When I got my first text from Bernie, I didn’t recognize the number, but I opened it anyway.
After all, I’m used to opening my texts. This one was from Amy, who said she was a volunteer for the Bernie Sanders campaign. She wanted to know whether I’d be supporting her candidate during the primaries, and included a link in case I wanted to donate.
That text message was generated using one of a new breed of peer-to-peer texting apps that exploit a gap in communications law to hit up your phone without your permission. They allow campaign volunteers to send a large numbers of texts to voters one at a time. The apps were a novelty in 2016, a staple in 2018, and will be indispensable for campaigns at all levels in 2020, with over a billion such texts expected to be sent by both Republican and Democratic candidates. And they’re being used for a wider variety of purposes now—not just getting out the vote and recruiting volunteers.
But are they effective? And do they risk turning off voters already inundated with text message advertisements? I soon deleted the text from Amy, and the two more that followed—Bernie just didn’t happen to be my candidate at the time. But if I’d received a similar text from the campaign I’m supporting, there’s a fair chance I’d have donated and maybe even volunteered. The real value of P2P texting apps may lie in their ability to reach large numbers of voters—not to convert them into supporters, but to help mobilize the true believers—with surveys showing that they helped motivate voters to get to the polls in past elections.
The main reason texts can reach so many voters and motivate supporters is because while most people screen phone calls, especially from numbers they don’t know, they associate texts with people close to them like friends and family. Right now the texts, generated by apps like Hustle and GetThru on the Democratic side, and RumbleUp and Opn Sesame on the Republican side, get read between 70% and 98% of the time. In 40% to 50% of cases, people actually respond. By contrast, emails only get read 20% of the time and responded to 6% of the time, according to Gartner.
And P2P texts seem to have a greater effect on certain voter demographics, especially when it comes to turning out the vote. Text messaging aligns well with the communications habits of younger demos, in the same way that Facebook ads seem to connect best with older audiences. Could that benefit benefit some candidates more than others? Could it boost Bernie Sanders, whose hardcore base skews young?
Daniel Souweine, the Sanders campaign’s text outreach manager in 2016, says it’s important not to overplay the demographic effect of P2P texts. While younger voters may be more apt to get into longer exchanges via text, pretty much anybody under 50 will be comfortable enough with texting to respond to political texts coming in on their phones. In fact, research shows that texting may have a bigger effect, politically, on slightly older voters. According to a study of texting in the 2018 midterms by the progressive digital advocacy group Tech for Campaigns, registered voters ages 26 and younger who were texted by a campaign turned out to vote 4% more often than voters who weren’t. But voters 27 to 50 years old turned out at a rate almost 8% higher than those in that same age group that were not texted, the study showed.
And the fact that virtually every campaign was using P2P texting in 2018 and will be using it even more in 2020 may have normalized it somewhat. “I think by now that effect has pretty well baked in,” says digital consultant Colin Delany. “The main age-related effect would be that campaigns have a lot of ways to reach older voters, but texts are one of the few things that younger voters will actually see.”
From millions to billions
It’s hard to get reliable estimates of how many texts have already been sent by all the candidates active in this election cycle, because campaigns often don’t like to disclose real numbers or they skew them for better optics. One campaign source told me that “tens of millions” have already been sent in the presidential primary race, but that number is probably too low.
The Trump campaign has been working on its 2020 texting game since 2017, and plans to send “almost a billion texts,” according to campaign manager Brad Parscale. The Republican Party and the Trump campaign use a P2P texting platform called Opn Sesame which was developed by Gary Coby, who was an architect of the campaign’s winning digital strategy in 2016 and now leads the texting program for Trump 2020. The Democratic National Committee says it’s already sent roughly 10 million P2P text messages this cycle, using the texts to locate and recruit volunteers all over the country. That goal became especially important in the wake of 2016, when the party didn’t organize early in key states where its eventual nominee might be weak, says the DNC’s mobilization manager Patrick Stevenson.
While campaigns are pushing out a lot more texts this cycle, they’ve also become much more ambitious about their goals. In 2016, Bernie Sanders used P2P texting to get people out to vote, and to recruit and organize volunteers. Today, the campaigns, PACs, and political parties use P2P texting for those purposes, plus fundraising, donor relations, promoting events like rallies, and inviting people to take surveys.
And P2P texting has become much more integrated with other communications channels like direct mail and online promotions, increasing its reach and effectiveness. For instance, I recently responded to a Trump Facebook ad, then filled out a short survey at a Trump site. The next day, I got a text from the campaign asking for a donation. Even though I’m neither a Republican nor a Trump supporter, the campaign had my cell number and was able to match my online identity and/or my email address to it.
A campaign typically starts with a list of cellphone numbers taken from a voter file it leases from the party. The voter file contains voter-provided phone numbers, email addresses, and demographic information on everybody with a recent voting history. My cell number wasn’t likely in the GOP voter file, but the Trump campaign gets millions of additional phone numbers by purchasing them from commercial data vendors like Axciom or i360. Whatever their source, the cell numbers can be queued up and placed into text messages in the P2P texting app, along with some boilerplate text that the sender can customize with their own words. Then they just hit Send and move on to the next phone number.
