Databases are crap for classifying openings. The lichess database labels the following as the French Defense: Normal Variation
Exactly. Its like i said, databases are terrible at deciding what to call an opening. 1.e4 e6 is a french but no would would call 1.e4 e6 2.d4 g6 a french yet the database has to categorize it as something so it keeps the French classification unless it can identify a direct transposition which it only sometimes prioritizes over the original designation. It may for example, change to call it a hippopatamus if it follows through with bg7, d6, b6 bb7. Lichess database doesnt do it but i believe chessgames database does, but it may be manual reclassification by an astute human.
Here is another example. is 1.nc3 d5 2.e4 the main line of the Van geet opening or merely a sideline for white of the scandinavian? likely is 1.e4 nc6 2.d4 d5 3.Exd5 Qxd5 the mainline of the d5 nimzowitsch defense or merely the d4 sideline of the mainline scandinavian? and when setting up the database to choose when to overwrite the original code classification (say from nimzowitsch defense to Scandinavian above) what is the tie-breaker?
sometimes, like in this example, the main line of one opening is merely a rare sideline in another, but the rare sideline may actually be the more common move order for reaching that position because the sideline of a main opening can till be more popular than the mainline of an obscure one.
I am a thorough pragmatist in this regard, opening classifications are best done by family-resemblance. The flavour of the position is what determines classification when ambiguity arises.
Databases are crap for classifying openings. The lichess database labels the following as the French Defense: Normal Variation