Seguimos con el proyecto mensual de ojear la revista Byte… con cuarenta años de retraso (tenéis todas las entradas sobre el tema, que ya son unas cuantas, en la etiqueta Byte de este blog). Y febrero del 86 se dedicaba… al procesado de textos (que, spoiler, no es lo mismo que los procesadores de texto).
Y comenzamos mirando publicidad. El primer anuncio, diría yo, de un programita que seguimos usando cuarenta años más tarde: ¡Excel! Dice la Wikipedia que fue lanzado en septiembre del 85, y si vais a nuestra entrada del número de mayo del 85 (sí, llevamos ya un tiempín con esta historia de la revista Byte) encontraréis el anuncio de que lo iban a lanzar, y corregidme si me equivoco (ya podría ser, ya), pero no lo habíamos vuelto a ver por aquí.
Y si os ha llamado la atención el ratón monobotón, o el disquete de 3,5″… sí, Microsoft lanzó originalmente Excel solo para Mac.
No pongo captura, pero también merece la pena pararse en la sección de cartas (página 24 y siguientes), en que los lectores revisan el programa para calcular π (¡del número de mayo!) y explican lo lentísimo que es convergiendo (pero destacan que es muy legible y un buen ejemplo para aprender) y algunas correcciones al programa sobre la distribución normal (esta vez solo tenemos que retroceder hasta octubre). Bravo por los lectores atentos.
Seguimos, esta vez con nuestra manía de pararnos en cualquier cosa que tenga que ver con el Amiga. En este caso, se trata de una introducción al Kernel, el software de sistema contenido en su ROM, escrita nada más y nada menos que por su creador, el mítico (en círculos reducidos, cierto) RJ Mical. Si alguien quiere leer más sobre el tema, en el mismo Archive podéis encontrar su manual. #YaNoSeEscribeSoftwareAsí
Y unas páginas más adelante nos encontramos un anuncio del Amiga que es un homenaje (merecidísimo) a Denise, Paula y Agnus, los tres chips especializados en vídeo, audio y gestión de memoria, revolucionarios para la época, que eran una de las partes vitales para hacer del Amiga la maravilla multimedia que era.
Y dejamos el Amiga (hasta que nos den la más mínima oportunidad de recuperar el tema 😅) y entramos en el tema del número, el procesado de textos. Hablando con la leyenda de la informática que es Donald Knuth (se lee Kanuz, por cierto), hoy profesor emérito de Stanford, creador de TeX y autor de la magna opus The Art of Computer Programming (in progress). Por aquella época ya hacía más de una década que le habían dado el premio Turing y en la entrevista, como no podría ser de otra forma dado el tema, hablan de tipografía digital y de la creación de Metafont, un software que se sigue usando hoy y que continúa siendo una [no tan] pequeña maravilla.
![COMPUTER SCIENCE CONSIDERATIONS
CONDUCTED BY G. MICHAEL VOSE AND GREGG WILLIAMS
Donald Knuth speaks on his involvement with digital typography
Text processing as a computer science problem has consumed a major portion of the time and energy of Stanford professor Donald Knuth over the past eight years. Knuth authored and placed into the public domain a highly regarded typography system that he calls TeX {pronounced "tech"), along with a font creation language called METAFONT. \n conjunction with the completion of T^X, Knuth and Addison-Wesley are publishing a five-volume work entitled Computers and Typesetting. Volume I is The TeXbook, volume 2 is the source code for TeX, volume 3 is The METAFONT Book, volume 4 is the METAFONT source code, and volume 5 is Computer Modern Typefaces.
To discover what so intrigued Knuth about this subject. BYTE senior editors Gregg Williams and Mike Vose conducted the following interview with Professor Knuth at Addison-VJesley's offices in Reading, Massachusetts, on November II, 1985.
BYTE: Dr. Knuth. how did you become involved with digital typography and the publicdomain system known as Tj:X? Knuth: I got interested because I had written books and seen galley proofs, and suddenly computers were getting into the field of typesetting and the quality was going down.
