From Marxism through 'the New Criticism' to Academic Nonsense 6 (Column 183)
With God's help
In the previous column I discussed a possible method for the scientific study of questions that lack criteria. I explained that comparison to an agreed-upon case can replace the need for sharp criteria. In this column I want to broaden the picture and show that this method is certainly applicable in other areas as well. This will take us back once again to the columns that dealt with poetry and Hasidism (104-113).
Postmodern art and ambiguous texts
In the previous columns I pointed to the similarity between claims made against "critical" articles on gender and the like, and claims made against postmodern art or postmodern texts of any kind. I described the difficulty of grounding such qualitative claims in a scientific and systematic way because of the difficulty of defining quality metrics in these fields, and in general. I explained that this is the essence of the comparative method adopted by the three pranksters who carried out the bogus-article experiment. I will now show that this method is definitely applicable in broader contexts as well. Instead of formulating quality criteria—a hopeless task (the "who made you the arbiter?" objection, and so on)—one should make comparisons to clear, agreed-upon extreme cases, that is, a text with banal content that by universal agreement has no value whatsoever.
To be sure, it is highly doubtful whether there really exists a text whose lack of value is universally agreed upon. After all, according to our postmodern cousins, a text as such has no meaning, and therefore no value either. Its meaning is created only in the reader and depends on the reader. Therefore, on their view, there is no way to judge the quality of a text. Yet one of the commenters on the previous columns pointed out that even a nonsense text can be judged in terms of its potential to arouse insights in the reader. Even if it has no value in itself, if its purpose is to arouse insights in the reader, then that too is how it should be judged.
You may be surprised, but I completely agree. More than that: I think this is one of the most important criteria for judging the quality of a poem. As I explained in the columns on poetry (107 and onward), a poem is not meant to convey information from the poet to the reader but to evoke in him one impression or another (catharsis), and therefore judging it in terms of its potential to arouse deep and varied impressions in a broad audience is relevant. From the postmodern point of view, this is also how a prose text should be judged. For them, everything is a poem. In contrast, in my view a prose text, such as an academic article, ought to be judged by the information it conveys from writer to reader. Prose is not poetry, and it has a meaning of its own, not only one created in the reader.
But in the previous column we saw, surprisingly, that even the champions of nonsense feel an incomprehensible need to justify themselves when their nakedness is exposed in this way. Instead of proudly insisting that a nonsense text is as worthy as any other text (after all, in every text meaning is created only in the reader, and so there is no way to judge the quality of texts), they find a need to apologize.[1] It seems that even to them it is clear that in these cases there is no justification for seeing any academic value in the text in question. Certainly there is no reason to give credit to the person who wrote it if all meaning is generated in the reader.
In this column I will try to examine a bit more deeply the meaning of the distinction between the meaning of a text and the meaning created in the reader. I will do so by expanding the comparative method. I have already mentioned that there are other candidates for such comparative study, such as (post)modern art, Hasidic texts, or texts in Jewish thought. Of course, there too, in order to make such a comparison, we need agreed-upon extreme cases.
Algorithms of correlations
Let us begin with (post)modern art. There, in principle, one can carry out such an experiment even without direct human comparison (as was done with the articles I described). One can create fabricated artworks by children, jumping monkeys, and best of all programs using random generators, and then test them. The test need not be conducted by way of the comparative judgments of readers and editors as we saw in the previous column (more on this below); one can also use algorithms that expose correlations.
Today there are algorithms that identify whether behind a data file (visual or verbal) there is some thought, that is, whether there is an organizing pattern indicating a plan or thought underlying that information, or whether it is merely a random outpouring with no idea or plan behind it. This is done by checking correlations between different segments of the information. Random splashing of paint will yield a lack of relation between the various parts of the work (a correlation on the order of 0), whereas a planned painting with an idea behind it (whatever that idea may be) ought to yield a higher correlation measure. I think the results could be rather embarrassing for some experts and artists in the field of (post)modern art.[2]
But here I will continue with the comparative method described in the previous column. It turns out that this too has already been done, and the basic tool is available to anyone who wants it, namely a text generator.
Text generators
Such a text generator was built by my friends Nadav Shnerb[3] and Moshe Koppel, together with Arel Segal-Halevi and others. You can choose a figure—Rabbi Kook, Rabbi Nachman, and so on—and with a single click the program produces four texts attributed to the selected thinker and displays them to the reader (as many times as you like, each time a different set of four), only one of which is original and the others fabricated by the generator. The reader is supposed to identify the original text from among the four. The generator is freely available online,[4] and readers are invited to go there and conduct as many experiments as they wish. I will already tell you that the average reader will fail the test in quite a few cases. A more skilled reader will do better (not always. Of course, someone who knows all of Rabbi Kook's texts will probably succeed. Although one can always try telling him that a new manuscript has been discovered).
Example
Here, for example, is a random click I made on the online generator requesting texts attributed to Rabbi Kook, and I received the following four texts (there are, of course, more and less successful attempts. It is all random). For sport, try to identify which of them is the real one before reading on:
|
Well, did you succeed?
Anyone blessed with a truly sporting spirit is now invited to give himself, or others, a lesson on each of these texts under the title "A lesson in Rabbi Kook's thought". Success is guaranteed. I am certain that every lecturer on Rabbi Kook's teachings can derive wonderful insights from any one of these texts, read every letter and line with exquisite precision, expound every jot and tittle, and bring forth a mouthful of pearls, and there is no chance that anyone in the audience will feel that something here is not quite right. I invite readers to post in the comments on this column lessons and ideas about each of these passages (it is highly preferable to do so before you have seen which one is genuine).
Initial analysis
So what exactly is the problem? The insights that will be produced will certainly be deep and important (truly!). Granted, but of course they have no connection to the text being studied. We saw that postmodernism holds that insights are always created only in the reader and are not in the text itself. Therefore, on that view, there is really no problem with the picture I have presented here. The only problem is that we are not studying Rabbi Kook but our profound lecturer. But that too is perfectly fine; there is something to learn from him as well. The question I am dealing with here is only what this says about the texts that Rabbi Kook did write. When we study them, have we really learned something from Rabbi Kook? Apparently not. Do we have any way to evaluate and judge, on their basis, either him or the value of his text? Apparently not. As I explained in the previous column, such an experiment does not testify about the fabricated texts, but about the ordinary ones. What this means is that there is no importance to the texts Rabbi Kook wrote. No less worthy insights can emerge from a text produced by a random generator. In other words, this text is not prose but poetry. It does not convey ideas; it arouses impressions (and perhaps insights as well) in the reader.
And again I will say, as I wrote in the previous column, that it is entirely possible that Rabbi Kook had wondrous and profound insights and original and important ideas. Such an experiment does not prove that he did not, but only that we have no way of knowing this. The insights created in us were created only in us, and they could just as well have arisen from contemplating a telephone pole or a stray cat (each according to his own sources of inspiration). The same is true of most Hasidic texts. I faithfully promise you that a random generator can produce for you material that will keep you occupied with Hasidic study until the end of your days. Again, one can certainly judge these texts as one judges a poem—asking which has greater inspirational potential (that is, a greater potential to arouse more varied and deeper impressions)—but that is not a judgment related to the author's ability as a thinker, only as a poet. The ideas that arise are not his but ours. Therefore I would not grant him a professorship or regard him as an important thinker on the strength of such texts, however popular they may be.
