A fascinating Financial Times article looks at Spotify’s new practice of charging record companies to include specific songs in Spotify’s algorithm (“Spotify must wait to find out if digital ‘payola’ hits the right notes”, 27 November 2020). Alex Barker, author of the article, notes that the famous economist Ronald Coase wrote a famous iconoclastic article about payola in the music industry and how it has existed since the early years of the last century. Music publishers would pay bands to play specific songs; later, in the 1950s, record companies would pay disc jockeys to play certain tracks.

So what is the problem? In 1960, the US authorities outlawed the practice, on the grounds that such payments were made without the consumer realising it. In their judgement, “the public is entitled to know by whom it is persuaded”. Similarly, users of Spotify premium service, who pay for a curated list of tracks, will want to know if Spotify is being paid to play individual songs.

The same applies to any recommender service. If users recommend products on Amazon, we feel cheated if we learn many of those recommendations were in fact paid (the reviewer was given the product free of charge in return for a review, and the reviewer can then sell the product after reviewing it). But for an ecommerce site like Amazon, we hardly expect the recommendations to be without bias.

A few minutes spent examining handbooks on configuring recommendation engines makes it clear they provide guidance to maximise sales, not to maximise consumer satisfaction – for example, Relevant Search (2016) states “To e-commerce, relevant means not just leading users to satisfactory purchases, but also making a buck.”

What about academic content? Do we half-expect that recommended journal articles are the result of a paid placing by the publisher? Or, to be more specific, by the subscription journal publisher, because open-access publishers, unlike subscription publishers, have no vested interest in the articles they publish being more widely read?

It is in the interests of the biggest internet companies, in this case Amazon and Google, to make free and paid listings as indistinguishable as possible. We agree it is unethical to disguise one as the other, and for paid services to deceive users of a free resource. There was an example some years ago relating to European tenders, of commercial organisations buying prominence at the top of search results in preference to the official body, the European Commission, who provide such information free of charge. While basic company information is available free in the UK, many organisations provide a paid service that simply displays what is available free elsewhere.

As far as I see it, the ideal situation would be:

  1. Users should always be told what is a sponsored listing and what is an editorial listing. Sounds simple, but it means Google has to keep its search listings business completely separate to its advertising business – clearly impossible.
  2. Academic recommender services should not include any payment for recommendations.
  3. Recommendations via social media should be clearly distinguished from recommendations via citations.

All easier said than done, unfortunately.