The DNC’s Stevenson told me the party has begun sending follow-up texts (using the Hustle app) to people to whom they’ve sent direct mail. This requires a voter file match between the person’s physical mailing address and their cellphone number. Stevenson said that when they do these follow-up texts, recipients are 21% more likely to make a donation.
More effective than email
Democrats were the first to discover the goodness of P2P texting. Back in the fall of 2014, an ex-Facebook employee named Roddy Lindsay was volunteering at the immigration nonprofit FWD.us, when a few of the organizers began using text messaging to contact supporters and invite them to events. “It turned out to be a really effective way to do that,” Lindsay told Recode in 2016. “We got people who never opened their emails to respond to a text.'”
After seeing that, Lindsay and two cofounders hurried to develop a texting app specially designed for outreach and organizing. Hustle, the company, launched in late 2014, and political campaigns began using the Hustle app in mid-2015. The Sanders campaign was one of them, using the app for get-out-the-vote efforts, but not on a grand scale—it was considered just one of many communications channels. However, the campaign’s text outreach manager at the time, Daniel Souweine, says they found that Hustle was an especially effective tool for signing up and organizing volunteers. Souweine was such a believer that he left the Sanders camp after the 2016 primary to found his own P2P company and platform called Relay (now GetThru).
By the 2018 midterms, P2P use had increased dramatically, with virtually all House and Senate campaigns using the technology to some extent. In the 2018 cycle alone, Democratic campaigns and politically adjacent organizations sent 350 million P2P text messages, six times the number sent in 2016 and 2017.
Love Bernie Sanders. Hate getting campaign texts from anyone, including Bernie Sanders. pic.twitter.com/95H4FTBVVM
— Matt Banks (@TheOCBossman) March 1, 2020
‘Is this a bot?’
The numbers will be far bigger in 2020. As November approaches, some voters–especially in battleground states–might start feeling some fatigue at getting so many text messages from so many campaigns, parties, and PACs.
“If you’re a voter in a battleground district, you might be besieged by text messages,” said digital consultant and epolitics.com editor Colin Delany. “Especially when you’re getting texted by campaigns that are not coordinated, or [legally] cannot coordinate.”
“The thing with P2P apps is that a lot of organizations are using them in a lot of different ways,” Delany said. “It behooves the state parties to have the different campaigns in the state coordinate, at least to extent that they all know who’s texting who,” Delany said.
But on the Democratic side at least, the technology infrastructure needed for such coordination is very new. The DNC, for example, is backing an independent Democratic Data Exchange that’s still being built. It was used to help some Democratic campaigns coordinate their activities and communications (door knocks, phone calls, etc.) in Virginia and Kentucky during the 2018 midterms, but that was a proof of concept. The DDeX will likely perform the same sort of tasks on a nationwide basis in the 2020 election. Other private efforts, like the Reid Hoffman-backed data exchange Alloy, are hoping to provide wider sharing of all kinds of data and tools to progressive organizations, but it too is still in its formative stages.
DNC mobilization manager Patrick Stevenson
“We don’t want to be out there salting the earth.”
The DNC’s Stevenson said he’s concerned about preventing text fatigue in the same way he’s trying to respect voters’ space with all of its outreach tools. “Generally, people really react strongly to being incorrectly targeted, so we’re careful to quickly update the voter file and take those people off the list,” he told me. “We don’t want to be out there salting the earth.”
Since P2P texting is relatively new to voters, people often aren’t sure if they’re being texted by a real person, Stevenson said. “Sometimes people write back and say, ‘Is this a bot?’ and the answer is ‘No, this is a 25-year-old staffer volunteering for free.'”
GetThru’s Souweine says he often advises clients to send fewer texts. “I might tell them that they shouldn’t bother trying to reach a certain demographic with a certain message, because it’s just won’t work,” Souweine said.
Enabled by a loophole in federal law
If voters begin feeling overwhelmed by P2P texts during the run-up to November, their text inboxes might start to seem less personal and relevant, and more spammy. P2P texts are unlike spam in that they have to be sent out one at a time, one sender to one receiver. But they are like spam in that a sender can send a promotional message via an electronic channel to a person they don’t know without consent.
In a way, the whole P2P texting boom is due to a gap in the communications law, one that consumer groups are trying to address in order to reduce the flood of texts bombarding Americans. The relevant statute, the Telephone Consumer Protection Act (TCPA), was created to rid the world of annoying robocalls. But it was written in 1991, well before cellphones took over the world. The law says phone calls made using autodialers are definitely not OK, but had no way of contemplating exactly what an “autodialer” might look like within a texting context.
The P2P app makers were so nervous about this gray area that they formed a trade group called the P2P Alliance, which then hired a powerful lobbying firm to intervene on its behalf with the FCC. In its May 2018 petition to the commission, the group’s lobbyists asked the FCC to officially clarify that P2P texts don’t fall under the prohibitions outlined by TCPA. A month later, the Republican National Committee filed a similar brief in support of the P2P Alliance’s request.
