Gladwell's book contains many good stories, but is rather misnamed, as many of the examples he provides are not about talking to strangers, but misreading people. As a psychiatrist, I often wonder if someone is being truthful or not, particularly when evaluating drug use and suicidal and violent intentions. It was comforting to realize that this is a hard task, and that some people are indeed mismatched, and that experts in interrogation can get it wrong, and also disconcerting. We are taught that suicide is difficult to predict, and that is true. There are lists of risk factors, the primary one being a history of suicide is most likely to predict future suicide. The software algorithm that predicted whether an accused person would show up to their hearing made sense to me, that focusing on risk factors without meeting the person had a better predictive rate than the judge.
Overall, I think there are situations where it pays to be a bit more skeptical than the default truth mode that he doesn't adequately introduce, and I don't think he drew many convincing conclusions. For instance, I agree that drinking too much will make judging other's intentions hard, but I don't think he talked about necessary cultural changes enough, though he mentioned the native tribe that with culture, had people drinking to excess with no issues. The victim of Brock Turner was right, alcohol did not rape her. Gladwell is right that it did make her more vulnerable, but there were many drunk people who didn't rape anyone that night, and I don't think he emphasized this enough.