As they say in the article, it’s intended to be better than the pattern where people have to turn left across oncoming traffic to enter or exit the highway. That often also has stoplights.

A cloverleaf requires more width to fit. The highway right-of-way may not be wide enough, or the width may have been “eaten up” by widening the highway. The more lanes going through, the less room on the sides for exit ramps.

The cloverleaf doesn’t need any stoplights so in theory it should have the best continuous throughput. However it does require mixing lanes, which can lock up more easily under heavy traffic than separated sequential exit and entry lanes.

Cloverleafs also suffer a lot more when merging traffic becomes backed up, since there's no guarantee that the traffic merging on/off will be at even close to the same speed as the traffic that isn't merging.

It turns out that engineers select different interchange designs based on the expected traffic patterns, and there is no one-size-fits-all method.

As a sidebar, my dad was part of the (huge) team of engineers that helped implement the first few diverging diamond interchanges, and so its a fun reminder of him, even if I'm not on a road he had any hand in designing.

I'd be interested in the story of how these designs got developed, but there's not much media centered on traffic engineering.

there are a lot more conflict points in the cloverleaf, the diverging diamond reduces locations where a crash may happen. and there are a lot of busy interchanges that don't have the land for a full cloverleaf. it may feel very similar to the driver; having to wait at one light, and then the next one. but we have a few diverging diamonds in my area and they work way better handling the higher traffic volumes we're seeing with recent development push.