back
45 comments
Hey HN, I'm the creator of this multiplayer tic-tac-toe game called tic-tac-tic-tac-toe. It's a Firebase-powered variant of the game everyone knows that pits GitHub against Twitter. Let me know what you think!
I would join, but it wants my email which I'm not sure why you need it since it gives you my other github information to insure I'm not a bot.
Sorry to hear that. I'm not actually using the emails in any way and don't have plans to. The reason it asks for your email is that I'm using Firebase Simple Login which requests the email scope by default.
I'm so impressed with the UI and execution. Very well done. No complaints, except that it's perhaps the move timer is a little fast.

I first learned about Ultimate Tic Tac Toe here[1]. Might be worth a link to that post -- it's a great explanation with surprisingly satisfying illustrations.

[1]: http://mathwithbaddrawings.com/2013/06/16/ultimate-tic-tac-t...

I've been playing a really similar game called Tic Tactics, the only difference seems to be you can play in a won square until it's completely full then it becomes a free move. Not sure what overall difference that makes to tactics but it's interesting they made that small change. A big difference is taking a board doesn't give your opponent a free move, which improves late game a lot.

Actually it also starts the game with 9 moves each which makes things interesting, any collisions match and you can't take three in a row or the middle tile.

Thanks for the comment! I considered giving each team 10 seconds per turn but settled on 5 seconds since I didn't want the games to last too long. I wanted to make it such that people kept saying "oh I'll just play one more game." And - for what it's worth - I think that's the same place I first heard about the game.
I still feel 5 seconds is too short.

I actually decided to stop playing the game rather quickly, because the team is so desperate to move in 5 seconds that they make several terrible moves in a row, easily losing the game in a almost minimal amount of turns possible to lose.

The 9 game grid makes for an interesting game with two people, but the crowd tends to rush for the wrong move. I wish there was a way to explain that you don't need to dive on a move to block the other team as long as you don't let them play in that sector. Unfortunately cooler heads are usually outnumbered.
This and Twitch Plays Pokemon have taught me a lot more about group efficiency than The Mythical Man Month
the worse is people piling onto a middle square when it allows the other other team to go anywhere if the middle is already won
What's the optimal strategy if you're playing solo vs solo (no team)? Regular tic-tac-toe is an obvious P1 draw due to the strategy-stealing argument, but that doesn't apply here because your move restricts your opponent's move choice.

This game can't end in a draw, bc majority gets the win.

I recall lots of people found winning strategies when the game first came out. They were based on always sending the opponent to the same field.
I meant optimal perfect strategy, either via a proof or a game tree search (alpha-beta pruning). With 3^81 game states (most not attainable), it should be searchable.

More precisely, the game has a finite # of states, and cannot end in a draw. It's always a win for P1 or P2. If both people play perfectly, who wins and what is the optimal strategy? (Note: It is possible that this game is a P2 win, because your move restricts your opponent's choice.) (For the person with the guaranteed win, what strategy forces the win? For the person who can't guaranteed win, what strategy gives your opponent the most opportunity to make a mistake?)

I'm not so sure that it's searchable. 3^81 is quite a lot. I don't know of any solved game with that many. Tic-tac-toe, connect 4, checkers, etc all have less. I think 3^81 is comparable to 9x9-Go and Chess which are both unsolved. I suppose it depends how aggressive you can prune (which, admittedly, is probably pretty aggressively).
To (perfectly) solve 9x9 go, you need to consider almost every move at every turn. That's a branching factor of 40+.

In this game, there are (usually) at most 9 legal moves per turn. The branching factor is a lot less. (Giving your opponent free choice to play anywhere is frequently a mistake.)

It's no where near 3^81 as there are winning strategy's that don't fill the board. Optimal play is really about delaying how long it takes to force a win.
Team Github seems to be the far more popular choice. Though strangely, it seems like the smaller team Twitter does better. Maybe crowds aren't as smart in this case?
Playing the game, it seems like Team Github doesn't actually know the rules of the game, and are just playing optimally as if for normal tic-tac-toe. They continually force Team Twitter into the middle area (by continually playing in the middle square of each area), allowing the most dominant meta area to be given up. Looking at the numbers of people that are choosing each square, it doesn't seem like it has anything to do with the number of people on each team. Literally every single person on Team Github chose the same squares in the most crucial mistakes.
I was the only person picking for Team Twitter (team size: 5) for a while, so I was able to pick strategically while Github was often split among 2-3 choices (team size: 20).
Team Twitter was going really well around the 8 person mark vs the 20 or so on GitHub.

