We are five years from the events depicted in Blade Runner, but happily, already have a tool more accurate than the Voight-Kampff test for uncovering robots among us. It is called Twitter. And we've applied this tool to the candidates in some of the most hotly contested 2014 Senate races, so that we can root out the robot menace before it takes hold of the American body politic.
(If you haven't seen the movie mentioned above, the point is: Are our senators robots? The answer might surprise you!) (If you think senators might be robots.)
We used two Twitter tools to evaluate two important robot questions.
- Are the Senate candidates robots? To answer that, we turned to Bot or Not, a tool created by Truthy, a group at Indiana University - Bloomington. Bot or Not — which was created in part with money from the Department of Defense — gauges how likely it is that a Twitter account is operated by a robot or not.
- Is this Senate candidate supported by robots? For this, we used Twitter Audit which as far as we know isn't funded by the military at all. It scours the followers of Twitter accounts and returns an estimated percentage of how many are robots.
Two quick caveats before we jump into it. First, the definition of "robot" in the Twitter sense is a bit different than the definition of "robot" in the Metropolis/Isaac Asimov sense. Twitter bots are just dumb bits of script that do things like send you links to methods of losing weight/gaining sex. For more on them, read this. But we're going to pretend we're talking about actual robots, because, well, that is funnier. The other caveat is that these tools probably don't work that well anyway.
How will this work? We'll use the Mitch McConnell- Alison Grimes race in Kentucky to demonstrate.
Kentucky
Candidate | Robot likelihood | Followers | Robot follower percent | Robot followers | ||
---|---|---|---|---|---|---|
![]() |
Mitch McConnell (R) | @Team_Mitch | 61% | 7,927 | 16% | 1,268 |
![]() |
Alison Lundergan Grimes (D) | @AlisonForKY | 69% | 14,900 | 14% | 2,086 |
OK. So we can see that Bot or Not is telling us that there's a 61 percent chance McConnell is a robot, but a 69 percent chance Grimes is.
As for followers, about 16 percent of McConnell's followers are robots, compared to 14 percent of Grimes'. But we figured two things. First, that there's got to be a margin of error on the Twitter Audit tool and, second, that robots would algorithmically follow the candidate they preferred. So when the robot follower percentage for the two candidates was within three percent, we deferred to the raw robot follower count. Make sense? (No? Well, it's all made up, so don't worry about it.)
Anyway:
MOST LIKELY TO BE A ROBOT
Alison Lundergan Grimes
MOST LIKELY TO BE SUPPORTED BY ROBOTS
Alison Lundergan Grimes
And that's how it works. Off we go.
Alaska
Candidate | Robot likelihood | Followers | Robot follower percent | Robot followers | ||
---|---|---|---|---|---|---|
![]() |
Mark Begich (D) | @MarkBegich | 37% | 2,844 | 17% | 483 |
![]() |
Dan Sullivan (R) | @DanSullivan2014 | 51% | 564 | 18% | 102 |
MOST LIKELY TO BE A ROBOT
Dan Sullivan
MOST LIKELY TO BE SUPPORTED BY ROBOTS
Mark Begich
Arkansas
Candidate | Robot likelihood | Followers | Robot follower percent | Robot followers | ||
---|---|---|---|---|---|---|
![]() |
Mark Pryor (D) | @PryorForSenate | 74% | 1,233 | 13% | 160 |
![]() |
Tom Cotton (R) | @TomCottonAR | 60% | 6,532 | 21% | 1,372 |
MOST LIKELY TO BE A ROBOT
Mark Pryor
MOST LIKELY TO BE SUPPORTED BY ROBOTS
Tom Cotton
Colorado
Candidate | Robot likelihood | Followers | Robot follower percent | Robot followers | ||
---|---|---|---|---|---|---|
![]() |
Mark Udall (D) | @MarkUdall2014 | 56% | 938 | 17% | 159 |
![]() |
Cory Gardner (R) | @CoryGardnerCO | 48% | 5,800 | 18% | 1,044 |
MOST LIKELY TO BE A ROBOT
Mark Udall
MOST LIKELY TO BE SUPPORTED BY ROBOTS
Cory Gardner
Louisiana
Candidate | Robot likelihood | Followers | Robot follower percent | Robot followers | ||
---|---|---|---|---|---|---|
![]() |
Mary Landrieu (D) | @MaryLandrieu | 54% | 3,149 | 20% | 630 |
![]() |
Bill Cassidy (R) | @billcassidy | 56% | 7,011 | 22% | 1,542 |
MOST LIKELY TO BE A ROBOT
Bill Cassidy
MOST LIKELY TO BE SUPPORTED BY ROBOTS
Bill Cassidy
Michigan
Candidate | Robot likelihood | Followers | Robot follower percent | Robot followers | ||
---|---|---|---|---|---|---|
![]() |
Terri Land (R) | @TerriLLand | 70% | 6121 | 13% | 796 |
![]() |
Gary Peters (D) | @Peters4Michigan | 55% | 2587 | 14% | 362 |
MOST LIKELY TO BE A ROBOT
Terri Land
MOST LIKELY TO BE SUPPORTED BY ROBOTS
Terri Land
North Carolina
Candidate | Robot likelihood | Followers | Robot follower percent | Robot followers | ||
---|---|---|---|---|---|---|
![]() |
Kay Hagan (D) | @kayhagan | 48% | 6257 | 18% | 1126 |
![]() |
Thom Tillis (R) | @ThomTillis | 69% | 23600 | 7% | 1652 |
MOST LIKELY TO BE A ROBOT
Thom Tillis
MOST LIKELY TO BE SUPPORTED BY ROBOTS
Kay Hagan
This post was written by someone who is 81 percent likely to not be a robot.