Earlier quoted context omitted.
In describing how he built his database Turchin says [0] it is about lethal events: "Instability events vary in scale from intense and pro- longed civil wars claiming thousands (and sometimes even millions) of human lives to a one-day urban riot in which several people are killed, or even a violent demonstration in which no lives are lost. In constructing the database I chose to include only lethal events." [0] https…
Does the model predict or account for the Oklahoma Federal building bombing (1996, 168 people killed)?
"Thus, although the most common fatality rate per event was 1 (48% of cases), in rare cases the ‘butcher bill’ could run into hundreds (less than 1% of events had a fatality rate of 100 or more). As a result, the rare but bloody events have a disproportionate effect on the trajectory. A good example of this effect is the latest ‘peak’ in the trajectory (during the 1990s) – it is entirely due to 168 deaths associated with a single event, the 1995 Oklahoma City bombing."