Incidentally, in a comment on column 5, David posted an amusing link in precisely this direction. It describes a similar experiment conducted by yeshiva students with the pamphlet "Olam Katan", to which they sent fabricated passages under the names of various strange and assorted types, and bet among themselves how many of them would be published in the pamphlet. Quite a few indeed were. This is a similar experiment, and its implications for the material published in that pamphlet appear, at least prima facie, to be similar as well.[5]
In the columns on Hasidism I explained that this is the problem I see in secular study of the Torah and the Talmud (especially the aggadic passages, of course). These learners see those texts as sources of inspiration rather than authority, and therefore their interpretation is free and creative. In such a case, meanings really are created only in the learner and are not found in the text itself. And of course this too belongs to the postmodern sphere ("Who made you the authority?!", "You have no monopoly on Judaism", and so on). The problem in such cases is that it turns the process of learning into inspiration-mining, whereas learning, in its essence, is supposed to be the transmission of information (or skill). That is not what happens in all these cases. There is no transfer of information here, only the creation of insights out of inspiration, and that could be done from a random source just as well. If you will pardon me, this is a degradation of the very concept of learning (which is very dear to me). In my view, learning is the close study of texts and the understanding of their meaning, and this is done by the transfer of information from the writer to the reader. Beyond that, there is of course also creativity in learning, and I do not disparage it. But I strongly reject the blurring of these two (which in postmodernism has become an ideology and a cornerstone of philosophy. On their view, all learning is nothing but creation), and certainly the complete elimination of the learning component while leaving creation as the sole component in learning.
Later I will clarify that this is not true of every text by Rabbi Kook, nor of every Hasidic text. But in my estimation it is true of many of them.
The solution to the riddle
For those interested, the fourth passage is the genuine one. Even if you managed to identify it, you must admit that it is not easy (I am sure that even among those who succeeded, most hesitated quite a bit and perhaps also decided somewhat arbitrarily). More than that: even where identification succeeds, one should remember that part of this is simply prior familiarity, and another part is due to the fact that the generator is not yet perfect (technological limitations). Beyond that, in this case the reader (= you) of course knew in advance that there were fabricated texts here (and also knew how many of them there were), and that obviously helps. In my estimation, even a skilled reader who encounters such a text in a printed book attributed to Rabbi Kook (an article from the papers of the late so-and-so, in the memorial volume for Rabbi so-and-so of blessed memory) will study these texts in all seriousness, and chances are that he will not sense it at all.
Beyond that, it is important to remember that Rabbi Kook's texts are indeed ambiguous, but still every one of his passages says something, or at least intends to say something (I at least hope so). But in extreme cases, like the deranged texts published in the above-mentioned "critical" writings, when you do not know in advance that these are products of a generator, you probably will not be able to identify the original at all, because they really do not say anything (there too the ordinary material is genuinely identical to the fake material). This is a fine and promising basis for the cyberattack I described in the previous column.
The next stage in that cyber struggle is of course a random article generator in the fields of nonsense. That is essentially what was done in the experiment described in the previous columns, except that it was done by human beings. I am speaking of a computerized generator that will produce an article in a tenth of a second (and not in six hours, as it took our acquaintances from the experiment. Incidentally, in one of their articles they too used a random poetry generator). That will of course make it possible to carry out a heavy cyberattack, a concentrated bombardment, against the editorial systems of the nonsense journals in the way described above. Quantity is very important, and therefore a computer will do this much better than we can (quality, as noted, does not exist here and plays no role). Let them now go and check, with private investigators, the names of all the senders and the contents of all those millions of articles. Let us see them do it.
Why is this not true of every text?
At first glance, all this says rather gloomy things about the writings of Rabbi Kook, Rabbi Nachman, and the like. If a random generator can create a Rabbi-Nachman-style or Rabbi-Kook-style text that is fit for study just like the original, does that not mean that these texts have no value at all? Seemingly, we have here form without content. If a random generator that imitates only the form creates a text that can be studied seriously in an afternoon yeshiva study session, and one can delve into the insights arising from it, what does that say about the connection between what we learn from Rabbi Kook and what he wanted to say (if anything)? Is there really any transfer of information here, or is this not genuine learning at all? Does this not mean that all meaning is indeed created only in us and not embedded in the text itself? And from this it would seem that the postmodernists are right.
There is a very important point here. They are not right, except with regard to a very particular kind of text. Their problem is that their texts belong to this kind, and there indeed the text itself has no meaning, and meaning is created—if at all—only in the reader, exactly as they describe. But they attribute this to the whole world and to all texts in every field, thereby disqualifying others by their own defect. In an ordinary prose text (remember from the columns on poetry: pure prose is an encyclopedia entry) it seems impossible to do a similar trick, and therefore for such texts the deconstructionist conclusion is false (namely, that meaning is created only in the reader and there is no meaning in the text itself).
I assume that someone who is not an expert in mathematics or physics could not do what our three pranksters did and send articles to journals in mathematics or physics. He could not learn how to get an article in physics or mathematics accepted without having the requisite knowledge, skill, and talent. Without a defined claim that he formulates, there will be no article. Incidentally, the same is true of a book by medieval authorities (Rishonim) or later authorities (Acharonim) on the Talmud. I do not see how a random generator could create a text attributed to Kehillot Yaakov, Ketzot HaChoshen, or Rabbi Akiva Eiger, and survive the scrutiny of skilled students of analytic study (and even of unskilled ones).
What is the reason for this? Very simple: because in those texts, just as in articles in physics or mathematics, we really understand what the author wanted to say. He writes arguments, and their content reaches us as he intended (within the relatively narrow margins of interpretation, which of course exist in every text). This is not a matter of inspiration or of expounding ornamental flourishes; it is a matter of learning and reading comprehension, as learning ought to be. You may be surprised, but there are "inferior" and "not profound" texts, such that reading them is nothing more than reading comprehension, not interpretation requiring inexhaustible creativity. There random generators will not be able to succeed (see the qualification below).
This is essentially the difference between poetry and prose as I defined them in the columns devoted to that issue. Poetry is read and one is impressed by it (or moved by it); prose is learned. Indeed, a random generator will be able to create poems that compete, with a fair degree of success, with poets (at least poets who are sufficiently avant-garde, but to some extent with every poet. Such is the nature of poetry). But no random generator will be able to imitate an encyclopedia entry. The problem with the postmodernists is that they see all texts as poetry and not as prose. Their texts are indeed of that kind, but as I said, they project their own defect onto all kinds of texts and accuse them all of the same thing. This of course circles back and gives them legitimacy to write academic articles that are really poetry rather than prose. If everyone does it, then we at least are aware of it and do it from the outset.
As stated, in my view (and as I remarked above, apparently even the nonsense people admit as much when they let their guard down, by virtue of their very apologies), academic research is supposed to produce prose, not poetry. It is supposed to convey information, not to arouse insights in the reader. As I wrote in those columns, a scholarly journal on poetry does not publish poems but analyses of poetry. That is the role of research, and that is the difference between research and the creation it studies.[6]
Form and content
The fundamental difference between the two kinds of cases—those that can be imitated and those that cannot—is the difference between form and content. What our three researchers did was imitate form. Wherever imitation of form is indistinguishable from an original text, that means that the original text itself is nothing but form without content. In a text that has content and substance (like the examples I brought above), such formal imitation cannot work. The conclusion is that articles in critical gender studies or queer geography are pure forms. They have no content at all, and therefore a random imitation that observes a few formal rules can completely resemble the original and one will not be able to tell them apart. Such an article is basically an empty structure, and if you have imitated the structure then you are not dressed up as a rebbe—you are a real rebbe. By contrast, an article in physics or mathematics, as well as novel interpretations by later authorities on the Talmud, are texts with content, and therefore such an imitation cannot work there.
