You’re starting an AI lab. Pick seven people to build its first model, with 20 game points to spend.
Start with any role.
Choose one person for each role. Open Your team to jump to any seat and see everyone available. After a pick, we take you to the first empty seat so nobody gets missed.
You don’t need to know the names already. There is no correct team. Change any pick if you change your mind.
The game keeps enough points to fill your remaining seats. See how points work.
- ChiefSets the direction.
- ResearchFinds the next idea.
- Post-trainingTeaches the model new skills.
- InfraMakes the systems run.
- SafetyWorks on safer behavior.
- EvalsChecks what works.
- WildcardAdds something unexpected.
Your seventh pick
makes the card.
Your team gets a name, a little commentary and a card with all seven people. Edit it, save it, or share it wherever your people are.
The image and shared team link show your seven picks. Anyone opening the link can make their own version. The question is simple: which pick would you change?
Finish below 20 points and you get a Bootstrapped badge showing exactly what you saved. Spending all 20 is fine too. Sharing is always optional.
Real research is bigger than a roster.
A starting point for curiosity.
A breakthrough can come from a student, an unknown researcher or a team this game has never heard of. Research also depends on collaborators, ideas, data, tools, time and luck. Seven public names cannot capture all of that.
A game, not a prediction.
The 108 people here are a small set of public references, not a list of the best scientists. Their cards can introduce you to their work. Their points cannot tell you who will make the next breakthrough, or whether your chosen people would work together.
Made for the fun of it.
Frontier Draft is a parody fantasy game. It is not affiliated with or endorsed by anyone who appears in it. Card details describe public work. Points are provisional game settings for fictional recruitment difficulty and opportunity cost, never a measure of ability, quality, prestige or predicted future impact. Points are not literal salary estimates or claims about anyone’s willingness to move. See How points work.
The jokes concern combinations, situations and the player’s choices. They do not attack a named person’s competence, ethics or character. References to disagreements concern public exchanges.
Game points, set by hand.
Points represent fictional recruitment difficulty and opportunity cost. A 5-point researcher is harder or more expensive to recruit in this fictional market than a 1-point researcher, not a better scientist. Points do not measure ability, quality, prestige or future impact, and are not literal salary estimates.
Something to change?
If you appear in this game and want out, message me and I’ll remove your card within 48 hours. No argument, no negotiation, no asking why. Message @falseconjecture on X for removal, a work-description correction, a point cost you want to question, or feedback on the rules.
Everyone on a card is a voluntarily public professional figure, described by publicly documented achievements only. No minors, no private individuals, no personality claims or rumours. Organization references describe documented work or founding roles, not a live employment directory.
Play on your terms.
There is no account, timer, purchase or share gate. No streaks, fake scarcity or artificial waiting. Optional challenges give you one extra goal, checked as you pick.
Your last lab, draft, completed challenges, appearance and sound preferences stay on this device. Privacy details.
Who’s in your lab?
Seven people. Twenty points. A team that’s yours to choose.