Monthly Archives: September 2014

@GunfightGame, a Shoot-Out Game for Twitter, Now in Beta

by

desert_BG

I have been interested in Twitter Bots for a couple of months now. With Jay Weingarten, I designed two bots. The first is called @SMOOTHTALKBOT, and it recommends Santana’s Smooth to people who ask “What’s the best song?” The second is @RegisPhilbot. It is a Weird Twitter recreation of America’s most beloved TV gameshow host from one decade ago.

I should say that we didn’t call it Weird Twitter at the time. But some people have referred to it as that. I think there are similarities.

Both of these bots violate Tiny Subversion’s Basic Twitter Bot Etiquette, in that they contact people without solicitation. But so does @RedScareBot (who TS’s Darius Kazemi calls “such an asshole”). Honestly, I’m not so concerned about the Ethics of Twitter. Does @RegisPhilbot cause harm to others? If you count the annoyance of receiving a weird tweet, then yes. But that harm is so small, and so inconsequential, and it’s counter-balanced by the joy that the Philbot brings to the world.

This perhaps introduces the field of Twitter Bot Apologetics: I am defending my creations, rather than pursuing truth.

Other bots I’ve created that violate these ethical guidelines: @GrootBot, @NotARacistBot, and @NotAFeministBot. The latter two do this only by retweeting with permission. I find the “harm” done by an non-consenting retweet to be so small as to be immeasurable. But it’s still bothered me.

However, there’s a problem.

Twitter Bots don’t really get “discovered” in the same way that a blog post might be. If I post a really good essay, people will share it. But people don’t really “share” Twitter accounts. They just retweet them. And if an account tweets into a void, without ever contacting the outside world, it sort of languishes in this place.

For example, I created @yoimsola.

Every minute, this bot tweets what I lovingly describe as “some LA bullshit.” Basically, celebrity-sightings, not-so-humblebrags, and moments of existential despair. It’s also only following people with the last name Spielberg.

He never tweets at anybody, though (too self-absorbed), so the only people who ever notice him are people who search for celebrity names and then see that he’s claimed to run into them. (People often tweet at the account, asking where these celebrities were sighted. Nowhere on the account does it say it’s a bot, and apparently the 44.5k tweets do not deter them from the belief that it is a Real Human Being.)

So I began to be interested in the idea of a Memetic Bot, that is, a bot whose programmatic structure naturally facilitates its sharing by others. In short:

Could I create a bot whose inherent share-ability was a function of its design?

My first attempt was @TheComplibot.

Every hour, it tweets a compliment to one of its followers. Since it only has 30 followers, pretty much everyone gets a compliment pretty much every day. Follow it and reduce this problem, please.

Though I believe in the inherent structure of @TheCompliBot, it hasn’t really panned out. People retweet @TheCompliBot, but it doesn’t seem to inherently spread itself. (Perhaps that is because it prefixes every tweet with “Hey” and thus shows up too often on people’s timelines? Maybe I should change that. Anyway, I’m getting off-topic. But one last off-topic thing: I’m thinking about making @TheInsultBot. I don’t know why, but I get the feeling that people will enjoy being mocked more than they enjoy being complimented.)

In addition to this drive — the Creation of a Memetic Bot — I’ve been wanting to make a Twitter Game for some time now. And I just haven’t had any good ideas. Some kind of Twitter Bot scavenger hunt? (“Find a tweet that references Obama and cream cheese.”) Maybe a game where you are assigned a phrase to convince another Twitter user to say.

But then I had an idea…

I am a huge fan of Westerns. That’s why I made a series of YouTube shorts with Doug Patterson called “Old West.”

What if I could make a Twitter Bot that would facilitate two people on Twitter having a gunfight? Through its automated procedures, it could facilitate a kind of person-to-person game on Twitter, giving the communications platform the sort of interactive playfulness that face-to-face communication has.

There are a few games on Twitter already, but — as far as I can tell — they are mostly trivia and word puzzles. This makes sense. After all, Twitter is a textual communications platform, so it makes sense that most of the games are textual communications. But I don’t think that has to be the case: I think there is room for a whole bunch of new games on Twitter. And I hope this spurs other ones.

