@biophil asks: "Why should we be skeptical of my simplified puppies/kittens example?"
The obvious answer is: "Everyone knows that puppies are better, and therefore the correct choice."
However, more seriously I would venture to guess the complications beyond the dichotomous choice, coupled with the visibility of choices to vote on based on appearance/availability altered by previous voters (which are also displayed and create influence) cause a situation that makes the simplified example (although educational) not applicable in full to #steem.
Admittedly, my reasons may be wrong, but may account for steem paying for groupthink.
The superiority of puppies not withstanding, I'd actually argue that being able to see other people's votes makes Steem more susceptible to groupthink. If somehow we didn't get to see people's votes until after payouts were made, I suspect we'd be more likely to vote for stuff we actually like, rather than vote on stuff just because other people are voting on it.
But I do like that you mentioned the oversimplified dichotomy; I wouldn't be surprised if that ends up playing a role.
Also that all votes are not equal because of investment incentives to have more Steem Power to have greater sway impacts visibility and payouts, which is also a differentiating factor not found in the simple example.
Thank you though for bringing up this topic.
I posted the same thing before I read this
Part 4 is up!