No more robocalls: New tech automatically hangs up on robots

Every month, the Federal Trade Commission receives 200,000 complaints about illegal robocalls, making it the most common problem reported to the FTC. Although the FTC has shut down some of the companies responsible for the billions of robocalls made in the US, it lacks a technological solution that would stop robocallers from dialing in the first place.

That's why the FTC last October offered a $50,000 prize to whoever could come up with the best technology for stopping robocalls. Today, the FTC announced three winners who devised robocall blocking systems that might help end the scourge forever. Two of the winning proposals detect illegal robocallers and disconnect the call before it can get through to consumers. The third would build a database of bad numbers by having consumers report robocalls.

The $50,000 prize will be split by two co-winners who separately built systems that "focus on intercepting and filtering out illegal prerecorded calls using technology to 'blacklist' robocaller phone numbers and 'whitelist' numbers associated with acceptable incoming calls," the FTC said. "Both proposals also would filter out unapproved robocallers using a CAPTCHA-style test to prevent illegal calls from ringing through to a user."

The FTC additionally honored two Google employees named Daniel Klein and Dean Jackson for devising a crowd-sourced robocall identification and blocking system. They won't receive a financial prize because none was offered for proposals from large businesses, but the system could still end up helping consumers.

The winners who split the $50,000 prize were Aaron Foss, a freelance software developer from Long Island, NY, and a computer engineer named Serdar Danis. Foss attended the winners' announcement press conference and talked to Ars on the phone afterwards.

Nomorobo—say it out loud

Foss' system is called "Nomorobo," and it blocks robocalls on any type of phone—landline, VoIP, or cellular—without requiring the user to buy any extra hardware. The customer would set up conditional call forwarding through their phone provider, allowing calls to ring both to their own number and to a Nomorobo number.

Nomorobo answers the calls first, passing legitimate calls (such as those from actual humans) through and hanging up on illegal robocalls. Rest assured this is more complicated than it sounds, and there is still work to be done before it can be deployed at scale.

One problem is that not all robocalls are illegal. The well-known "Rachel at cardholder services" and other such scams clearly are illegal, but robocalls are legal for groups such as charities, government and political organizations, and pharmacies notifying customers that their prescriptions are ready. Thus, a comprehensive whitelist to pass through legal calls and blacklists to block illegal ones must be developed.

The blacklist would be populated with the worst offenders in an FTC database, and Nomorobo uses algorithms to identify other illegal robocallers. If the system isn't sure, it presents the caller with an audio CAPTCHA to verify whether it's legit, as you can see in a Nomorobo demo video:

"You might be a robocaller," the system asks the illegal robocaller. "To prove you are a human please enter the number 71. … I'm sorry that is incorrect. Your number has been blacklisted."

Foss said he developed an algorithm to detect whether calls are illegitimate, inspecting the caller ID header and frequency of calls. He said it can detect illegal calls even when robocallers use spoofed caller IDs, but said he will not reveal exactly how it does that because it's proprietary technology he intends to sell. (Presumably, the FTC got a more detailed rundown of the technology during the awards process.)

Creating a comprehensive whitelist is also a major challenge, as it requires anyone making legitimate robocallers to identify themselves. Foss believes legitimate users of robocaller services will be eager to participate, however, because the proliferation of illegal robocalls makes it harder for consumers to trust the legal ones.

"The not-bad guys in the robocall market are just as much against illegal robocalls as the FTC is," he said. "The major companies want to make sure their phone calls get through."

Nomorobo itself runs in the cloud. Foss said he uses Twilio for phone connectivity and would likely run the server on the Amazon Elastic Compute Cloud. "It would be very lightweight. The system can could do analysis rapidly without a lot of computing power," he said.

If the Nomorobo system suffered an outage, the worst that would happen is that customers would receive all the calls directed at them (both legitimate and illegitimate).

Foss set up a website for Nomorobo and plans to use the $25,000 as seed money to help bring the system to market. While he wants to commercialize it himself, he is also open to partnerships and licensing the technology to other companies.

The company operating the service would be able to manage call numbers through an interface like this one:

Credit: Aaron Foss

Foss hopes to get Nomorobo on the market within the next six months. He envisions the service being sold to consumers for a "small monthly fee," billed either through the phone company or directly through Nomorobo.

"Does it work? Is it easy to use?"

The FTC received nearly 800 proposals, judging them on the criteria of "Does it work?," "Is it easy to use?," and "Can it be rolled out?" The judges were FTC Chief Technologist Steve Bellovin, Federal Communications Commission Chief Technologist Henning Schulzrinne, and AllThingsD editor Kara Swisher.

One interesting proposal that did not win the contest involved a Raspberry Pi, additional hardware, and software to implement a blacklist/whitelist system. We wrote about it in February, noting at the time that it could fail the easy-to-use test by requiring consumers to buy hardware. It also didn't work across all types of cell phones.

The winners' systems are supposed to work across any type of phone system and shouldn't be difficult for consumers to set up. Besides Foss, Robocall Challenge co-winner Danis developed a system called "Robocall Filtering System and Device with Autonomous Blacklisting, Whitelisting, GrayListing and Caller ID Spoof Detection." It works similarly to Nomorobo, but can be implemented through hardware or software. The FTC describes it thusly:

This solution involves a software application that can authenticate Caller ID information as either authentic or spoofed, and display this information to the customer. It can be implemented through a customer-installed software application on smartphones and certain telephone systems, through updates to smartphone operating systems or carriers’ software, or through a hardware device at the customer premises. In addition to authenticating Caller ID information, the system depends on white and black lists that can be populated manually or autonomously and then aggregated into global white and black lists. Calls with authentic caller IDs on private or global white lists can be put through to the customer and calls from spoofed caller IDs or authentic callers' IDs on the private or global black lists can be dropped. Any number not on a white or black list, or not authenticated, can be handled based on customer preferences, such as forwarded to voice mail or subjected to human verification without ringing the customer phone. Human verification would rely on continuously changing pre-recorded questions presented to the caller, which would be difficult for a computer to answer. The solution stops those who abuse the telephone system without inconveniencing regular callers.

The system developed by Google engineers Klein and Jackson would rely on consumers to report illegal robocalls, making it easier to block calls from known bad numbers. We've contacted a spokesperson for Klein and Jackson to find out whether and when the engineers or Google plan to release this system to the public, but haven't heard back yet.