Then I was working on a committee at Stanford planning an exam, and we got a hold of some drafts of Patrick Winston's book on artificial intelligence. We were looking at it to see if we should put it on the reading list for a comprehensive exam. It had just been brought in from Los Angeles where it had been done on a digital phototypesetter. This was the first time that I had ever seen digital type at high resolution. We had a cheap digital machine at Stanford that we thought of as a new toy. But never would I have associated it with printing a book that I'd be proud to own. Then I saw this type, and it looked as good as any I had ever seen done with metal. I knew that it was done just with zeroes and ones. I knew that it was bits. I could never, in my mind, ever, conceive of doing anything with lenses or with lead, metallurgy, and things like that. But zeroes and ones was different. I felt that I understood zeroes and ones as well as anybody! All it involved was getting the right zeroes and ones in place and I would have a machine that would do the books and solve all the quality problems. And, also, I could do it once and for all. I still had a few more volumes to write [of his seminal work. The Art of Computer Programming, a seven-volume series of which three volumes are finished] and](https://i0.wp.com/obm.corcoles.net/wp-content/uploads/2026/01/image-23.png?resize=749%2C1024&ssl=1)
Y, para hacer más énfasis en lo que decía de que procesado de texto no se refiere a los procesadores de texto (al menos, no a los que nos vienen más rápidamente a la cabeza), nos podemos dar un chapuzón en cómo estaba por aquel entonces el estado del arte de la interpretación del lenguaje natural:
(Como es costumbre de la casa, tanto Pollack como Waltz son no solo expertos, sino pioneros en la materia.)
Seguimos con el tema. Nos quejamos (con razón) de que artes y humanidades están excesivamente separadas en las cabezas de muchos, y de que esto es fuente de unos cuantos de nuestros problemas. En los ochenta ya era en gran parte así, no nos engañemos, pero de vez en cuando podíamos ver cosas como un artículo en una revista tecnológica dedicada al tema del procesado de… poesía.
No os perdáis, por favor, la discusión sobre cómo sacar la métrica de un poema automáticamente (en inglés, además, donde la cosa depende más de sílabas átonas y tónicas que en español):
![Machine Reading of Metric Verse
by Paul Holzer
A computer can definitively scan a line of poetry for its stress pattern principally in one of two ways: (I) an algorithm can deduce the syllabic structure and the stressed syllables from analysis of the letters that make up the word, or (2) the computer can look up every word in a dictionary database that holds the syllabification and accentuation of every word. The lookup method requires a large database, and the algorithmic approach is complex and requires a deep analysis of English phonetics and spelling.
One of the features of a poetry processor is that the poet-user can specify the meter of every line of a poem (see photo A). For example, the string .-/.-/.-/.-/.-/ represents iambic pentameter. Dots (.) indicate an unstressed syllable and dashes (-) represent a stressed one. The slash (/) indicates the end of a foot, the basic metric unit. The first line of Shakespeare's Sonnet 18
shall I comPARE thee TO a SUMmer's DAY?
is an example of a line of iambic pentameter. The stressed syllables are in uppercase.
After writing a poem, users might request a metric scan of the poem. I will describe here a method fordoing this that is not based on one of the two general solutions I mentioned in the first paragraph. Instead, the processor will break each word into its syllables and then redisplay each line, with each syllable in uppercase or lowercase according to the position of the dots and dashes in a user-specified metric form. So. were Shakespeare trying to compose trochaic pentameter, with the metric pattern -./-./-./-./-./. the processor would reply with
SHALL i COMpare THEE to A sumMER'S day?
He would read this to himself, trying to put the stress on the uppercase syllables. Noting the rhythmic clumsiness, he might rewrite his line as follows:
To a summer's day I shall compare thee
and the processor would respond:
TO a SUMmer's DAY i SHALL comPARE thee.
Sounds better!
The main task for the computer is to break each word into its syllables. The algorithm is based on a systematic application of what appear to be the general rules by which English words break into syllables. Of course, there are no fixed rules, as evidenced by the fact that different dictionaries give different syllabifications for the same word.
