This is not a reading of Audrey Tang (唐鳳, b. 18 April 1981, Taiwan's first transgender and first non-binary cabinet minister, and its first Minister of Digital Affairs). It is a methodology post, and her chart is only the example that forced the question into the open. We make no forecast about her, nothing about her private life, and nothing she would not say of herself.

The question is this: BaZi and Purple Star both need a binary male/female input to run — and for a transgender person there is no honest value to feed. What does an astrology system that takes honesty seriously actually do at that point? Our answer is the point of the post.

The four-pillar structure comes from the date — it does not use the luck-pillar direction, and it does not use gender. So this much we can read cleanly, and it is the same at any of the contested inputs:

• Day Master 丙 (Yang Fire); the month gives 食神格 — the "output / creation" structure: making, expressing, teaching.
• 食神 → 財 → 官 (Earth → Metal → Water) — a clean flow from creative output, into value, into public office, with 印 (偏印, 寅 Wood) shielding the self from the 七殺 (壬 Water) pressure. In plain terms: technical/creative output channelled into public authority, tempered by learning.

If you wanted to ask "does the chart have a 科技官 (tech-into-office) shape?" — the honest answer is yes, structurally, and that answer needs neither her hour nor her gender. That is as far as a forecast-free reading goes, and we stop there.

Both engines compute the direction of the timeline — the order in which the 10-year luck pillars (大運) and Purple Star decades (大限) unfold — from a single rule:

陽年-male / 陰年-female → forward (順) · 陰年-male / 陽年-female → backward (逆)

Audrey Tang was born in a 辛酉 (yin) year. So if the engine is told male, the timeline runs backward; if told female, it runs forward. These are not nuances of each other — they are opposite sequences, and they would produce opposite forecasts. The single binary we are asked to supply is the forecast.

That is exactly why we are not forecasting here. To do so we would have to assign her a gender for the sake of the arithmetic — and there is no version of that we are entitled to do.

The deeper point — and the one most systems get wrong by collapsing it — is that a chart uses gender in two structurally different ways:

• Type A — the timeline direction (順/逆). Its classical rationale is the yin/yang of the body at birth. This is the one that flips the forecast.
• Type B — relational role. Which star stands for a spouse (男命看財為妻 / 女命看官殺為夫), and certain gendered life themes. Its rationale is lived/relational, not natal yin/yang.

For a transgender person these two can legitimately resolve to different answers — a relational reading may follow lived gender and partnership, while the direction question is a separate, physiological argument. Forcing one binary through both at once is precisely how a reading becomes incoherent — or disrespectful.

This is not a gap only we notice. Practitioners openly debate whether a post-transition chart should run on original biological sex or current gender identity, with no consensus. The tradition simply never anticipated the question — so anyone who hands you a single confident trans chart is concealing a choice they made for you.

The obvious objection: we recover an unknown birth hour by matching it against life events (校時) — so why not recover the direction the same way? We ran exactly that test. The method, plainly:

• 1. Run both directions — build the 大運 sequence as if 順行 and as if 逆行.
• 2. Place her dated public events on each timeline.
• 3. Check the match — does the active 大運's ten-god fit the kind of event (public office, business, creative output)?

Read honestly, the birth-sex (逆) direction lines up with the career themes somewhat more cleanly — most visibly, only it places 正官 (public office) across the very years she held office. So the events do produce a lean.

And we still will not use it. A visible lean is not a verdict:

• Theme, not fortune. Matching the kind of event is far softer than calling a year good or bad — what a forecast actually needs — and on that harder axis neither direction is clean.
• The fit is half-built-in. Her structure is 食神格 (output-rich) to begin with, so a 食傷-heavy run "matching" a life of technical output is nearly circular — and we went in already knowing the ending.
• One life, almost all of it upward. Her public record is nearly all ascent; with so few hard years, there is very little for the comparison to be tested against.
• The lean points where we have least right to go. The better-fitting branch is the birth-sex one — i.e. "the arithmetic is tidier if we read her against her identity." A thin retrospective lean cannot earn that, and we will not assert it.

