Where he really lost me was that every fascinating and surprising claim presented by him with absolute certainty that I looked into was either an exaggeration or wrong. As an example, he had a striking statistic that the modal number of friends people used to be able to rely on in an emergency was something like 4 or 5. Ten years later in the present day that number has dropped to 0. Putting aside the fact that a mode can be pretty misleading, I still thought it was an interesting difference in only ten years so I was disappointed to find that the most credible explanation in the literature seems to be that it's an artefact of survey design. Even the original paper he got the statistic from was pretty equivocal on the validity of the finding, so even if he was a bit lazy in not reading through later papers, he still should have been less confident.
I haven't read his book though. The above is just to say, caveat lector.