Incidentally, from here you can understand in which fields and kinds of articles in the humanities and social sciences such an imitation will not work. Articles that make a clear claim that reaches the reader as is, without too much creative interpretation, are articles laden with content, and therefore cannot be imitated by a random generator. Such articles and journals certainly do make claims, and there such an experiment will not work. This is the main limitation on any generalization we might wish to draw from the results of this experiment to the humanities and social sciences as a whole. This experiment attacks very particular fields and genres, and it is not correct to generalize from them to the rest of the fields (journals and areas of research). As for the question of what these experiments mean for the humanities and social sciences in general, we will get to that in the next column.
A random generator for content
But the situation is not quite as simple as I have described thus far. First, I should remind you here that the father of the comparative method as a substitute for criteria is our teacher, R. A. Turing, of blessed memory. The Turing Test in artificial intelligence[7] seeks to do exactly this. It wants to define human conversation or communication, and to ask when a computer may be regarded as human for this purpose. When does it exhibit the characteristics of a conversation with a human being? And again, there is a problem of criteria. There are all kinds of human beings, with different interests and abilities and different levels, and therefore it is hard to formulate criteria that sharply define what human conversation is. Turing proposed replacing the criteria with a comparative method: we place a person in a room with two computer screens, each of which is connected to an entity in another room, one to a computer and the other to a human being. I sit in the room and have to guess, based on a conversation I conduct through the screens, which of the two is the human and which is the computer. Turing proposed the following criterion: once I can no longer distinguish who is the computer and who is the human, the computer is entitled to be regarded as human (or more precisely, its conversation is human conversation). This is precisely the comparative method I described in the previous columns. Turing takes a being that everyone agrees is human (this is the need for a clear and agreed-upon case), and assumes that comparison to it can replace the theoretical definition of human conversation, which as noted is very difficult to formulate.
But now a fundamental and important question arises regarding our discussion. Today it is known that a computerized generator can do the same thing with thinking in other domains as well. Thus one can create a text generator in various fields of knowledge, a generator that will produce meaningful texts with content. In principle, there are computer programs that can prove a theorem in mathematics or analyze a problem in physics (I do not know what level things are currently at, but clearly such computerized generators are possible in principle). These are artificial-intelligence programs, where today we can create learning programs, train them in one field or another, and bring them to the point where they can produce outputs of value in that field.[8] First and foremost among them is the Turing Test itself. Nowadays there are already programs that perform not badly on Turing tests,[9] and it is likely that in the not-too-distant future there will be programs that do far better. Thus, for example, there is no reason why there should not be programs that produce Torah insights at a high level, and learn to create imitations of Kehillot Yaakov or Ketzot HaChoshen.
If so, then seemingly the ground has been pulled out from under the distinction I made above. We are forced to see that all fields, nonsense and non-nonsense alike, are exposed to the same attack I have been speaking about here. We have now learned that even in the exact sciences and in Torah novellae one can create texts and articles with meaningful content by means of a computerized text generator. So the question returns: do the tests and experiments I have described here really single out only postmodern gobbledygook, or are our postmodern cousins right that this is true of all kinds of fields and genres? Perhaps all texts really are forms without content, and the distinction I proposed above is empty.
A formal and a content-based generator
So no, they are not right. Generators of the new type that I have just described do not reflect the problem with which we are dealing here, and therefore they do not lead to the conclusions I was speaking about. Such generators simply imitate our thinking and do what we do, and so it is no wonder that they too can produce outputs like ours. Just as a standard computer today performs a mathematical calculation, and this in no way undermines either the meaning of the calculation or the value of an article that presents such a calculation.
Such a generator comes about because we have simply taught the machine to do what a human being does. Some will say that we have taught it to think (see column 35). By contrast, in the nonsense experiments the imitation was formal only, that is, the generator (which in that case was flesh-and-blood human beings) merely imitated keywords and atmosphere and did not really create content. It did not undergo the thought processes of real researchers, if only because in the fields under discussion there are no such things (neither real researchers nor their thinking). It merely imitated the form of expression and the expressions they use. This is a formal generator, not a content-based one. The experiment we described was based on formal imitation, and therefore it really did prove that there is no meaningful content in the articles in those fields. A content-based generator, by contrast, does not prove that the articles being tested have no content, but at most the fact that a machine too can create content. That is something entirely different.
The conclusion is that there are two kinds of generators: a formal generator and a content-based generator. A content-based generator may perhaps imitate and create valuable content (for example, the programs used in Turing tests imitate content and not form)[10], but that has no principled significance. Here we are speaking about a formal generator. Only a formal generator deserves the title 'random generator,' because it preserves only the form, whereas the content is accidental and arbitrary. A content-based generator is not random in any sense, since its content satisfies rational criteria just like parallel human content.
Incidentally, the experiment mentioned above that was conducted in "Olam Katan", at least if I understood correctly what happened there, is based on a content-based generator rather than a formal one. The boys fabricated material that imitated the content of relevant responses and did not create nonsense passages. They wrote a text appropriate to a wavering single woman or a typical IDF officer, and so it is no surprise that the editors could not tell that it was fabricated. Anyone can write plausible responses by an army officer or a frustrated divorcée, and why should someone who reads them be able to detect that they are fabricated? This is not formal imitation but content imitation, and therefore it does not raise a real problem. The experiment is indeed embarrassing, and personally I would very much like to draw conclusions from it about pamphlets in the style of "Olam Katan" (which I really do not like)[11], but on sober reflection it seems to me that this is not the case.
Demonstration
To sharpen the point, I will take another passage, deliberately also by Rabbi Kook, and we will see why a formal generator could not imitate it. One of his well-known passages is the opening of Orot HaKodesh, where he writes as follows:
Holy wisdom is exalted above all wisdom in this: it transforms the will and the psychic disposition of those who study it, bringing them near to that very sublimity with which it itself is charged. This is not so with all worldly forms of wisdom. Although they depict lofty, beautiful, and noble matters, they do not possess that operative quality of drawing the essential being of the one who reflects upon them toward their own level; indeed, they bear no relation at all to the other powers and essential dimensions of the human being, apart from his intellectual faculty alone. The reason is that all matters of holiness come from the source of the life of all life, from the foundation of life that gives being to everything, and the power of holy content is sufficient to bring forth multitudes of creatures without end, to plant the heavens and found the earth, and all the more so to impress a new and distinct form upon the soul that contemplates it. All secular sciences lack this power, for they do not renew or generate novelty on their own; they only depict and present before the intellectual gaze that which already exists in reality. Therefore they also cannot make the one who contemplates them into a new creature, uproot him from the evil traits of his essential nature, and establish him in a state of new existence, pure and alive in the light of true life that endures forever and ever.
There is no doubt that the formulation is poetic, as is Rabbi Kook's way, but here the meaning is completely clear. There is room for interpretations, explanations, and various connections, but the literal meaning of the sentences in this passage is entirely clear. No one will dispute that his claim is that the study of Torah differs from all other kinds of study, because in Torah the learner himself changes and is acted upon by his study, whereas other studies are information that enters him but does not necessarily act on him and change him. Up to this point every reader will agree, and therefore this is the solid content of the passage. The words have a literal meaning, and it does not depend on interpretation. From this it follows that this is prose and not poetry, and that cannot be imitated by formal imitation.
As stated, one may agree or disagree with what is said, one may seek further depths and meanings, interpret what exactly that external or internal influence is, and whether there is a sharp boundary between the two. All these are good questions, and the answers certainly do depend on interpretation, but first and foremost there is a clear, basic, literal meaning to this passage. Its wording is indeed poetic, but it is prose. It has content and not only form, and therefore a formal generator could not imitate it. The use of characteristic words and inflections, and the application of common formal connections between words in the rest of Rabbi Kook's writings, could not create such a passage. The conclusion is that it is not correct to treat a passage like this as nonsense in the sense described above.
