线段不是梯子,是账
The Divided Line Is Not a Ladder but a Ledger
太阳那个比方讲完,格劳孔还没缓过来,苏格拉底就接着往下画了。他说,你再拿一条线来。
他前面拿太阳抵过一回,说本钱今天付不出,先给点利息。这条线是第二笔利息。本钱还是没还,他还的又是一样看得见的东西,好让人顺着它去想那样看不见的。
这条线在洞穴的前一页。顺序值得留意:先有太阳,再有这条线,最后才是洞穴。三样东西是一口气讲下来的,讲的是同一件事的三个面。
线怎么画,他说得很具体。把一条线切成不相等的两段,一段归看得见的那一头,一段归想得到的那一头;然后每一段再照原来的比例切一次。这样一共四段。
四段各是什么。最下面那一段是影子和倒影,水面上的,光滑东西上映出来的。往上一段是投下这些影子的东西本身,活物,草木,人做出来的器具。再往上一段是数学那一路,几何学家画一个方出来,讲的却不是这个画出来的方,是方本身。最上面那一段,他说的是另一种走法,不借任何图像,从一个念头走到另一个念头。
第三段那里有个细节值得停一下。他们画出来的那个方,属于第二段,是看得见的东西;可他们嘴里讲的不是这个方。也就是说,第三段的人一边用着第二段的东西,一边说我讲的不是它。这个又用又不认的姿势,柏拉图写得很准。
到这里,一般的读法就说,你看,四层楼,越往上越真。这个读法不算错,他自己后来也是这么排的。可是四段之间最要紧的那一处分别,不在这个排法里。
最要紧的一处在第三段和第四段之间,而且它跟对象没有关系。
他说,几何学家拿奇数偶数,拿各种图形,拿三类角,当作已知的东西。这些东西他们不再交代,不管是对别人还是对自己,都不再交代一遍为什么。他用的话很不客气:他们不屑于给出交代。拿这些当出发点,一路往下推,推到他们本来要证的那个结论为止。
另一路不这样。另一路也从假设出发,可它把假设当踏板,不当地基。它踩着假设往上走,去问那个假设自己凭什么。
所以第三段和第四段的分别,不在于你研究的是图形还是理念,在于你拿自己的起点怎么办。当作已知,不再交代,就是那一段;当作还可以再问的东西,往上追,就是这一段。
这条分界线,SAE 这边整个认下来。
SAE 那边一路在说的一件事就是,你承担的那些前提得一直摊在受审的位置上,不许供起来。前提当然要有,不设前提连话都开不了口;可是设了之后不许贴上免检的条子。一贴条子,那句话就从可以被问的东西变成了不能被问的东西,而那正是麻烦开始的地方。
柏拉图那句不屑于给出交代,说的就是贴条子这个动作。他自己看得极清楚,而且他把这个看清楚的东西当成了分界线。这一处不是我们替他找的对接,是他亲手划的那一刀。
可是接下来他做了另一件事。
他把这条关于做法的分界线,折算成了一把梯子。
他不是偷偷折算的,他说得很明白。要紧的是它明说得有多早。
回头看画线的第一句指令:切成不相等的两段。高低在这一句里就已经露面了,比任何论证都早。哪一段长哪一段短,比例定下来的那一刻,清楚和不清楚已经分了先后,后面的话都在这条线上说。
得说准一点:排序不是从这三个字里长出来的。他在别处另有明说,四种状态按各自分到多少真理,就分到多少清楚。所以把线改成等分,排序不会跟着消失,消失的只是它被画进图形的那一次露面。不相等这三个字要紧,不在于它是排序的来源,在于它是排序最早的一次露面,而且早到人不会把它当成一个主张。
折算之后发生了什么,值得说慢一点。
原来那条分界线问的是:你对你自己的起点交代过没有。这是一个可以对账的问题。有人问你凭什么,你答得出或者答不出,答不出就补,补不上就承认这里欠着。整件事在同一个平面上进行。
折算成梯子之后,问题变成:你在第几层。这就不能对账了。你在第二层,那是你的位置,不是你的欠账。补不了,只能往上爬。而爬这件事有没有路,路怎么走,谁说了算,全是另外的问题。
我这一刀漏了什么,和你站得低,是两句完全不同的话。前一句可以一笔一笔对,对不上就补,补不上就写在那里;这不等于每一笔都能在原来那本账里还清,有些还不清,可它至少还在账上。后一句连账都没有,只能认。
而且梯子一立起来,就会长出一样东西,位次。上班的人对这个过程都不陌生。本来问的是这件事你想清楚了没有,问着问着就变成了你是几级。那个级别一开始是拿来标记想清楚到什么程度的,标着标着,级别自己成了要紧的东西,原来那个问题反倒没有人再问了。梯子不是不好用,是它一好用,就容易把它本来要衡量的那件事替换掉。
SAE 那边也有四格。
先把话接上。认识一样东西,说到底是划界,把这一边和那一边分开,本系列一直把这个动作叫作落刀。四格说的就是这一刀。
