OPINION
Text in which the author defends ideas and reaches conclusions based on his / her interpretation of facts and data

No one wants to date in a ghost town

Many pioneering platforms depend on large companies as they need vast amounts of data just to make a product viable

The logo for Facebook Dating.
The logo for Facebook Dating.Jakub Porzycki (NurPhoto via Getty Images)

For many years, the dating app Tinder had Facebook integration. This was even after Facebook publicly disabled such integration (in full disclosure I had a research app that was rendered inoperable by such changes in 2015). Yet, Tinder’s access to Facebook notably disappeared in April 2018. This meant users could no longer see if they had Facebook friends in common with a potential match. Further, the change was swift and briefly rendered Tinder inoperable for some users. A mere month later, Facebook launched its own (seemingly underwhelming) Facebook Dating.

This strategy appears pervasive on social media. It’s a cautionary tale we might call “externalizing the skunkworks.” A skunkworks is usually an internal group, like Microsoft Research, that produces products not necessarily meant for public consumption but useful to determine the vector of future products and company strategy. Maintaining internal R&D is expensive and risky given that so much software never makes it out the door. Yet, even if such research is expensive it benefits from internal access to big data. For example, in my own time at Microsoft Research in 2005, I had access to considerable internal mail data that would simply be inaccessible otherwise. In the case of Tinder, it became clear to Facebook that they could set up a competing dating app using their own massive social graph, rather than loan this graph to an external company. In other cases, companies will simply buy the competitor or emerging product. For example, the product I worked on, SNARF, had some features rolled into Microsoft Outlook, while a similar startup called XOBNI, was acquired by Yahoo Mail.

Small companies latch onto larger ones for data access or app integration

For companies and startups looking to make it big, this strategy is full of risks. On the one hand, they can leverage existing platforms with vast stores of data to train their algorithms or inventively repurpose existing data with new interfaces. So that mitigates the potential startup costs. But on the other hand, these companies are at the mercy of continued access to the platform. It seems that opening up APIs (the technology used to link data across companies) to smaller firms means that the firms absorb the risk of failure, while the larger data controllers can sit back, discover who wins and then simply close up the APIs once they have anointed a winner. Twitter, for example, still makes it easy for third parties to display newsfeeds. After a raft of Twitter clients a decade ago, they acquired TweetDeck and integrated its services. Google was notorious for a similar practice, acquiring and shutting down software like Etherpad and Dodgeball. Reddit made it easy for third parties to display a newsfeed, and the company itself similarly acquired AlienBlue, which then became the official iOS Reddit app. On the other hand, Facebook has disabled third-party access to its feed, and the feeds on Instagram. They have discovered that for them, there is more value in controlling the mobile experience than learning from third parties about what works best.

For every TweetDeck, there are dozens of failed clients, broken extensions and abandoned projects. Why do companies bother then? It’s not just the long prospect of striking gold. Many companies simply need vast amounts of data just to make a product viable. Without access to that data, they cannot prove their effectiveness or profitability. Worse, in modern artificial intelligence, small companies might not even be able to train on data at a scale required to make their algorithms work. No one wants to date in a ghost town. So small companies latch onto larger ones for data access or app integration. The reward is rarely to compete with the big platforms, but rather to be acquired by them. Tinder being a notable exception here, so far. For Tinder, the game was played with shrewd marketing and a successful launch (which ironically leveraged Facebook accounts to ensure prospective dates were trustworthy). Yet, the cost startups face is not just in their own investments, but in the risk that their very product could be shut down at a moment’s notice, particularly if another app fits the bill and is acquired first.

There’s an old adage on the internet that if you’re not paying for a service you are the product. Perhaps it’s time to add to that: if you’re not controlling the data used on your app, you’re the free Research and Development team for someone else’s platform.

Dr Bernie Hogan [@blurky] is a Senior Research Fellow at the Internet Institute at the University of Oxford. His work specializes in social media data collection, social identity and social network analysis. He has published peer-reviewed work using data from Facebook, Reddit, Twitter and a host of other platforms. He is currently the director of Oxford’s MSc in Social Data Science and is writing a book on the topic for SAGE Publishing.

More information

Recomendaciones EL PAÍS