There is no single classification of Greek. There are scholars' proposals, each answering a different question, for a different period and a different area. The Atlas stores each scheme with its author, work, edition, purpose, time window and criteria, and each membership of a unit as a separate cited assertion. There is no "parent dialect" column, consensus tree, invented similarity score or count of "all dialects".
Three original approaches, three scales
The Atlas has directly examined three original approaches: Hatzidakis's broad division, Trudgill's feature map and Pantelidis's narrower historical zone proposal. They work at different scales and must not be presented as three rival exhaustive taxonomies. GRDD+ is a different kind of object: a partition of a computational dataset.
Disagreements that must stay visible
North and south vocalism is a useful criterion, but it cannot stand for every Greek tradition. VLACH explicitly limits its reach and distinguishes its collection organisation from linguistic classification B05 Further classifications. Newton's Cretan and Cycladic group and Trudgill's smaller feature areas are not interchangeable nodes B02 §3, §6. Disagreement is not settled by majority vote. The time window, inclusion rule, feature coverage and dependency are recorded. Trudgill relies substantially on Newton and Kontossopoulos, so three names are not three independent field surveys.
Tsakonian's Doric trajectory needs specialist historical qualification. "Not through Koine" in a broad overview does not prove there was no contact with Koine. The place of Pharasiot and Silliot relative to Cappadocian is unresolved. Modern standardisation is not an ancestral node.