Of course a content-based generator could imitate such passages as well. If we were to teach it Rabbi Kook's mode of thought and his conceptual world, then just like a human student of Rabbi Kook it too could create texts simulating his content. But as I explained, this is an imitation of Rabbi Kook's mode of thought, and therefore it does not necessarily say anything about the quality of his original writings. The passages brought above reflect the character of his poetic writings, where the poetry lies not only in the form of the wording but also in the ambiguity of the content. These are passages in which the fundamental statement itself depends on interpretation and is not self-evident, and therefore they can be imitated formally. Passages written in a poetic way but containing substance whose basic meaning is clear to every reasonable reader—no.
This is all the more true when we are dealing with texts like novel interpretations by medieval authorities (Rishonim) or later authorities (Acharonim). In a typical text of that kind there is a difficulty and a resolution, and therefore their literal meaning is self-evident to every reasonable reader. In such cases there is no doubt that formal imitation cannot create a similar text. Formal imitation cannot give us a meaningful difficulty, nor a resolution relevant to a given difficulty. Content imitation, of course, can do this even in such texts (like a student who learns from his rosh yeshiva and manages to think and create on his own), but that is not important for our purposes.
Is the boundary between content and form sharp?
When I searched in Orot HaKodesh for a passage to bring as an example containing clear and sharp content (and therefore not amenable to formal imitation), I had quite a difficult time finding one. While leafing through it, I understood that the boundary between passages that contain prosaic content (= 'plain,' in Rabbi Kook's terminology) and those that are pure form is not sharp. Rabbi Kook's language is always poetic (to varying degrees), but as we have seen here, even a poetic formulation can contain content. I hesitated over quite a few passages as to whether it was clear to me that they could not be imitated formally, and my answer was not unequivocal.
I assume that the same ambiguity exists between a formal generator and a content-based one. It is not always easy to determine when a generator is entirely formal and when it also contains a content component (or how one constructs a text generator that is purely formal). In many cases, imitation of the form also contains at least some imitation of the content. Thus, for example, what those three researchers did was formal imitation, since they used a given set of words and expressions and created a desired atmosphere without regard to content. But even there it is clear that there was not a complete disregard of content. There were claims there that literally spoke about oppression and the like. So even there, it was not a matter of pure form. Still, it seems to me that the idea is clear and the principled distinction is correct. There are gray areas, of course, because our world is complicated.
One more sentence about futurism
As an aside, I will say that the implication of the picture described above regarding content-based generators is that a cyberattack like the one I proposed in the previous columns could also be directed at serious journals in serious fields. Once we reach a point where content-based generators can produce valuable material, it is not clear what added value human beings will still have in scientific and academic research. At the moment it seems that we are not yet there, because otherwise we would already be flooded with professors claiming tenure on the basis of thousands of articles they had created using computerized content generators. But that time is probably on its way, and not very far off. It will be interesting to see what happens then, but I am not concerned here with futurism.
Interim summary
For the time being I have tried—and I hope succeeded—to demonstrate sharply the dimension of nonsense in the extreme fields discussed here. These are fields in which one can propose a comparative scientific experiment that tests and confirms their bogus character: if random equivalents (that is, the outputs of formal generators) can be presented in such a way that experts (?!) fail to distinguish them from the real material, that is an indication that we are dealing with nonsense in the grave and severe sense of the term.
But it is clear that in a considerable part of the humanities and social sciences, and in the journals in those fields, this is not the case. In most of the articles in those fields there is substance and content, and it is likely that the products of formal generators would not be able to imitate them and get accepted for publication in most normal journals. Normal articles, even in the humanities and social sciences, serve for learning—that is, for conveying ideas and information from the author to the reader—and not merely for generating inspiration. These are texts of prose and not of poetry, and therefore one cannot say there that we are dealing with nonsense in the extreme sense defined here, namely nonsense that at most serves as a basis for inspiration (if at all). The description I have given here characterizes mainly the "critical" fields, which as noted are esoteric and wild margins of the humanities and social sciences.
But it seems to me that the extreme cases I have described here nevertheless say something about the normal material in these fields as well. I have already mentioned the fact that in mathematics and the natural sciences they did not try—and very likely, even if you tried, you would not succeed—to create nonsense that would be accepted for publication (incidentally, it would still be interesting to try). What is the significance of this fact? Does it also say something about the normal material that is produced and published in the humanities and social sciences? In the next column I will try to examine the broader implications of the experiments and of the analysis of them as presented here.
[1] To be sure, we saw that the main thrust of their self-justification concerned the fact that they had not disclosed the forgery of the author's name. I already remarked that apparently they really had no serious problem with the nonsense content itself.
[2] An old dream of mine together with my friend Nadav Shnerb (dating back to our cheerful doctoral days) was to conduct such a study. We thought of approaching Kishon with a proposal that he help finance such an experiment with postmodern paintings (which were so dear to him). In the end the matter never came to fruition, but just now I suggested to a former student of mine that he try to work on a thesis in this direction (for a master's degree in philosophy and computer science). He told me that he had begun to examine it seriously, and I am very curious about the results.
[3] See this in his book Keren Zavit, in the essay on Nitzavim. The generator I used later is found on his website.
[4] See, for example, here: http://imitatorgwt.appspot.com/
[5] I am not exactly familiar with the structure of this pamphlet, but I suspect that it is not entirely so. According to the description, these are readers' responses, and there it is indeed harder to detect fabricated material. Who knows what sort of lunatic-of-the-day is writing to you… Later in the column I will return to this and formulate the problem more precisely.
[6] See again my article on academic research and the prohibition of bias.
[7] See also about it in column 35.
[8] Incidentally, in mathematics it is probably easiest to create such generators, because there, at the end of the day, one is really dealing with entirely mechanical manipulations, without interpretations and other human skills. The human advantage over a machine in mathematics is the ability to skip formal steps—for example, to identify promising directions in advance and eliminate options that probably will not work (especially in fields that are not computational, or are NP-complete, where the computer cannot reach a result in a reasonable amount of time and therefore such shortcuts are absolutely necessary). But in most cases even that advantage is not guaranteed, and programs probably can or will be able to do this automatically, better and faster. Even in non-computational fields, there may in the future be quantum-computing programs and the like that will be able to do everything better than we can. Consider yourselves warned.
[9] However, see for example the criticism described here.
[10] Incidentally, the boundary is very far from simple, and this is not the place to discuss it. I am only pointing to the principled distinction. It is possible to teach a program the form of human discourse, and that itself will create discourse with intelligent content (even without our manually inputting the learning of the content, but only of the form).
[11] My impression is that this is a pamphlet of very one-dimensional preaching under a veneer of openness and the raising of various, burning, topical issues. Full disclosure: I have a personal stake in the matter, since an interview was conducted with me on their behalf and was ultimately not published (as part of that openness and willingness to put problems on the table).
Discussion
I actually thought it was passage 2. In all the other passages there are grammatical problems with subject and predicate. You can say many things about Rav Kook, but he was careful about syntax and grammar.
Truth be told, I didn’t check. The program said it was the fourth passage. Really amusing.
I also thought it was passage 2. Passage 3 has obvious problems; I didn’t understand the fourth passage, so I assumed it wasn’t that one; while the first passage seemed strange to me.
By the way, if the generator is random, that’s really not okay. It could lead to a situation where people understand what is correct and yet the generator tells them they’re wrong, and then they think they have data showing that they’re wrong… and then, when it’s “data,” they may draw mistaken conclusions (about Rav Kook’s writings).
