This video was made possible by Audible. Try out Audible plus for free for thirty days
by going to audible.com/HAI. So, you want to totally ruin a fun little
game from your childhood? You’ve come to the right place. Here’s how to never lose tic tac toe. Step one: go first. It’s usually best to say something like,
“do you want me to go first, I don’t really care,” and then make a move before they
can respond. And when you do go first, go in one of the
corners. Now, if your opponent goes in the middle,
you might end in a draw. But, if they go anywhere else, you can guarantee
a win. There are basically four non-middle places
your opponent can play—technically eight, but really four because the positions are
duplicates if you rotate the board. They can go next to you, in a corner near
you, in the middle away from you, or in the opposite corner. If they go next to you, go in your nearby
corner, but not with them in between you. They’ll block you, then you go in this corner,
and now you’ve won—because either they’ll block you here, and you go here, or they’ll
block you here, and you go here. Now, if after your first move they hit you
with the corner near you, just take the opposite corner. Then same thing: they block you, you go in
the other corner, and boom, pop the sparkling apple juice, you’ve won again. If they start in the middle away from you,
go in the corner touching them. Then same thing, they block, you corner, you
win. And if they go in the far corner, just go
in any other corner, and again, block, corner, win. Now, if they were smart, and they did go in
the middle at first, don’t panic. Play it cool, and go catty-corner to your
first move. If they then go in a corner, you go in the
opposite corner and you’ve won—they’ll block here or here, and you go in the other
one. If they go in a middle, though, you have to
block them and sadly, you’re probably headed towards a tie. So, how do I know all this? Well, because I read the tic-tac-toe Wikipedia
page. Well, actually I didn’t read the Wikipedia
page, I paid a writer to read it for me, which I was able to do because years earlier, I
read the Wikipedia page on capitalism. But how does Wikipedia know this? Well, it’s because tic-tac-toe is what’s
called a solved game. Like ice cream flavors at a store that has
very few ice cream flavors, solved games come in three types. The first type of solve is called “ultra-weak,”
which incidentally is also the type of jokes we use in these videos. Basically, it means that the value of the
start position is known—that value can be one of a few things: either the first player
can always win, the second player can always win, or one or both players can always force
a tie. The thing about an ultra-weak solve, though,
is that while you know the starting position’s theoretical value, you don’t actually know
what the winning strategy is. For example, there’s a mathematical proof
by Nobel Prize Winner and subject of the 2000s’ worst Best Picture Winner, John Nash, showing
that a game of hex on any NxN board can always be won by the first player. While we know the winning strategies of small
boards—from 1x1 up to 9x9—once we get to bigger boards, we know the math says there
definitely exists a way for the first player to guarantee a win, but we don’t actually
know what that way is. The second type of solve is a weak solve. That means that the value of the start position
is known, and an algorithm knows how to get to that outcome, but only starting from the
beginning. If you give a weak-solve algorithm a board
already in play, with mistakes made, it’ll go all blue screen of death and won't know
what to do. The third type is a strong solve—that means
that you, or an algorithm, can figure out the best possible move from every possible
game position. Mathematicians who apparently have nothing
better to do have developed strong solves for a ton of games, from tic-tac-toe, to Connect
Four, to Chopsticks, often using something called a minimax algorithm. To understand, let’s rip-off Black Mirror
and become an algorithm, by doing a minimax-style analysis of a game played between our two
beloved mascots that we’ve always had: giraffe as interesting, and calf as interesting. The only way to have a simple example is to
invent a really boring game that would suck, so, in this game, there are only two move
options, left or right, and there are only three turns: giraffe, then calf, then giraffe. Here’s what’s called a game tree, which
shows all the possible outcomes: giraffe wins are in yellow, calf wins in purple. So let’s work backwards. We see here that in this outcome, it’s giraffe’s
turn, and if they move left they win, if they move right they lose. They’ll always move left to win, so this
position is a guaranteed giraffe win. And in this position, they win either way. That means on this half of the tree, no matter
what calf does in its move, giraffe will win. But on the other side, in this position, calf
always wins, no matter giraffe’s second turn. In this one, giraffe can win—but, on calf’s
turn, they can always just go left, meaning this side guarantees a calf win. But if we go to the top, we see that all giraffe
has to do to guarantee a win is go left at the beginning, meaning that this game has
been strongly solved, and that the first player will always win. Theoretically, that same minimax algorithm
analysis could solve any two-person game with finite board positions; but before you start
building a Magnus Carlbot, you should know that games like chess or Go—both of which
have more possible outcomes than there are atoms in the observable universe, which is
probably at least a thousand—are too complicated for a minimax to be practically feasible. For the time being, you’ll just have to
settle for ruining tic-tac-toe for your friends and family. Remember: it’s technically not cheating,
it’s just being a tool. You know what else feels like cheating but
isn’t: listening to audiobooks. You get all the incredible storytelling of
reading a book, plus the superior, smug feeling of reading a book, without actually having
to make your face holes read a single word squiggle. And the best place to listen to audiobooks
is on Audible, the leading provider of spoken-word entertainment. Their newest plan, Audible Plus, gives you
full, unlimited access to their Plus catalogue, letting you listen to thousands of audiobooks,
plus original audio dramas and podcasts. It’s the perfect option for audiobook and
podcast power-listeners like me. I’ve been using Audible Plus since it came
out, and the absolute favorite thing I’ve listened to with it was Forward—a collection
of short stories by some of the top science-fiction writers out there, including Andy Weir of
the Martian. The best thing about Audible Plus is that
it’s unlimited—you can try as many titles as you want, and it still costs the same,
so you’re guaranteed to find something you like—and you can listen to Forward or anything
else in their catalogue for free for 30 days by going to audible.com/HAI.
