Chapter — Arbor and the Tree of Every Future

Arbor sees games as trees, and she means it literally — she can’t help sketching them, branches spreading across any scrap of paper she can find.

“Here’s now,” she says, drawing a single dot. “You have four moves.” Four lines branch out from the dot. “Pick any one — say this. Now it’s my turn, and I have five replies.” Five more lines fan from the end of each. “And from each of those, more branches, and more, spreading out and out — every possible future of this game, growing from this one dot like a tree.” She taps the sprawling sketch. “When someone says ‘think ahead,’ this is what they mean — climbing a little way up the tree. And when someone says a game is solved —” her eyes shine — “they mean somebody, usually a computer, climbed the whole tree, all the way to every leaf, and knows, for certain, who wins if nobody makes a mistake.”

Arbor was the kid who drove people mildly crazy asking “but then what?” She couldn’t take a plan at face value — she always wanted to know the next step, and the step after that. For a while it seemed like worry. Then she realized it was a superpower with no off switch: she was building trees in her head, following each choice to its consequences, and it meant she was rarely surprised. She’d already visited the futures. She wasn’t anxious about what might happen. She’d walked what might happen, several branches deep, and come back knowing which paths were traps.

“Every choice is a branch,” she says, “and most people only look at the branch they’re standing on. But the whole tree is right there, waiting to be climbed. You don’t have to climb all of it — nobody can, some trees are bigger than all the atoms in the universe. But climbing even three or four branches ahead, and imagining your opponent picking their best reply at each fork — not their laziest, their best — that’s the difference between reacting and seeing.”

She came to the GridForge academy because Gridley, the family host, needed the computational-thinking capstone taught — the “how a game gets solved” story that ties every grid game together. On her first day Gridley asked her to explain why Connect-4 is “a first-player win.” Arbor drew the tree. “It means someone climbed the entire Connect-4 tree,” she said, “every possible game, and found that if the first player plays perfectly — always the best branch — the second player cannot stop them, no matter what. It’s not that the first player usually wins. It’s that with perfect play they win always, and we know it, because the whole tree was explored.” She showed how the value of a position gets figured: “You look at the leaves — the finished games, win or lose — and you back the answer up the tree: at each fork, assume each player picks the branch that’s best for them, and the value flows down to now.” Gridley, who wanted students to glimpse how minds and machines think ahead, said quietly, “You didn’t just plan a move. You explained how a game gets known all the way down.”

In her workshop she teaches students to think in branches. A boy came in stuck: “I can’t plan ahead, there are too many possibilities, my brain just freezes.” Arbor nodded warmly. “Because you’re trying to climb the whole tree at once,” she said. “Nobody can. Even computers can’t climb a big tree fully — they climb part of it, cleverly.” She drew a small tree. “The trick is: don’t look at every branch. Look at the few good ones. At each fork, ask ‘what are my two or three best moves?’ and follow only those. And crucially —” she tapped the opponent’s forks — “at their turns, imagine them playing their strongest reply, not the one you hope they’ll make. Plan against their best, and you’re never ambushed.” The boy tried following just two branches, three moves deep, assuming his opponent’s best each time — and found that the freeze lifted, because a tree of a few good branches is climbable, where a tree of everything is not.

“That’s how thinking ahead actually works,” Arbor told him, “for you and for the machines. You can’t see every future. But you can see the important ones, if you prune away the silly branches and always assume the other player is smart. See a few branches clearly, beats seeing all of them not at all.”

Later the boy asked the question. “But if computers can solve games completely — if they know who wins — then what’s the point of me playing? Isn’t it already decided?”

Arbor smiled, because it was the question she loved most. “Only a few games are fully solved,” she said, “and only small ones — the trees of chess or Go are so enormous that not all the computers on Earth could climb them before the sun burns out. So ‘solved’ is rare and precious, not the normal case.” She tapped the sketch. “But even for a solved game — even Connect-4, where perfect play is known — you haven’t solved it. The point isn’t to be the computer. The point is to climb the tree yourself, a little farther each time, until you can see the traps and the wins. The tree was always there. Solving is just knowing it completely, and thinking ahead is knowing it partly — and knowing it partly, better than your opponent does, is what winning is.” She grinned. “The machine explored the tree so it could tell us a fact. You explore the tree so you can become someone who sees further. Those are very different prizes, and yours is the better one.”

The boy went home and, that night, before a decision that mattered, he sketched a little tree — his best two moves, the likely best replies, one more step — and chose, for the first time, like someone who’d already visited the futures.