Or maybe not. It really is hard to know. Maybe it takes long original passages that do have meaning and pastes them together. I wonder whether reading the original itself and in the proper context would help with understanding.
I’ve now checked. The feature produces only one correct passage and three fake ones.
Hahaha, I understood what happened. This generator brings 3 original texts and one invented text (not the other way around). The fourth answer was the invented one. The problem is that in the texts it relies on there are some typos, which makes identification harder.
If so, then I had a misunderstanding of the experiment there. I need to check.
In any case, note, Roni, that in your first message you wrote that clearly the last three have no meaning. It has now turned out that at least the second and third are original texts.
It apparently takes original passages and mixes them. There’s an option there to work with Rambam, and it’s very easy to identify laws from the Mishneh Torah—whether the laws are in the same context (and therefore an original passage) or not. Of course with Rav Kook it’s much harder. I also thought passage 2 because in the others the syntax is poor or they say nothing, but as stated the program points to 4 and someone here identified 1 as original
One must distinguish between the clarity of a text and its content. A text can be clear and shallow, or obscure and deep, or obscure and shallow, or clear and deep. When the style is obscure, it is hard to understand what a text is saying. Then it is easy to confuse shallow content with deep content, between genuine tekhelet and indigo imitation, and that same confusion points both to a lack of clarity in the writer’s style (in his thought as well, even if such a style is intentional) and to a shallow reading by the reader (even if he is accustomed to reading such texts). By the same token, a clear text that speaks to the reasoning of most people may easily reflect simplistic and mediocre thinking on the part of both reader and writer. If in the classical period clarity and simplicity of text were considered virtues, then in most modern movements (Expressionism, Marxism, Symbolism, Hermeticism, Surrealism, Futurism) writers, whether in poetry or prose, preferred to write complex and obscure texts.
You certainly can say many things about him, but you cannot say that he was careful about syntax. That is often the reason it is hard to understand him. For example, he starts a sentence with a subject and adds so many modifiers that the modifiers effectively become the predicate (the substantive one, not the grammatical one—there is no grammatical one) without anyone noticing. When he wanted to, he certainly knew how to write with correct grammar, but in his notebooks he was writing for himself and did it very quickly—to the point that he says that dipping the pen in ink disturbs him.
And all that, whether true or not, is not worth a garlic peel to the discussion here.
Rav Kook’s texts reflect a romantic-modern mode of writing. What you are asking for is scientific prosaic writing rooted in the Aristotelian tradition or the Enlightenment. On the contrary: just as to you and your friends non-scientific texts seem like nonsense, so to romantics and modernists “enlightened scientific thought” dealing with spiritual matters sounds tasteless and lifeless. In truth, on both sides there are very few writers who wrote good prose, but on both sides there are people who wrote meaningful content. It is possible to study the writings of Rabbi Avraham Yitzhak HaKohen Kook carefully, interpret them, and criticize them. It is not possible to make a mockery of them by means of a text generator, because that is precisely a kind of postmodern “new criticism,” even if it has a scientific form. By the way, there are forgers who succeeded in forging paintings of great artists or works of great composers and deceived major experts—so what does that mean? That sometimes the “great experts” are not really experts, and that imitation too is a kind of talent.
M., with all due respect, it seems that you simply do not understand the discussion here at all, or what it is about.
If a random formal (!) generator imitates a text in a way that confuses learners, that means that even when they study original texts they are studying themselves. That’s all. What does any of this have to do with what I do or don’t expect from Rav Kook? I expect nothing from him. He has every right to write poetry or in any form he wants. That is not what the discussion here is about.
The forgeries also have no connection whatsoever to our discussion. What has forgery to do with this? If one forges a real painting (copies it), that is obviously unrelated here. And if one forges a style (creates another painting in the painter’s style), that is parallel to a content generator, not a formal one.
Rabbi Michi, with all due respect,
I agree with you that postmodernism is highly problematic in terms of both content and form. However, you extended the criticism to the writings of Rabbi Avraham Yitzhak HaKohen Kook and to Hasidic texts. And you brought an imitation of such texts by a random formal generator as evidence for your approach. In my humble opinion, that is a mistake for two reasons.
First, your claim that poetry inspires, as opposed to prose which conveys ideas, is true only of mediocre poetry and mediocre prose. The accepted definition of poetry is that whose purpose is to instruct and delight. Therefore true poetry requires careful study, except that the way one studies poetry is not the way one studies prose. As David already requested, that they not be read as one reads books, but rather that they be read and meditated upon, and one receives reward for them as for the laws of leprosy and tents. The Book of Psalms, the classics of Chinese poetry, the writings of Plato, Dante’s Comedy, Shakespeare’s plays—these are poetic works that were studied carefully by sages in every generation. Not because of their magnificent style, but because of the deep and exalted moral thoughts hidden within them.
Second, the supposed success of a formal generator in imitating texts in a way that confuses most readers teaches that today most readers do not know how to read and study poetic texts. The idea that an algorithm can imitate an original work is foolishness.
Content and form are interdependent. Wordsworth would say that the poet has deep reasons why every word in the poem was chosen rather than another word, and why it is found in the line precisely where it is.
But what then? Postmodernism is a vulgarization of modernism. For modernist writers and Rav Kook had good reasons for writing in an esoteric style, and each of them had an original style and original thoughts. By contrast, the esotericism of most postmodernist texts stems from poverty of thought and heaps upon heaps of cleverness. Rav Kook wrote: “I need to free my literature from its chains. Why am I unable to write the depth of my ideas in a straight way, without complication, without excessive composition, but rather things in their plain sense, clothing them according to the order of their formation? This is a hidden secret.”
Again, this is going off to places unrelated to the discussion. I did not speak about the question whether Rav Kook or some Hasidic thinker had deep ideas. Some did and some did not. Nor did I speak about whether they expressed them properly or in the right way. I asked one simple question: if it is possible to imitate this text formally, what does that say about the person who studies it? That is all.
If you argue with me, you are in effect claiming that someone who knows how to read deep texts will know to tell me, for every text, whether it came from Rav Kook/R. Nachman or from a generator (a Turing test). My bet is that he won’t. But that is an empirical question that can be checked experimentally. Everything else is irrelevant to the discussion, and the arguments and quotations you brought (Wordsworth and others) do not touch the question I asked, only questions I did not ask. On those you can see my columns about poetry (in which I wrote similarly to what you are writing here).
By the way, a content generator could certainly imitate them.
The conclusions of a new study on Rav Kook’s writings are presented in the book Kabbalat Ha-Ra’ayah by R. Yosef Avivi, where he shows that hundreds of Rav Kook’s paragraphs are nothing but a translation (not conceptual but literally word for word) of the words of the Ari and other kabbalists. In my opinion this is proof that those paragraphs do have meaning (though he agrees, and regrets, that all those who meditate on Rav Kook’s ideas did not realize this and therefore wrote nonsense and gibberish in interpreting his ideas)
How many texts written by people are truly original, and how many are an imitation of original texts, and how many an imitation of an imitation?
It is possible to imitate literary texts formally, and over the generations quite a few excellent formal imitations of literary texts have been written. The question is how one imitates. In any case, literary writing needs interpretation, and the question is not whether one can imitate it but whether one can study and understand it. How many people are capable of writing a careful, orderly, and profound interpretive composition like Orot HaKodesh?
I know a bit of Avivi’s work, and in my opinion he exaggerates in identifying Rav Kook with Kabbalah one-to-one. I suspect there is no proof there of an unequivocal meaning for such paragraphs, and the same can be done for imitative paragraphs as well. But I do not know what he says there.