那四格是横着排的,不是竖着排的。它们不是四层高低,是同一刀可能欠下的四种账。
你划一条线,把红划出来。这一刀之后,可能欠下的账有四种。
第一种,收进来的这一边没有收干净。红这一侧还有你没有收进去的东西,亮度,浓淡,此刻它落在谁的眼睛里。这一种最普通,也最躲不掉。
第二种,划错了边。本该在这边的被划出去了,或者本不该在的被收了进来。这一种事后能纠。
第三种,两边都够得着。同一样东西,按这个条件算红,按那个条件算不红。这不是它模糊,是你没说清按哪个条件,条件一补上,两属就散了。
第四种,这把尺子根本管不着。有些东西落在这套判据的场子外面,你说它红或者不红都不对,因为这个问题在这里不予结算。它不是假的,是没有资格进这本账。
这四样在 SAE 那边各有名字,收纳之亏,错置之亏,两属,判外。名字用不用都行,要紧的是它们是四种问法,不是四个高度。
家里换季整理衣柜,四样一次全齐。你把衣服分成夏天穿的和冬天穿的。分完之后,夏天穿的那一堆里,还有一大堆这个分法照顾不到的事,哪件配哪件,哪件领子已经磨了,这是第一种。有件外套你顺手放错了堆,这是第二种。有件薄外套春秋都穿,两边都想要它,其实是你没说清按气温分还是按场合分,这是第三种。角落里那把伞,既不是夏天穿的也不是冬天穿的,它根本不是衣服,这个分法管不着它,这是第四种。
四种里没有哪一种更真。它们记的不是东西的高低,是你这一刀的亏欠。而且这四种不称穷尽,也不互相排斥,同一刀可以同时欠好几种。
再换个说法。同一件事,你可以说我大概知道,可以说我算过,可以说我亲手做过一次。这三样常常被读成一样比一样扎实。可仔细看,它们是三种不同的知道法,各有各的漏。大概知道的人抓得住整个形状,说不出细节;算过的人数目对得上,可能从没碰过实物;亲手做过一次的人手上有实感,却容易把那一次的偶然当成普遍。你要把它们排个高低,总能排出来,可是排完你就看不见各自漏的是什么了。
这里得说一句公道话。往上一层这个感受是真的,不是幻觉。学会一件难事,确实像是上了一格,站的地方不一样了,回头看以前的自己看得清清楚楚。这个经验 SAE 不否认,也不该否认。分歧只在于这份感受该记在哪一栏。记在实得那一栏,它是真的,你确实取走了东西;记在高度那一栏,它就开始替你回答一些它其实回答不了的问题。
拆掉梯子的人,得回答梯子原来管的那件事。
他划的那条分界线,你刚刚整个认下来了,而且认得很痛快。可是我那条线,本来就是用来分高低的。不给自己起点交代的那一路,和肯往上追问的那一路,在他那里不是并排的两种做法,是一低一高。他说得明明白白:清楚的程度,按它们分到多少真理来定。
于是他会问:那条线你认,那条线分出来的高低你不认。你把线拿走,又把线分出来的东西丢掉,这是想两头都要。
他还有第二句。我画这条线,是画给一群将要管事的年轻人看的。他们要知道该往哪儿使劲,先学什么后学什么,什么时候算是可以往前一步了。梯子做的就是这件事。你那本账,一条一条记着这一刀漏了什么那一刀漏了什么,一个想学点东西的人拿在手上,该怎么办。
还有一件他可以顺手指出来,而这件事是 SAE 自己说的:那四种问法不称穷尽,欠账的种类本身没有底。
于是他问:你那本账,一共几笔。梯子上的人至少知道自己在第几级。
这一问落得很实。梯子告诉你脚往哪儿放,账本只告诉你还欠着什么。要一个人真动起来,多半还是梯子更管用。
最后回到那条线本身。
如果当初那第一句指令不是切成不相等的两段,而是切成相等的两段,会少掉什么。
排序不会少。他在别处照样说得出,谁分到的真理多,谁就分到的清楚多。四种分辨方式也照样是四种,几何学家照样不交代他的起点,另一路照样往上追问。
少掉的只有一样:那把梯子就不再是画出来的了,它得靠一句一句讲出来。一件要靠讲的事,是可以被问的;一件已经画在那里的事,通常没有人问。
Four Segments, One Hierarchy
After the sun, Socrates draws a line. Divide it unequally into the visible and intelligible, then divide each segment again in the same ratio. The result is the familiar fourfold scheme: images and reflections; visible things; mathematical reasoning; and dialectical understanding that proceeds without relying on images.
