[cbhl]

I think it would be disingenuous to consider two years worth of studies "insights"; I think the truly interesting stuff will likely come out of clustering machine-learning algorithms applied to the data ten, maybe twenty years down the line (when some Googlers' children have grown up and we can see the effect or lack thereof on them as well as on the Googlers themselves).

I get the impression that Google errs on the side of "provide more information"; for example, food in the micro-kitchens has been labelled with traffic-light colours for many years now. Many of the interns I met at my last internship there went out of their way to eat more of the food that was labelled "red" because "that clearly means it tastes better". Whether that made the life of the interns in question "better" or depends entirely on the metric you choose to measure with.