With God’s help, 21 Heshvan 5779
Words of wisdom and knowledge are acquired through study and toil. Someone who has not studied mathematics or physics, history or Gemara—every text on a subject he has not studied will appear to him as a ‘meaningless jumble,’ and he will not know how to distinguish between a real text and one produced by a random generator.
The writer of the post testifies about himself that in most places he examined in Orot HaKodesh he does not understand the author’s intention—so how does it occur to him that he could distinguish in the Rav’s writings between source and imitation?
And what is the remedy? Let him go to a sage and he will teach him. Today, on quite a number of sites, there are regular lessons on Orot and Orot HaKodesh from Torah scholars who studied the books in order from the mouth of Rabbi Tzvi Yehuda Kook of blessed memory or from his outstanding disciples, may they live long (for example: the lessons of Rabbi Uri Sherki on the “Arutz Meir” site, or the lessons of Rabbi Amir Doman and Rabbi David Turner on the “Yeshivat Ma’ale Eliyahu” site, and elsewhere as well).
As far as I remember, I did not testify anywhere that I do not understand the text, or even the author’s intention. That would be false testimony. I also understand books of Hasidism. And still, what I understand there is to a large extent a product of my own thinking, and my claim is that this is the problem with texts of this kind. In my estimation, this is true even for experts far greater than I am in Rav Kook’s writings (in the type I am speaking about, the poetic kind, as I divided it in the post).
Over the last year the “Nezer David” institute has begun publishing Hug Ha-Ra’ayah, the lessons on Orot HaKodesh given by the editor of the book, Rabbi HaNazir of blessed memory; and what could be better than receiving a lesson on Orot HaKodesh from the one entrusted with arranging it by the Ra’ayah himself, and who in the earlier parts merited private guidance from the “master of the teaching,” who went over the material and commented on it?
Another important book for understanding the Ra’ayah’s words is Milon Ha-Ra’ayah by Rabbi Yosef Klner, in which one can understand the unique meaning of each expression in the Rav’s writings in light of the parallels where a similar expression is found elsewhere in his writings. And R. Yosef Avivi’s book exposing the foundations of the Ra’ayah’s words in Kabbalah has already been mentioned here—and “give to the wise and he will grow yet wiser”!
With blessings, Shatz Levinger
As for the words of mockery heard from people who have not properly studied the Rav’s words, the parable of the drunken beggar is apt: one of the nobles who saw him took a fancy to a prank and brought him to the palace, ordered him washed, and placed in a magnificent room in the palace. When the beggar awoke, he did not know what to make of himself: “Am I a beggar as I remember, or a nobleman as I see? What is my true identity and what is only a dream?” The beggar decided to conduct an experiment: to open a book and try to read. If he succeeded, that would be a sign he was a nobleman; if not—a sign he was a beggar. To his dismay, the man discovered that he did not know how to read, but he recovered himself and reached the decisive conclusion: “In truth I am a nobleman, only all noblemen do not know how to read and merely pretend to read.” The man calmed down and went back to sleep the sleep of the righteous in his noble bed 🙂
And in His Torah he meditates day and night. Rashi explained: at first it is called the Torah of the Lord, and when one toils in it, it is called his Torah.
The difference between a scientific text and a literary text is that a scientific text conveys information, and once the information has been acquired the text becomes unnecessary; therefore most scientific texts of past generations are no longer studied. A true literary text, by contrast—its wonder is of its very essence (“Torah from Heaven”), and its study depends on an interpretive tradition that explains it (“they were taught with the signs of the cantillation.” Rashi explains: He established for it a tradition and signs, both in the words of Scripture and in the wording of the Mishnah), and its general understanding is an ongoing and eternal process of self-creation.
The first is actually real (a familiar paragraph). The fourth is clearly fake (there is no connection at all between the first sentence and the second, nor between the second word in the second sentence and the rest of the sentence).
Shatzl, you wrote, “Another important book for understanding the Ra’ayah’s words is ‘Milon Ha-Ra’ayah’ by Rabbi Yosef Klner, in which one can understand the unique meaning of each expression in the Rav’s writings in light of the parallels where a similar expression is found elsewhere in his writings. And R. Yosef Avivi’s book exposing the foundations of the Ra’ayah’s words in Kabbalah has already been mentioned here”
You should know that in that book Avivi claims that Klner does not understand a single word of Rav Kook and mocks his Milon Ha-Ra’ayah. That only reinforces the rabbi’s claim that everyone understands according to the imaginings of his own heart
With God’s help, 21 Heshvan 5779
To A.H.—many greetings,
Differences of opinion in the interpretation of a text exist with every text. Shall we therefore say that the text has no meaning and “everyone sees his own heart’s imaginings”? When we find a dispute among commentators, we labor to understand the method of each one: what led Rashi to explain it this way, and why did our Rabbis the Tosafists explain it differently, etc.
This is how scholars approach a text and identify the different lines of reasoning that can be said, the linguistic nuances arising from the text, and the parallels that can teach us the intention of the words. When one understands the system of considerations that brought each commentator to his unique approach, one can try to decide between them or propose a new interpretation, and in any case understanding is broadened.
The method of Milon Ha-Ra’ayah is important even if we reach conclusions different from those of the author, just as when we study Tanakh we use a concordance to trace the various meanings of the same word in Scripture, and it is certainly possible that we will arrive at new interpretations.
Likewise, the method of Avivi, who finds sources and parallels for the Rav’s words in kabbalistic literature, is very important, even though it is quite possible that the Rav developed a kabbalistic idea in new directions. And there are writings in which the Rav’s teaching and the ways to apply it through the generations were explained. And other scholars compare the Rav’s thought with other currents of thought, from Israel or from the nations of the world, with which the Rav engaged in intellectual give-and-take.
Both methods are important, and to them one must add the living tradition. After all, the Rav’s writings were not discovered in some cave where they had been hidden away for two thousand years. The Rav spoke about these matters with his great disciples—Rabbi Tzvi Yehuda, Rabbi HaNazir, and Rabbi Charlap—and they left behind
An attempt to understand deep thought must combine all possible methods, and from that combination the overall picture becomes clarified.
With blessings, Shatz Levinger
Paragraph 5, line 3:
…and they left behind students and writings.
The Ra’ayah Kook wrote: Truth is not revealed to a person in fragments, but through a general appearance all at once; and all the toil of studying the fragments comes only to prepare the person for this general appearance. Therefore one who truly understands the writings of Rabbi Avraham Yitzhak HaKohen Kook, after laboring in the study of his writings in their general principles and their minute details, his thought encompasses the paragraphs until they become one clear unified thought (this is what his disciple, Rabbi HaNazir, did in his compilations and editing); whereas one who readily finds it difficult to understand clearly one passage or another from his writings is still at the beginning of the study of his Torah-wisdom. In any case, one may argue that a new original system of thought can become revealed to the many only in a style to which they are unaccustomed, and one may argue that there is no point in publishing things understood only by a select few. However, it is worth remembering that what has always distinguished important poetic texts from empty esoteric texts is mainly commentators who succeeded in explaining and interpreting the text and explaining its value to the many.
The words “and writings in which they explained the Rav’s teaching and the ways to apply it through the generations,” which slipped into paragraph 4, line 2, should be at the end of paragraph 5, which should conclude (after the correction):
…and they left behind students and writings in which they explained the Rav’s teaching and the ways to apply it through the generations.
