Scope
Each stage states what it produces, the rule it operates under, and one worked example from the corpus.
No stage may pass on material that the preceding stage could not verify.
The Manifesto sets out what is owed to a life. The Standard sets out what a finished record is checked against. This page sets out the process that produces it.
A guarantee that cannot be checked is only a statement of intent. The process below states what we do at each stage, what we may not do, and the point at which a reader could establish that we had failed.
Each stage states what it produces, the rule it operates under, and one worked example from the corpus.
No stage may pass on material that the preceding stage could not verify.
The stages run in fixed order. Each works only from material the preceding stages verified. Where the record is silent we shorten the output rather than fill the gap.
Academic and historical literature yields tens of thousands of individually sourced entries, covering mainland Greece and the diaspora in Germany, Australia and the United States. One records that 1,236,280 Greeks emigrated permanently between 1955 and 1977, from a population of 8.77 million (Kasimis, pp. 1-2). We audit entries by hand.
Placement determines which baseline applies. Greece-born residents of Australia numbered 12,291 in 1947 and 25,862 by 1954 (Cartledge, pp. 3-5). A man who arrived in 1948 entered a community half the size of the one his cousin entered seven years later.
The United States abolished its national-origin quotas in 1965, after some 120,000 Greek-born immigrants had entered between 1945 and 1970 (Kitroeff, pp. 7-8). A person who applied to enter in 1962 met a quota; one who applied in 1966 did not.
Men were 61.1 percent of Greek migrants to Germany between 1955 and 1973, and 50 percent of those to Australia (Lianos, pp. 120-121). A woman who left for Germany sat further from the ordinary case than one who left for Australia, and we say how far. The measurement also carries the relation established at Stage II. Where the evidence allows, it distinguishes a trait that grew freely from one that formed as protection over a wound the whole community shared. We never manufacture a rupture.
We bind it to the documented interior norms of the person's community and generation, and to what the family recorded. Örs found that Istanbul Greeks frequently declined to turn 1955 into a categorical judgement of Turks (pp. 82-84). Without both bindings, we make no statement.
An editor then checks the draft against the Standard, each factual sentence against its citation and each interpretive sentence against its label. The family may correct the record, approve it, or have it deleted.
We do not publish the structure of the corpus, the models, the prompts, the thresholds or the scoring. The guarantees above do not depend on them: each one holds against the delivered record and the citations printed in it.
Coverage is uneven. Some communities are documented in depth and others thinly, and a record drawn from a thinly documented community rests on fewer anchors than one drawn from mainland Greece.
Auditing is continuous rather than finished. Where a claim cannot be anchored, we shorten the record rather than fill it.
Take any sentence in a delivered record. It states whether it is fact or interpretation. Ask us for the citation behind a fact, for the community and years an interpretation is bound to, and for the reviewer who approved it.
The complete checklist is the Standard.
Six stages in fixed order, each able to reject the output of the one before it. Every sentence we deliver can be traced back through them to a citation or to a label.
Monica WagnerFounder