So the event test does not rescue us from abstaining — it vindicates it. We looked at the very evidence that resolves an unknown hour, and on the gender question it is too weak, too circular, and too loaded to decide. We show the comparison for transparency about the method — not to assign her anything.

When the direction input is genuinely under-determined, the honest move is the same one the engine already makes for an unknown birth hour, or for a health question: it abstains, and it says so. For a chart like this that means:

• Read the gender-independent structure (above) — and only that.
• If a timeline is wanted, show both directions side by side, label each as a convention, and commit to neither — never silently pick one.
• Treat "which convention is right" as an open question, not a fact. We could not claim one predicts better without a proper study we do not have — so we claim nothing.

We will not assign Audrey Tang a gender so that our software can give you a number. That refusal is not a limitation of the method — it is the method working: an honest system tells you where its inputs run out, instead of dressing a guess as a verdict.

• No forecast — not in either direction, not conditionally.
• No claim about her private life, identity, or body — none of that is ours to read.
• No "birth-sex is more accurate" — that would need a study we do not have, on data too scarce and sensitive to gather lightly.
• No health, no longevity, no death — the engine declines these for everyone.

Most charts do not raise this problem — and when one does, askmingli.com runs BaZi + Purple Star together and tells you plainly when it is sure and when it is withholding.

這不是一篇「唐鳳命盤解讀」(唐鳳,1981 年 4 月 18 日生, 台灣首位跨性別、首位非二元性別的內閣政務委員,亦為首任數位發展部部長)。 這是一篇方法論文章,她的命盤只是把問題逼到檯面上的那個例子。我們不對她作任何預測、不談她的私生活,也不會說任何她不會這樣說自己的話。

問題是這樣的:八字與紫微斗數要運轉,都需要一個「男／女」二元輸入 —— 而對一位跨性別者,並沒有一個誠實的值可填。一套把誠實當真的命理系統, 走到這一步時究竟該怎麼做?我們的答案,就是本篇的重點。

四柱的結構由日期而定 —— 它不用到大運方向, 也不用到性別。所以這一部分我們能乾淨地讀,而且在任何有爭議的輸入下都一樣:

• 日主 丙(陽火);月令成 食神格 —— 「輸出 / 創造」之格:製作、表達、教導。
• 食神 → 財 → 官(土 → 金 → 水) —— 一條乾淨的流:由才華輸出,化為價值,再入於公職; 並有印(偏印,寅木)護身,化解 七殺(壬水)之壓。 白話說:技術 / 創造的輸出,導入公共權位,並以學養調和。

若你想問「這張命盤有沒有『科技官』的格?」—— 誠實的答案是結構上,有,而這個答案既不需要她的時辰,也不需要她的性別。 一個不帶預測的讀法,到此為止,我們就停在這裡。

兩套系統都用一條規則,去算時間線的方向 —— 也就是十年大運、紫微大限展開的次序:

陽年-男 / 陰年-女 → 順行 · 陰年-男 / 陽年-女 → 逆行

唐鳳生於 辛酉(陰)年。所以若告訴引擎男, 時間線逆行;若告訴它女,則順行。 這兩者並非彼此的細微差別 —— 而是相反的次序, 會產生相反的預測。我們被要求填的那一個二元值, 其本身就是那個預測。

這正是我們不在此預測的原因。要預測, 我們就得為了算式而替她指定一個性別 —— 而這件事,沒有任何一個版本是我們有資格做的。

更深一層、也是多數系統因混為一談而搞錯的點:命盤用到性別, 有兩種結構上不同的方式:

• 甲類 —— 時間線方向(順／逆)。其古典依據是出生時身體的陰陽。這一種會翻轉預測。
• 乙類 —— 關係角色。哪顆星代表配偶(男命看財為妻 / 女命看官殺為夫),以及某些帶性別的人生主題。 其依據是生活 / 關係,而非出生陰陽。

