Mark is suffering from a pretty classic engineer's problem. Somebody asks what they think is a simple yes or no question, but they either don't understand the complexities of the question, or won't be able to properly understand the implications of the yes or no.
I felt the same. I've never been a fan of his but I have to say, after watching bits and pieces of the testimony, I felt sympathetic for his position. Senators pushing for yes/no answers to questions where there are many many variables and different situations to consider.
Pretty much any question on the topic is too complex for a simple yes/no. And it's abundantly clear to me that 99% of the general public just do not understand that at all. Most people (senators included) were just looking for him to say something damning which they can endlessly bring up in the future.
I don't believe Mark got into this position out of malice. Facebook just grew naturally over time and turned into the beast it is today. How many times have you been building an app and thought "If only I knew <x> about my users I could provide <really cool feature>"?. Each one of these little steps encroaching on user privacy is small and hard to notice for the users and even for the engineers. But over time they add up and at some point there is an "oh shit" moment where you realise what you have.
Yes, some of us have been warning about this for _over a decade_. But the benefits clearly outweighed the costs in the eyes of the users. Only now are the general public starting to realise what the actual costs are.
> For instance: are whales fish? That's a simple yes or no question, right?
You make a good point. While Mark was presented with a similar situation, the answer to the question, "Are whales fish?" could be, "That isn't a 'yes-or-no' question because [reason]," as opposed to, "I'll have my team follow up with you after my congressional hearing."
The latter is safer. I guess it's kind of the same thing as "don't talk to the police" meme in US - everything you say can, and will, be used against you. So correcting malformed/misinformed questions is dangerous, because you may trip up when explaining - or say something that will be misinterpreted.
For instance: are whales fish? That's a simple yes or no question, right? http://inference-review.com/article/on-being-a-fish