A password for Netflix or HBO Go can be a precious thing that’s offered to a romantic partner or gifted from a family member (and saved for future use). But for UK company Synamedia, a password like that is a password to be hunted down. At CES this year, the firm unveiled a new service that uses machine learning to spot shared passwords.
It works like this: a streaming service buys access to Synamedia’s platform, which analyzes data from all its users. It looks at a range of factors, like where an account is being accessed from, what time it’s used, what content is being watched and by what device, and so on. It then looks for patterns that indicate a shared password, and it gives the service provider a probability score — a guess at how certain the system is that it’s found an infringer.
“A typical pattern would be you have a subscriber that is simultaneously watching content on the East Coast and West Coast of the US,” Synamedia’s CTO, Jean-Marc Racine, tells The Verge. “That’s unlikely to be the same person.”
Companies are more likely to up-sell you than shut you down
After that, the service provider can choose what to do. If the sharing pattern is extreme, indicating that the credentials have been sold online to multiple users, for example, they can just shut down the accounts. But, if it’s something a little more harmless — maybe a password shared between a family — they might just send a nudge over email, suggesting they upgrade to a premium account.
“The approach is that people tend to be not too punitive about it,” says Racine. “They up-sell services instead.”
Machine learning is particularly well-suited for this sort of task because it can spot patterns in a big pot of data. More importantly, Racine notes that the “consumption patterns” his company studies are always evolving. What people watch and how they do it has changed hugely in the last few years. It’s better to have a system built with machine learning that can adapt to these shifts, rather than a hard-coded algorithm that has to be updated manually.
The data also reveals some interesting patterns. Synamedia’s algorithms can spot university campuses, for example, as they tend to be homes to a “peak of sharers.”
Racine notes that that increased demand for Synamedia’s services is evidence that the new streaming video market is maturing. At the beginning of a service’s life, he says, companies don’t tend to care about password sharing since it’s an organic way to introduce new users to the service. “But after a while, there is a growing concern about piracy, and [companies] want to ensure they’re maximizing their revenue.”
The new system — called Credentials Sharing Insight — is currently being trialed by a number of firms, though Synamedia won’t say which ones. However, the firm currently sells other services to some of the biggest names in the business, including AT&T, Comcast, Disney, Verizon, and Sky. Just today, Sky announced it was taking a stake in the company.
But what about those doing the password-sharing? Is there any advice for how they might escape the algorithm’s grasp? “Well, I would say, you know, if you enjoy the content, you should pay for it. Then you have nothing to worry about,” says Racine. No sympathy there.