A number of consumer groups, including the National Consumer Law Center, Consumer Reports, Public Knowledge, and others, filed their own petition with the FCC last July, urging the agency to resist granting the texting lobby’s request. “The structure of the TCPA shows that Congress intended to require consent not just to automated calls, but to all calls made on systems that meet the definition of an automated telephone dialing system,” the letter states. It also argued that, based on the definition provided in the TCPA, P2P texting apps do indeed use a form of automated dialing, and so should be prohibited by the statute.
Digital consultant Colin Delany
“But some people are telling clients not to count on it for 2022.”
The FCC has asked for public comments on the matter, but so far has not made the clarification the P2P Alliance asked for. In fact, it’s taken no action on the matter, and has given no indication in public that it intends to do so. It did, however, publish a 2016 advisory to political campaigns saying that only texts sent using an autodialer would be prohibited by the TCPA.
In the absence of an official clarification, P2P texting lives on shaky regulatory ground, and campaign consultants are exploiting the loophole, fully aware that it might not last forever.
“When you talk to political consultants, the general consensus is that this is one regulatory decision away from disappearing,” Delany told me. The attitude among consultants is “it’s available now, so we’ll use it,” he adds. “But some people are telling clients not to count on it for 2022.”
In the current political climate, in which misinformation has become a major concern, some experts worry that P2P texting apps, because of their unregulated nature, could be used by bad actors to sow discord and even suppress the vote (via texts informing voters that their polling place or election day has been changed). So far, those fears have remained mostly theoretical.
Mostly. Just before early voting began in the 2018 gubernatorial election in Tennessee, an unidentified party (using a Tennessee phone number) sent anonymous text messages bashing the Republican candidate Bill Lee. “I’d never vote for Bill Lee. He gave $ to liberals Megan Barry and Phil Bredesen, who’s running against Marsha (Blackburn)!,” one of the texts read.
But the texts weren’t exactly misinformation, because they referred to real news stories from legitimate sources. It was their timing and anonymity that raised flags. There is no federal law requiring texters to identify themselves or provide disclaimers in political texts, and Tennessee authorities at the time weren’t even sure if the texts violated any laws. The Federal Election Commission declined to require disclaimers on political texts back in 2002, saying that there are too few characters in texts to reasonably require them. And the FEC hasn’t seriously revisited the subject since, even though texts can now handle lengthy word counts. It’s more likely that a state law would prohibit anonymous political texts.
“The instances of impersonation and other kinds of malfeasance are vanishingly rare,” GetThru’s Souweine said. “I can think of one time that it happened with one of our clients in three years, and it was to the tune of a few hundred text messages.” Souweine said the problem was quickly identified, and the offending texter was removed from the platform. GetThru encourages its users to identify their campaign, and make it clear that it’s a real person texting. “And there are starting to be rules that specifically encourage this kind of clarity, such as the recently passed updates to the Disclose Act in California,” Souweine said.
One reason that P2P texting apps might not be the optimal platform for spreading misinformation is simply because the texts are a one-to-one, not a one-to-many, form of communication. That makes it harder to make a text go viral, in the way that such information spreads on Facebook and Instagram. P2P platforms exert control over the content flowing through their platforms, but it’s certainly possible that some anonymous bad actor could send out a lot of texts before the company became aware of it. It’s also possible that a well-funded misinformation campaign could build its own P2P texting platform.
In any case, some of the most harmful campaign communications in 2020 could come from the campaigns themselves. The Trump campaign used advanced microtargeting to convince certain groups to stay home on Election Day in 2016. Just before the day, it targeted black voters in swing states with messages saying “Hillary Thinks African-Americans Are Super Predators,” as McKay Coppins mentions in his recent Atlantic feature on misinformation in 2020. An anonymous Trump campaign official bragged to Bloomberg Businessweek that the tactic was one of “three major voter suppression operations underway.” The other two targeted young women and white liberals.
It’s almost certain that this type of Big Data-powered voter suppression will happen in 2020. And it could just come via text message. That’s because the same dense and multilayered data profiles used to target individual voters on Facebook can be used to target voters’ cellphones with texts. It’s two different delivery modes, but the same targeting data. As in the Tennessee case, some of the voter suppression ads or texts can’t exactly be called misinformation, because while they contain half-truths and exaggerations and often twist the meanings of news stories, they can’t be clearly debunked as lies.
In short, there’s cause for concern about misinformation in texts. And if P2P texts start to be used to spread such malicious content, consumers might stop opening them and regulators might take a new interest in P2P texting platforms. Until that happens, we should all get used to political texts, and learn to assess their authenticity and accuracy.
CORRECTION: An earlier version of this article erroneously stated that Hustle launched in late 2015 and campaigns began using its app in 2016. Actually, the company launched in late 2014, and campaigns began using the app in mid-2105.
This story is part of our Hacking Democracy series, which examines the ways in which technology is eroding our elections and democratic institutions—and what’s been done to fix them. Read more here.