With blessings, Shatz Levinger
In 1905, someone who presented himself as a Sephardi under the name “Algazi” (apparently he was actually Ashkenazi and his real name was Friedlander) circulated several tractates of the Jerusalem Talmud on Kodashim, which he claimed to have found in the only manuscript in the world. (I seem to recall he claimed he bought the manuscript in Istanbul, but the details are hazy in my memory and not important here.) In time it was proven that this was a forgery, and in fact a rather poor and amateurish forgery. R. Meir Dan Plotzki (author of Kli Chemda) even took the trouble to detail the astonishingly primitive method of forgery used by that Algazi-Friedlander. And nevertheless, great and worthy scholars were taken in, including R. Menachem Ziemba, of blessed memory, may God avenge his blood (who later admitted his mistake), and some say even the “Chafetz Chaim.”
Another example is the SCIgen program, a random text generator that produced computer science papers which were accepted for publication in the most respected venues. (For anyone interested in trying it: https://pdos.csail.mit.edu/archive/scigen/).
Another famous example is the violinist Fritz Kreisler, who at first published his own compositions as “discoveries” of works by earlier composers (including great classical masters such as Beethoven and others). In this case too, leading musicologists were fooled, until Kreisler admitted it was his joke. Today there are random generators capable of composing original music in any style.
I do not claim on the basis of these examples (and many others) that the Turing test proposed by Rabbi Michael is worthless, but that it should be used cautiously. The fact that one can forge texts and fool even major experts does not necessarily indicate that the field is “nonsense.” The rabbi will surely say that all these examples relate to content imitation rather than formal imitation, but how shall we distinguish between the two? (Perhaps that is the subject of the next column…)
Indeed, a content generator can certainly do that. When the technology advances, it will be able to do so even for papers submitted to serious journals. As for the question of distinguishing between the two, I cannot go into that here, particularly since I am not an expert in artificial intelligence.
In previous columns I already explained that forgeries like those that existed in the natural sciences as well (cold fusion, etc.) are also irrelevant to the discussion. Certainly the above-mentioned forgery of the Jerusalem Talmud, which included serious statements (by the forger) about the laws of sacrificial matters. In any case, I completely agree that one should use the Turing test cautiously.
An interesting article on the ynet site:
https://www.ynet.co.il/articles/0,7340,L-5383266,00.html
Kreisler composed works that were pastiches in the style of Baroque works, at a time when it was popular to compose classicizing pieces and when most works by Baroque composers had been forgotten and were unknown to the wider public. He did not compose them as a joke, but because he needed short pieces as encores for concerts and did not want to mention his own name too many times. Later he revealed that he had composed them, and when the critics complained he told them that the name had changed but the value remained the same—and indeed most of them are still performed today.
The “discoveries” of musical works by earlier composers was a common 19th-century technique used by an industry that forged works of Pergolesi and Mozart. Today it is believed that the overwhelming majority of works attributed to Pergolesi are not authentic.
The violinist and composer Marius Casadesus composed a concerto for violin and orchestra and attributed it to the discovery of an old manuscript by Mozart at age 10, proof of Mozart’s genius (a manuscript he did not allow anyone to see). The musicologist Alfred איינשטיין analyzed the work and reached the conclusion that it was a mystification in the style of Kreisler, but the musical world preferred to believe that it was a work by Mozart, included it in the catalogue of Mozart’s works, and it was performed in concert halls by the greatest violinists. Only 50 years later Casadesus admitted that he had composed the work and even requested royalties.
The French writer Alexandre Dumas had assistants whom he taught to write novels like Alexandre Dumas, and they participated in writing the many novels he composed. This is somewhat similar to artists’ workshops in the Renaissance period, where the apprentices and students of the artist participated in the work of painting his creations.
John Milton is considered one of the greatest English poets, and the greatest master of style among English poets. The poet John Philips managed to internalize Milton’s style and imitated it well in composing a poem describing in great detail how to grow apple trees and make apple juice.
In Mozart’s time there were dozens of talented composers so influenced by Mozart’s works that they composed pieces very similar to his (some were even called “the French Mozart,” “the Spanish Mozart,” “the Danish Mozart,” etc.). So what distinguishes Mozart’s works from those of his peers? Simplicity and depth, which an untrained listener is not necessarily able to appreciate.
In summary, until the modern period imitation was not considered a flaw, but a legitimate mode of creation. For example, all of Shakespeare’s plays are based on stories by other writers. But Shakespeare turned mud and silver into gold, whereas most imitators at best managed to turn gold into a bit of silver.
With God’s help, 22 Heshvan 5779
On the limitations of expression, which cannot fully express the depth of thought, the Ra’ayah writes in his letter to his disciple Dr. Benjamin Menashe Levin (Igrot Ha-Ra’ayah, I, Letter 216):
“And as for your presenting European literature as an example, with its precision and clarity—insofar as it concerns me personally, I can admit that I do not possess a properly explanatory style. In any case, generally speaking I will say that I do not envy those people of disclosure who express everything clearly… In many places there are such matters that only if we crush them, if we blur their form until they become so small, as small as our speech is in comparison with the flight of thought—only then can we express them clearly.
But so long as we penetrate into the inwardness of the soul within things and yearn for matters in their very height, in no way will we be able to clarify everything… and often very often we can feel that one who comes to reveal everything to us gives us a trustworthy certificate of his own inner poverty of spirit.”
The way to express deep ideas is gradual revelation, which slowly makes it possible to absorb and internalize:
“Let our words go gradually, so that as the soul becomes fit to sense on its own, from subtle hints, what is stored in the inwardness of things, deeds, and sayings—then that capacity is transformed within it into a force of life, not merely some bare addition of some abstract knowledge,
and as this life-force develops and comes into action, the power of speech is enriched as well, until it can reveal large parts of what is hidden in the inwardness of the soul. And if in the very mode of revelation there come crowns upon the letters, by which loftier, stronger, and greater matters are hinted at—matters that stand in the might of thought’s concealment, preserving the wealth of its life inwardly.”
On the ability of a brief expression to open hidden treasures of spiritual life, the Rav writes in Letter 178 (also to R. B.M. Levin, p. 232):
“In such an atmosphere, in which metaphysical studies were common, expressions like these became current, and a treasury of words for such hidden things was used. And indeed, there is a wondrous active power in each of these expressions, rich in content and brief in wording; it is a dynamic force that works wonders.
In the powers of the soul we see the might of the power of a brief and rich expression, in social life and moral life. There is sometimes an inner quality to feel the force of the richness of some expression even without intellectual cognition, but by means of some hidden feeling. And when immense treasures of knowledge and feeling are enclosed in a single word, when it is accompanied by mighty spiritual preparation on the part of the one expressing it—it acts upon reality in accordance with the measure of spiritual force accumulated in it for the sake of its coming into being.”
In short: speech reveals a handbreadth of the depth of the idea hidden within it; listening to it and internalizing it open a path to deeper understanding!
With blessings, Shatz Levinger
'
:
Paragraph 6, line 2:
…and with the very mode of revelation…
Paragraph 9, line 2:
…even without intellectual cognition…
It seems to me there is an imprecise inference here:
The fact that I am able to give meaning to something that has no meaning
does indeed indicate that there can be interpretation unrelated to the speaker’s intention (because we are dealing with a generator that has no intention),
but it does not indicate that every interpretation is unrelated to the speaker’s intention.
A statistical example:
A police officer flashes red-blue lights, drivers understand there is police and slow down
A malfunction in an ambulance causes red-blue flashing, drivers think there is police and slow down
In both cases an interpretation was given to the message; once the message was real and intentional, the second time it was produced by a random generator
Analogy:
A person reads Kedushat Levi and understands that it is preferable to sleep before midnight and at midnight get up and say Psalms, and not the reverse
A person sees a black cat devouring a white mouse and decides this is a hint to sleep before midnight and say Psalms afterward
In the first case there was a real message and it was conveyed correctly; in the second the message was created by the person himself
But “just because I’m paranoid doesn’t mean they aren’t after me,” and the fact that students of Hasidism want to draw certain conclusions from certain books does not mean they have not correctly understood the intention.