As such, I named it after the first game to ever have a microchip. I hope this connects it to the past, while also opening doors to future games. I also hope that I don’t get sued. (Can I? I don’t think you can own a title. Man, I hope this second-hand legal knowledge is worth relying on…)

With the creation of this game, I had four considerations to take in mind:

1. Not involving people who don’t want to be involved
2. Not allowing other people to misuse the bot in a way that violates 1.
3. Making it fun and entertaining
4. Making it non-violent, or at least non-violent enough that it doesn’t offend bystanders.

Here was how I tackled those four considerations:

1. Not involving people who don’t want to be involved.

I required a duel to be initiated between two people. The first would challenge another by saying “@GunfightGame I challenge @ATwitterUser to a duel.” Then, a tweet would be sent to @ATwitterUser, telling them of the problem and asking if they agreed. If they did, the game would begin.

2. Not allowing other people to misuse the bot in a way that violates 1.

Of course, this being the Internet, I was worried people would take advantage of the challenge mechanism. Even though the bot requires the consent of both parties to proceed, that first message could still be considered spam, and I didn’t want to enable the entire Twitterverse to use my bot as a sort of spam.

So I decided to limit who you could challenge to people who were already following you. This puts the impetus on you. If people don’t like you challenging them, they can stop following you. If they do, you can’t challenge them anymore. I figured that this solved that problem (though I guess we’ll see…).

3. Making it fun and entertaining

Once the second party (the “challengee”) agrees to the fight, another tweet is tweeted that tells both players that a duel has begun, and that the first person who replies “bang” will be victorious. I thought about requiring multiple shots, but I didn’t for three reasons:

i. I didn’t want to have to code it. The bot uses Google App Scripts, and stores no information in a database. Keeping track of multiple shots would require a database (or at least some cleverly structured replies). I didn’t want to bother.

ii. I thought it was more fun for it to just be quick response. The concept of the gunslinger is all about you being quick or dead. And since there’s no way to have users affect the accuracy of their shot, I thought this was the best way to create a sense of meaningful play for both participants. Everyone knows how the game works.

iii. I thought people would be more likely to play the game the less effort it required from them.

With the basic mechanic understood, I then went through and punched up the tweet text itself. I wanted to have a sense of those great Elmore Leonard novels. Quick prose. Declarative sentences (or fragments). I wanted to capture that feeling from the beginning of Once Upon a Time in the West, the heat, the boredom, the tension, and the vast emptiness.

giphy

4. Making it non-violent, or at least non-violent enough that it doesn’t offend bystanders.

Normally, this isn’t a concern of game-makers. In fact, quite often the opposite concern is the case: people want to make their game as violent as possible. I’m not a huge fan of violent games, though. And I wanted to be especially cautious with this one, because it had the possibility of involving people who had not consented to the violence.

With a personal tweet, I can judge whether I think it’s appropriate for the person, the situation, and the time. But the bot can’t have that level of oversight. It needs to be programmatically designed not to be too violent in any circumstance. To accomplish this, I did a couple of things:

i. I made the introductory tweet totally non-specific. Though the name of the account is @GunfightGame, the message simply says that you’ve been challenged to a duel.

ii. Once you play the game, the language is archaic, metaphoric, evocative of violence, but not violent itself. When someone is killed, they are “gunned down.”

iii. The graphics are bright and cartoonish.

@GunfightGame is now in Beta.

If you would like to play the game, you can tweet “@GunfightGame I challenge @SoAndSo to a duel.” (You don’t have to say “to a duel” or “I,” but you do need to say “challenge” and the name of a follower. Otherwise, it rejects it.)

There are still a couple of bugs in the game, and I’m sure there will always be situations in which it malfunctions. Opening yourself to a wide Twitter audience is always going to cause problems. With a browser game, you can control the ways in which an ordinary user interacts with it, what kind of input they provide.

But with @GunfightGame, there’s no controlling that. Anybody on Twitter can tweet anything at it at any time. If it starts to get too many tweets, it will definitely begin to malfunction. People say that you should prepare for success, but there’s really no way to do that. If and when it starts to malfunction, I’ll try to fix it.

For now, I invite you to challenge your followers, and please tweet any bugs or suggests to @GregKarber.

A website is coming soon.