greg Wood, co-founder of Archivist. “The image of three guys in the basement running around dungeons, trying to kill anything that moves is an outmoded stereotype that quite frankly probably only existed in very small percentages, despite the outsize public perception. The data says most tables are looking for something quite different, and it says it pretty loudly.”
What’s Behind The Shift?
Archivist doesn’t collect demographic data, so the “why” behind the trend is still an open question. Theories on the table include:
Talk trends among GMs
Across 54.7 million words of analyzed session audio and transcripts, Archivist discovered that GMs speak an average of 46.5% of a table's words, with a standard deviation of just 12 points — meaning most tables cluster tightly around that midpoint rather than spreading evenly across the full range. Eight in ten tables fall between 31.3% and 61.6% GM share, and the extremes are rare: only 1.6% of sessions have a GM speaking less than a fifth of the table's words, and just 3.7% see a GM take up more than 70%.
The pattern holds remarkably steady regardless of system, party size, or play style: a GM at one-third of the talking or two-thirds is unusual, and most sessions settle in that 40–50% band no matter what's being played. At an average of 21,000 words spoken per voice session, that means any group recording their own game will see exactly where they land on this curve within a single afternoon of play.
A Deeper Dive
This is the first release in what Archivist plans as an ongoing look at trends across its dataset of real campaigns. Additional findings on class, species, and cast composition are available on the Archivist AI website.
About Archivist AI
Archivist AI is the campaign memory platform built for tabletop roleplaying game groups. Founded in 2025 by Greg Wood and Greg Harzelak, Archivist captures, organizes, and surfaces everything that happens across a campaign – session transcripts, character histories, factions, quests, locations, and world lore – so players and Game Masters can spend their time at the table instead of managing notes. Archivist supports more than 250 RPG systems, has processed more than 20,000 sessions, and integrates with Discord, Foundry VTT, Obsidian, and other tools through its MCP platform. The company has publicly committed to never building an AI Game Master. More information is available at myarchivist.ai.
New analysis of real tabletop campaigns finds Evil-aligned characters are nearly extinct, while Paladins and Bards surge past Fighters and Wizards under the D&D 2024 rules.
SAN LUIS OBISPO, CA, August 24, 2026 – Archivist AI, the campaign memory platform built for tabletop roleplaying game groups, today released new findings drawn from its dataset of real, in-progress tabletop campaigns, built from the actual characters players are bringing to the table. The headline: the hobby’s reputation for chaotic, morally grey antiheroes doesn’t hold up, and the newest edition of Dungeons & Dragons is reshaping which classes players reach for.
Archivist has spent the last year turning session content into structured campaign records for thousands of tables. That same record-keeping now offers something rarer in the TTRPG space: a real, aggregate look at who players actually choose to be, rather than who the internet assumes they are.
The murder hobo is basically extinct
Across the analyzed D&D 5e and D&D 2024 characters, only about 2% have an Evil alignment, while nearly two-thirds skew Good. The chaotic-neutral rogue who burns down the tavern for fun remains a meme rather than the majority – the data show most tables are populated by earnest, good-hearted characters, whatever their class or backstory. A thief with a heart of gold? Call it the Chris Pine Effect.
Paladins and Bards are having a moment
The shift is sharpest at the class level. Under the classic 5e ruleset, Paladins accounted for roughly 5.5% of characters. Under the newer D&D 2024 rules, that number has climbed to about 7% – enough to leapfrog both Fighter and Wizard into the top three classes. Bard has climbed alongside it, while the classic “lone adventurer” archetypes have quietly lost ground - less Strider, more Shakespeare. Something is pulling players toward oath-bound, party-first, support-driven characters instead of the classic damage-dealer.
“We didn’t set out looking for this – it just kept showing up in the data,” said Greg Wood, co-founder of Archivist. “The image of three guys in the basement running around dungeons, trying to kill anything that moves is an outmoded stereotype that quite frankly probably only existed in very small percentages, despite the outsize public perception. The data says most tables are looking for something quite different, and it says it pretty loudly.”
What’s Behind The Shift?
Archivist doesn’t collect demographic data, so the “why” behind the trend is still an open question. Theories on the table include:
- A post-pandemic pull toward “Found family” storytelling
- The influence of narrative-driven actual-play shows that reward camera-friendly heroics over combat optimization
- A broader cultural turn toward cozier, less grimdark entertainment, or
- A simpler mechanical explanation – the 2024 rules made Paladins and Bards more flexible to build and play.
Talk trends among GMs
Across 54.7 million words of analyzed session audio and transcripts, Archivist discovered that GMs speak an average of 46.5% of a table's words, with a standard deviation of just 12 points — meaning most tables cluster tightly around that midpoint rather than spreading evenly across the full range. Eight in ten tables fall between 31.3% and 61.6% GM share, and the extremes are rare: only 1.6% of sessions have a GM speaking less than a fifth of the table's words, and just 3.7% see a GM take up more than 70%.
The pattern holds remarkably steady regardless of system, party size, or play style: a GM at one-third of the talking or two-thirds is unusual, and most sessions settle in that 40–50% band no matter what's being played. At an average of 21,000 words spoken per voice session, that means any group recording their own game will see exactly where they land on this curve within a single afternoon of play.
A Deeper Dive
This is the first release in what Archivist plans as an ongoing look at trends across its dataset of real campaigns. Additional findings on class, species, and cast composition are available on the Archivist AI website.
About Archivist AI
Archivist AI is the campaign memory platform built for tabletop roleplaying game groups. Founded in 2025 by Greg Wood and Greg Harzelak, Archivist captures, organizes, and surfaces everything that happens across a campaign – session transcripts, character histories, factions, quests, locations, and world lore – so players and Game Masters can spend their time at the table instead of managing notes. Archivist supports more than 250 RPG systems, has processed more than 20,000 sessions, and integrates with Discord, Foundry VTT, Obsidian, and other tools through its MCP platform. The company has publicly committed to never building an AI Game Master. More information is available at myarchivist.ai.
