Emma Zajdela

It's hosted at the University of Rhode Island.

And you could go and you could download images of these dresses dating back to, I think the latest one in the collection is actually in the 1860s.

Wow.

Although those were all floor-length dresses.

So what we found in the data is these really cool two different clusters, I would say.

The first one is that there's always floor-length dresses.

So the maxi dress is always in, always has been.

In the past, it's actually kind of funny, in the 1860s, for example, they would have these long trains, so in some ways would even extend past the length of the floor, if you think of it that way.

But so there's this maxi cluster that always exists.

And then there's an upper cluster of shorter dresses that oscillates.

So this looks like the sine wave that you might have seen in school.

This just goes up and down over time.

But then something happens in the mid-1980s.

In the mid-1980s, you have this oscillation, and then this actually starts to split.

So instead of having two clusters where you had sort of short skirts that would oscillate and then floor-length skirts, now you have three clusters.

You have many skirts.

midi or these kind of ankle length skirts and then floor length skirts.

Yeah, that's exactly our idea.

So the model that we developed to explain these cycles is based on an idea from psychology called optimal distinctiveness.

And the idea of optimal distinctiveness is that for innovations to be successful, they need to be different from others, but not too different.