Identifying a would-be terrorist: An ineradicable error in the data processing?

Processing fragments of data collected on a monitored person to find out whether this is would-be terrorist (WT) very challenging. Moreover, the process has proven be deceptive, with repeated dramatic failures. To address issue I suggest mirror simple model mimic at stake. The considers collection ground items which are labeled either Terrorist Connected (TC) or Free (TF). extract signal from implement an iterated coarse-grained scheme, yields giant unique item label TC TF. results obtained validate processing scheme correct outcomes for full range proportions items, beside in specific sub-range. There, systematic wrong labeling benefit WT, who wrongly not (NWT). This flaw proves irremovable because it anchored within itself connexion treatment uncertain aggregates, inevitable appear. “natural” allocation aggregates TF label, tune ethical application presumption innocence force democracies, confines diagnosis systematically some NWT. It happens that applying guilt, instead innocence, indeterminate shifts errors another sub-range then zero identifying WT. But instead, NWT Despite being quite far reality, my that, just case, intelligence agencies should investigate overall effect seemingly inconsequential applied locally when uncertain, without any immediate detectable impact processing. If bias true, making public choice could avoid misunderstanding people mistakes repeated.