A different output should lead back to a named input, dataset, or rule.
First check whether the calculators received the same instant
A date, clock time, and city identify an instant only after a historical time-zone rule is applied. One calculator may use the current offset, another the recorded historical offset, and a third may silently guess a nearby city. Daylight-saving changes can create two possible instants for the same clock label or make that label nonexistent.
Coordinates matter when True Solar Time is used. A city-centre coordinate and an exact birthplace coordinate can differ enough to cross an hour boundary when the time is already close. Before comparing pillars, compare the resolved IANA zone, UTC offset, coordinates or precision label, and the instant each calculator actually used.
Year and Month Pillars depend on solar boundaries
Some products change the BaZi year on January 1, Lunar New Year, a date-only approximation of Li Chun, or the exact Li Chun instant. Those choices can produce different Year Pillars around early February. MingWeave uses exact solar-term instants from its versioned data and reports the nearest boundary in the calculation trace.
The Month Pillar does not follow Gregorian month starts. It changes at the twelve Jie solar terms. A calculator that rounds a term to midnight, uses a different ephemeris, or confuses Jie with every solar term can shift the boundary. The difference is concentrated near a term, so the timestamp and time zone must be compared alongside the pillar.
Day and Hour Pillars expose school conventions
Many systems treat civil midnight as the start of the new sexagenary day. Others advance the day at 23:00, the start of Zi hour. Near that interval, the convention changes the Day Pillar and therefore can change the Hour Stem as well. A calculator should name the choice instead of presenting it as an invisible default.
True Solar Time adds a longitude correction and the equation of time to the civil clock. That can move the selected time into another two-hour branch or across the chosen day boundary. It is not a time-zone replacement, and applying only a longitude shortcut without documenting the equation of time is a different method.
Use a difference table, not a winner label
Record each calculator’s input normalization, year boundary, month boundary, time basis, day boundary, solar-term source, and data version. Then compare Year, Month, Day, and Hour separately. If only the Hour Pillar changes, investigate clock conversion or the two-hour branch. If Year or Month changes near a term, inspect the exact boundary.
MingWeave’s comparison tool holds the birth input constant and changes only an approved Method Profile. It labels which pillars stayed the same and which changed, with warnings when True Solar Time crosses a boundary. A reproducible explanation is a stronger result than confidence unsupported by a visible method.