Cumulative Advantage

What it is

Matthew Salganik, Peter Dodds and Duncan Watts published the design in Science in 2006, and the design is the contribution.

They built an artificial music market. 14,341 participants were given previously unknown songs to listen to and download. Some could see how many times each song had already been downloaded by earlier participants; others could not. That is the whole manipulation: identical songs, identical interface, with or without a record of what other people did.

Then they ran it as parallel worlds, so the same songs faced the same choice repeatedly with different populations.

Increasing the strength of social influence increased two things. Inequality, meaning the gap between the hits and everything else grew. And unpredictability, meaning which songs became hits varied between worlds.

In effect

The second result is the one that matters, because inequality alone is unsurprising.

If success were mostly quality, the same songs would win in every world. They did not. A song could finish near the top in one world and near the bottom in another, against the same competitors, decided by what the first few hundred people happened to do.

For anybody choosing from a sorted list, whether songs, apps, books or restaurants, the consequence is specific. The sort order is not a measurement of the field. It is partly a record of the order in which people arrived, and the ranking helped produce itself.

What it does not say

It does not say quality is irrelevant, and the paper is careful here. Success was only partly determined by quality: the best songs rarely did poorly and the worst rarely did well, but any other result was possible. Quality sets the floor and the ceiling.

It does not transfer automatically to real markets, which have marketing budgets, playlists, press and prior reputation, all of which are forms of social influence arriving before the first listener.

And it is 2006, before streaming. Whether algorithmic recommendation amplifies or dampens the effect is a reasonable question and a judgement rather than a finding.

---