I find the turkey illusion interesting: a turkey is fed by the butcher every day. Day 1, food arrives. Day 2, food arrives. Day 999, food still arrives. Based on the data, the turkey may believe the butcher is a friend and the future is safe. Then day 1000 comes, and it is Thanksgiving. The turkey had a lot of data, but not enough context.
I do not read this as a reason to ignore data. The data was not wrong. Maybe it was just incomplete. The turkey saw the pattern, but not the context around it. That is the part I find useful to keep in mind when looking at data, forecasts, and repeated signals.