The correct inference would be if the distribution of conclusions were identical with a real text and a random one, but if a real text leads more people to a certain conclusion (which the author indeed intended), why not assume that the writings conveyed the message correctly?
Well, we are already at truly exalted peaks of the art of pilpul.
Before you are four texts, one real and three from a formal generator. You are unable to locate the real one among the four. Conclusion: even when you study the real text, the connection between what you understand and its meaning or the intention of its author is entirely accidental. True, it is possible that by chance you hit upon the author’s correct intention. By the same token, it is possible that you will interpret a gibberish text exactly in accordance with some random intention that passed through its author’s mind. So what? Even a stopped clock shows the correct time twice a day. I understand that in your view this is a good reason to buy a stopped clock? Happy are you.
Even before Enigma was cracked, one could guess that there would be a submarine operation in a certain place at a certain date and time. There is some chance that the guess would hit the target. Why did they bother seating Turing and his group at Bletchley at all? What fools they were…
With God’s help, 23 Heshvan 5779
For understanding the writings of the Ra’ayah, one may also benefit from the lessons of Rabbi Dr. Reuven Fireman, in the series “The Writings of Rav Kook—Reading Comprehension,” on the “Arutz Meir” website (for Rabbi Fireman and his books Emunah Berurah, Besod HaTefillah, and others—see his entry on “WikiYeshiva”), and “give to the wise and he will become wiser still.”
With blessings, Shatz Levinger
Rabbi Michi,
You argue that one cannot imitate a clear literal text formally, and that a poetic text can be imitated formally in a way that makes it hard to distinguish it from the original text. But your argument relies on a literal reading that does not distinguish between a poetic text and a formal imitation of a poetic text. In truth, however, a poetic text requires a poetic reading, and therefore one who studies poetic thought does not merely read words and sentences and passages and their literal meaning, but listens patiently to what is hinted and hidden in the thought as a whole.
Therefore anyone who has succeeded in learning the original thought thoroughly and inwardly will easily identify what is the original and what are the imitations.
With God’s help, eve of Shabbat, Chayei Sarah 5779
In trying to clarify which passage was the original one from Rav Kook, my eye was caught by the strange phrase that appeared in one of the passages: “All is foreseen and the kingdom is given,” which is a play on words based on the well-known saying, “All is foreseen and permission is given.” The word reshut has two meanings: (a) freedom/permission of choice; (b) a governmental authority (for example, “be wary of the authorities,” “do not become known to the authorities,” “the public domain”).
The generator apparently received an instruction to find familiar phrases that form the beginning or continuation for the word “given,” and so, for example, it was drawn from the word “all” to the phrase with the meaning: “All is foreseen and permission is given.” Another instruction the generator received was to replace words with synonyms. Thus “the authority/permission” became “the kingdom.”
The product is apparently random, but an insight emerges from it: the permission given to a person to choose his deeds is not given to him merely as such, but insofar as he is part of the governing authority. The Creator gave man kingship and dominion in the world so that he would be the “executive authority” of God’s will in His world; therefore he must use the freedom given to him מתוך recognition of the great responsibility placed upon him..
We must fulfill: “Look to Abraham your father and to Sarah who gave birth to you,” to learn from their ways how to use the power of independent choice in order to choose what is good and upright in the eyes of God and man.
With blessings for a peaceful Sabbath, Shatz Levinger
The word reshut with a sheva means power, ability, something whose doing depends on a person’s will. The word rashut with a kamatz means government, rule. According to Rabbenu Yonah, the word rashut with a kamatz derives from the word reshut with a sheva. Therefore, to expound “All is foreseen and the authority is given” (with a kamatz) means: “The king’s heart is in the hand of the Lord; He turns it wherever He wishes.” According to Radak, the word rashut with a kamatz derives from the word rosh (head). Therefore, to expound “All is foreseen and the authority is given” means: “Walk before Me and be wholehearted.”
A peaceful and blessed Sabbath
With God’s help, 26 Heshvan 5779
To M80—many greetings,
It seems that according to Rabbenu Yonah’s explanation, kingship is called reshut because the king is permitted to act as he wishes and those subject to him are not permitted to protest. According to Radak’s explanation, kingship is called reshut because the king is the head and leader of his people.
It may be that the connection between reshut and leadership is also that one who is at the head, the leader, must “use his head” and think very carefully about his actions, for if he errs in judgment there is no one to stop him, and his mistaken decision may bring harm to many. And perhaps a “king” is called that because it is his duty to deliberate carefully before each action of his.
In Aramaic, reshu means “debt” (and perhaps rash too is a person sunk in debts and therefore poor in means), and from here there is a hint for one who has authority and leadership in his hands: the great power in his hands imposes upon him as well a great obligation to fulfill his role with care and responsibility.
With blessings, Shatz Levinger
Shatzl’s response, which began to expound on the generator’s mistake, reminded me of Umberto Eco’s book Foucault’s Pendulum. In the book, a group of intellectuals use their broad knowledge, an obscure list from the 14th century, and a random computer program to create a conspiratorial history of humanity involving Templars, the Freemasons, and countless esoteric and historically insignificant phenomena. The book is divided into 10 chapters according to the order of the sefirot (Kindness, Severity, Beauty…). The group gets into trouble when various charlatans and cranks become convinced that their alternative history is real and try to obtain it…
Indeed. He anticipated the era of the flood of books about mystical-historical conspiracies in the style of Brown’s Da Vinci Code and the like. Midrashic readings in the style of the ones Shatzl is making here only prove my claims here.
With God’s help, 4 Kislev 5779
To Y.D.—many greetings,
It is possible to find connections between things that have no planned connection, but in the present case it seems that the generator’s result is not entirely random, and the proof is that in all the texts there are at least sentence fragments that are logical in themselves, and only later do the statements get stuck.
It seems to me that the generator was probably programmed to identify associations, complete sentences that commonly appear in a real text, and also identify synonyms. Thus, for example, when it was given the word “all,” it attached to it the familiar phrase “All is foreseen and permission is given,” and replaced “permission/authority” with “kingdom,” which are words used sometimes in the same meaning.
With blessings, Shatz Levinger
As for texts of esoteric groups as well, it is possible that the associations to kabbalistic concepts such as the ten sefirot may not be random. People who dealt in mysteries sometimes engaged with and knew kabbalistic literature, and it may be that they used concepts from it in one way or another.
In paragraph 2, line 1:
…to complete sentences that usually appear in real texts as a continuation of a certain word, and also…
As far as I know, this generator is entirely formal. It is not a neural network that imitates Rav Kook’s thinking, but a formal splicing together of expressions. That is precisely why I made the distinction between the two kinds of generators.
The properties I identified, which lead to the construction of logical sentences or sentence fragments, are formal.
(a) Associative dragging—completing a sentence based on a given word; thus, for example, the word “all” drags after it the familiar continuation “All is foreseen and permission is given.”
(b) Replacing words with synonyms—thus “permission/authority” is replaced by “kingdom.”
By means of associative continuation and synonym replacement, a formal generator can create logical sentences without understanding a thing or half a thing about the system of thought..
With blessings, Shatz Levinger
That is not correct.
On the contrary, דווקא the first passage is the correct one.
And I didn’t know it beforehand; rather, I inferred it from reading the four passages (only the first one has any meaning).
Afterward I googled it and found that indeed the first passage appears in Shemonah Kevatzim.
So apparently, after all, there is real content there, contrary to what you said.