The mathematical segment contains an exact observation. Geometers draw visible figures while claiming to reason about the square or diagonal itself, not about this imperfect drawing. They also treat odd and even, the kinds of angle, and other starting points as given. They do not provide an account of them. Dialectic treats a hypothesis differently: not as a foundation but as a stepping-stone from which inquiry ascends to ask what supports it.
The decisive contrast therefore concerns method before it concerns objects. What do you do with your own beginning? Treat it as settled and reason downward, or keep it exposed and reason toward an account of it?
SAE accepts that distinction. Premises are unavoidable, but none should acquire immunity merely because reasoning had to begin somewhere. Plato sees the danger clearly when he says mathematicians decline to give an account of their assumptions.
When an Audit Becomes a Ladder
Plato then converts this methodological difference into rank. The instruction to divide the line unequally already gives the image a vertical logic: degrees of clarity correspond to degrees of truth. What began as a question—have you accounted for your starting point?—becomes a location—at what level do you stand?
Those are different grammars. A debt can be identified, paid, revised, or left openly unpaid. A lower level is a status. It cannot be corrected line by line; one must ascend, and the map of ascent requires someone to identify the route and its authority.
Institutions repeat this conversion constantly. A question such as “Has this person understood the problem?” becomes a grade or professional level. The level begins as a shorthand for understanding, then replaces it. Rank is useful precisely because it is easier to administer than the quality it was meant to record.
SAE therefore lays its four categories horizontally. They are not degrees of reality but four possible liabilities incurred by a distinction.
Four Entries in the Ledger
Suppose a criterion marks something as red. The cut may leave at least four kinds of account unsettled.
First, the included side remains incomplete: hue, brightness, context, and the perceiver's condition exceed what “red” has collected. Second, the boundary may be misplaced: something may have been wrongly included or excluded. Third, an object may belong to both sides under different conditions; the overlap is resolved only when the governing condition is stated. Fourth, the criterion may not apply at all. Asking whether an umbrella is summer clothing or winter clothing fails because the object is outside the field of that division.
SAE calls these, respectively, a deficit of inclusion, a deficit of placement, dual belonging, and being outside adjudication. The labels matter less than the form. They are questions addressed to the same cut, and more than one can arise at once. None is “truer” than another.
Different kinds of knowledge illustrate the point. Having an overview, calculating a result, and doing something by hand are not simply three heights. Each gains something and misses something. The overview can preserve shape while losing detail; calculation can be exact without contact with the object; practical experience can mistake one encounter for a universal rule. A ranking can always be imposed, but the ranking hides the different remainders.
What the Ladder Still Does Better
The feeling of ascent is not an illusion. Learning a difficult practice genuinely changes where one stands, and the former self can become newly visible. SAE records this as acquisition rather than altitude. What it refuses is the further inference that greater acquisition answers every question of direction, value, or finality.
Plato can object that SAE has taken his best distinction while discarding what made it educative. His line is drawn for future rulers. They need to know where to direct effort, what to study first, and what progress looks like. A ledger identifies debts but does not tell a learner where to place the next foot.
The objection intensifies because SAE does not claim its four liabilities are exhaustive. The kinds of debt may themselves be open-ended. Plato's learner at least knows which segment he occupies. The SAE learner receives an account that can always acquire another line.
This is the real exchange. The ladder risks turning inquiry into rank; the ledger risks turning inquiry into endless audit without orientation. Plato provides direction at the price of fixing a hierarchy. SAE preserves revisability at the price of leaving motion contingent.
If the original line had been divided equally, Plato could still have argued that some cognitive states contain more truth and clarity than others. What would disappear is only the silent geometry by which the ranking is made to look already present. A hierarchy stated in sentences can be questioned. A hierarchy built into the picture is often accepted before anyone notices that a claim has been made.