對一位跨性別者,這兩者可以合理地得出不同的答案 —— 關係的讀法可隨生活性別與伴侶而定,而方向問題則是另一個、屬於生理的論證。 把一個二元值同時硬套進兩者,正是一個讀法變得不自洽 —— 或不尊重 —— 的方式。

這不是只有我們注意到的缺口。命理界本身就公開爭論:變性之後的命盤, 該以原生理性別還是現性別認同來排 —— 而並無共識。傳統根本沒料到這個問題 —— 所以任何遞給你一張篤定的跨性別命盤的人,都隱藏了一個他替你做了的選擇。

最直接的反問:未知的時辰,我們能用人生事件回推(校時)—— 那方向為什麼不能照辦?我們正是跑了這個測試。方法,講白了:

• 1. 兩個方向都排 —— 分別當作 順行 與 逆行,各建一套大運。
• 2. 把她有日期的公開事件,放上每一條時間線。
• 3. 比對 —— 當值大運的十神,是否吻合事件的類型(公職、創業、創作輸出)?

誠實地讀,原生理性別(逆)向,與她的事業主題吻合得稍乾淨些 —— 最明顯的是,唯有它把正官(公職)放在她實際任官的那幾年。 所以事件確實給出一個傾向。

而我們仍然不會用它。一個看得見的傾向,並不是定論:

• 是主題,不是吉凶。吻合事件的類型,遠比判一年好壞要軟 —— 而後者才是預測真正需要的;在那條更硬的軸上,兩個方向都不乾淨。
• 這份吻合一半是內建的。她本就是 食神格(輸出豐沛),所以一段 食傷 偏重的運「吻合」一個技術輸出的人生,幾近循環論證 —— 而我們本就知道結局才去找。
• 一個人生,且幾乎一路向上。她的公開紀錄幾乎全是上升;艱難之年太少,可供比對檢驗的東西就太少。
• 而這個傾向,正指向我們最無權去的方向。吻合較好的那一支是原生理性別 —— 即「若違著她的認同去讀,算式比較整齊」。一個單薄的回溯傾向擔不起這個結論,我們也不會這樣斷言。

所以事件測試並沒有把我們從「不判」中救出來 —— 反而印證了它。 我們看了那份能解開未知時辰的同一種證據,而在性別這個問題上,它太弱、太循環、也太沉重,無法定奪。 我們把這個比較列出來,是為了方法上的透明 —— 不是要替她指定什麼。

當方向的輸入真正無從確定時,誠實的做法, 與引擎面對未知時辰、或面對健康問題時所做的一樣:它選擇不判,並且說明。對這樣一張命盤,這意味著:

• 只讀與性別無關的結構(如上),也只讀那部分。
• 若需要時間線,就把兩個方向並列呈現, 各自標明為一種約定,並不選定任一 —— 絕不暗自挑一個。
• 把「哪一種約定才對」當作開放的問題,而非事實。 沒有一份我們並不具備的恰當研究,我們無法宣稱哪一種預測得更準 —— 所以我們什麼都不宣稱。

我們不會為了讓軟體吐給你一個數字,而替唐鳳指定一個性別。 這份拒絕並非方法的不足 —— 它正是方法在運作: 一個誠實的系統,會告訴你它的輸入在哪裡用盡,而不是把一個猜測裝扮成定論。

• 不作預測 —— 兩個方向都不,連有條件的也不。
• 不對她的私生活、身份或身體下任何斷語 —— 那些都不是我們該讀的。
• 不說「原生理性別比較準」 —— 那需要一份我們沒有的研究,而其數據太稀少、太敏感,不該輕率蒐集。
• 不論健康、不論壽元、不論死亡 —— 對任何人,引擎都拒絕這些。

多數命盤不會引出這個問題 —— 而當某張命盤引出時, askmingli.com 並列跑八字 + 紫微斗數,並明白告訴你:它何時有把握,何時選擇保留。