This was a blast. Great presentation. Who had the idea to pit GitHub vs Twitter?

It was ultimately my idea to pit these two teams together in this format but I initially got the humans vs. humans suggestion from a coworker. I initially had it as a single human team versus a computer AI but I think this makes for a more exciting matchup!
I'm currently addicted to TicTactics and know a few decent strategies, so the experience of wanting to use a non-obvious move required for planning-ahead but having all the other players do the obvious move, is very frustrating.

BUT, that's just me. Nice app anyway!

Nice, looks like this is also using the Firebase Hosting beta.

First time I've seen that in the wild but I've been using it in lightning talks for a couple of months now. It's pretty slick.

This is great! The mob seems to make some funky decisions. I'm curious to see what happens when more people join; if the mob gets smarter or if they get worse. Tell you friends and let's find out!
I think the rules are implemented incorrectly. Multiple times I've seen a move get made that seems to violate the rules.

As an example, I just saw team Twitter place a piece in the center square of the top left board (the center board had not yet been won). The Github team then proceeded to place their piece in the bottom right board rather than the center board. It may have been the case that there was a split vote for center and bottom right in Twitter's move that resulted in this bug manifesting itself.

I'm glad I'm not the only one who's seeing glitches. I've had instances of the right squares being lit up in green, but then a square on another board, which shouldn't be permissible, actually get selected.
I am seeing some of these issues as well. Also seeing the clock do some weird things. I am going to reset the host in a game or two and hopefully things will get back in sync. The game will probably be way out of sync for a few seconds while I do the reset.
Things should be better now!
Very cool. How do you make this with Firebase? Do you have a central server polling the Firebase collection or how does it make the decisions if all logic is on the clients?
Each client writes to a "suggestions" node in the Firebase with their suggestion. Each client also listens for changes to that "suggestions" node to show you how many people have voted on a particular square. Firebase internally does all the work to send the changes on the "suggestions" node to every client. I also have a single host computer (not a server, just my laptop next to me), which goes through all the suggestions once the timer runs out and makes a move for the current team. This move is written to the Firebase and every client is notified by Firebase that the move has been made. They each update their own UI and the process starts over.
Very interesting. I half to imagine most of the crowd doesn't understand the rules, or is intentionally trying to throw the game. Many of the moves that I saw happening were exactly the wrong move to make.

Here are rules and strategy as discussed here prior: https://news.ycombinator.com/item?id=5898506

Are there loads of trolls or do I not see in what universe it is a good strategy to send the enemy to a block we almost own?
I think that's due to the timer causing people to make mistakes.
It's pretty fun, one complaint is that it sometimes drops moves or uses ghost votes to put us somewhere we didn't vote for... still cool for a weekender!
i sense an off-by-one error? i was the only one playing for team Github:

Team GitHub played [8,1,0] @kn0ll chose [8,2,0]

Any thoughts about adding chat? =)
Yeah, because most ppl don't read the rules and don't understand how the game works, with chat we could try n' educate 'em...
Collaborative gaming has a lot of promise, but so far it's been implemented poorly.

You need to allow gamers to create small teams, and allow them to talk with one another to discuss strategy. Then they decide on a move.

Someone make chess using the above.

To be fair, these implementations are only poor in your opinion. And your suggestion might be poor in other ways.

For example, I would suspect far fewer people are willing to invest upfront in a game like yours. It's much easier to find two people who want to play chess against each other where they are fully in control, than two groups getting only one single, slow game out of it.

One way a more thoughtful, strategic game like yours might work would be to achieve critical mass with a simpler, fast paced game and then add depth.

That would make a way better and smarter game. But for 5 minutes of fun, this unstructured style is probably the way to go. I don't think this Show HN is intended to become the next Starcraft.
Reminds me of this: https://en.wikipedia.org/wiki/Kasparov_versus_the_World Not exactly what you describe, though, but a fun experiment.
Very simple UI with a small learning curve. Bravo, this is fun.
Team Twitter > Team Github! :) tribal chest pound
Twitch plays Tic-tac-toe!
for some reason it causes 100% CPU load after some time (Chrome 34)