If I had to take a bet, I would bet on a combination of a shortened attention span and unknown long-term effects of technology on the brain development plus environmental factors (for example, it seems like plastics are having an impact in things such as fertility rates and our hormonal balance. This could cause other unknown effects).
My guess would be that we are too rushed, munching way too much information that scatter our brains, too prone to use shortcuts to get ahead (can't afford being deep), too busy to let ourselves sink into the very much needed meditative state, paradoxically too result-oriented to have time for quality, too fast-paced to let ourselves relax, get bored, and sleep enough. We are too reactive, too shallow, too busy, too stressed. We are constantly in an overexcited, coffee-pumped state following the patterns required to grind the results that meet the deadlines. And that's the opposite of a happy, alert, fresh, calm, sensitive state of being, in my opinion, that leads to good IQ results.
I would bet on too much screen time for the whole family; seen it first hand. Parents sitting on their devices all evening and not interacting with their children. It's a missed opportunity for learning, especially at a young age.
That's a disturbing image I hadn't considered. The parents I know (mainly CS people) have talked for years about "limiting screen time" of the kids, but I didn't consider that some other parents might have too much screen time themselves, to the point of shortchanging child early development.
I suspect it means that kids today are a little bit less interested in logic puzzles than their parents, who were themselves turn a lot more interested in them than their grandparents, and IQ testing simply isn't robust enough to accurately capture generational differences in cognitive performance if there actually are any. Instability of IQ testing over time seems a less contentious hypothesis than earlier generations being on average too stupid to accomplish anything...
Interesting topic, in any case.