The following is a simple version of the algorithm:
1. Break the word up into a sequence of alternating vowel and consonant groupings. Thus microcomputer becomes micro computer. Wherever there is a vowel or group of contiguous vowels, there will be a syllable. We need only assign the neighboring consonants to the syllable on the right or to the syllable on the left.
2. If the first vowel group has a consonant group to its left, then assimilate this consonant group to the vowel group. This leads, in our example, to microcomputer.
3. If the final vowel group has a consonant group to its right, then assimilate this consonant group to the vowel group. We now get microcomput er.
4. For the remaining unassigned consonants, do the following:
. a. If the consonant stands alone, attach it to the following vowel. Thus we get mi cr ocompu ter.
b. If there are two consonants, split them. We get mic ro com pu ter.
c. If there are three consonants, then i. If there is a doubled consonant, split the pair; thus apply becomes a ppl y and finally ap ply.
ii. If there is no doubled consonant, but the first of the three consonants is n, r, or [, then split between the second and third consonants.
iii. In all other cases, split between the first and second consonants.
Before applying this algorithm, however, we must preprocess the initial string of letters in order to take into account certain peculiarities of English orthography:
1. Final e is silent (with certain exceptions); treat it as a special consonant. Thus compute becomes compu te, then compute, and finally compute.
2. Translate many two-letter sequences into special single consonants, e.g.. sh, th, gu, qu. and ck.
3. Identify common suffixes. For example, the algorithm applied to blameless would yield blameless and then bla me less. However, when less is removed as a suffix, then the e in blame to thinking of the program as something for me to use— the relational table of contents was so the user could access my work. The program was originally to have been just a floppy solution to my table-of-contents dilemma. But you don't get that involved in a software application without elaborating and generalizing. In that way software is very much like'
poetic forms. You use it for the sake of using it. It generates its own kind of trance. Poetry and programming, once you look at them in context were just made for each other.
Marriages like this one, made in heaven, often are so because they are marriages of convenience. One of the impediments to formal verse writing is the inconvenience of having to
make repeated book accesses for rhymes, just when the form has prompted some involvement. You stop and look and lose something. That's one reason people have tried to do without forms. But that's throwing out the baby with the bathwater. You don't stop measuring and sounding things out, and you don't abandon would be recognized as silent, yielding blame less.
4. Identify some prefixes. For example, if en is recognized as a prefix, then enact becomes en act, rather than e nact.
It seems to be impossible to come up with a reasonably small set of rules and preprocessing steps to guarantee correct syllabification of all words. Two examples will illustrate some of the inherent difficulties:
1. Compound words: The algorithm will not detect the silent e in snake within the compound word snakebite unless the fragment bite is recognized as a word or treated as a suffix. Avoiding the problem would require either extensive word or prefix table lookups.
2. Successive vowels in different syllables: In reach, the ea is a single vowel sound, and the algorithm would treat it correctly. In react, we pronounce the e and a separately and the correct syllabification is react. Were the algorithm modified to isolate re as a prefix, it would treat react correctly, but turn reach into re ach.
Where ambiguities can arise, the best approach is to formulate a rule that leads to the smallest number of cases requiring table lookups for resolution. The present algorithm is not perfect, but it produces a readable, if not dictionary-perfect, syllabified word 95 percent of the time.
I have provided a Pascal program that implements the syllabification algorithm and illustrates how The Poetry Processor "reads" a user's poem according to a user-specified metric scheme. Editor's note: The Microsoft Pascal source code and executable version are available from BYTEnet Listings, telephone (617) 861-9764. as SCANPOEM.PAS and SCANPOEM.EXE. The executable version requires any MS-DOS or PC-DOS machine] To run the program, prepare two files. TESTPOE must contain the lines of poetry. You can write TEST.POE as a text file with each line of the poem on a separate line. A second text file. TESTFRM. should have a line containing a string of dots (.) and dashes (-) indicating the accentual scheme that each line of poetry is supposed to follow. Slashes indicating the end of a foot are optional.
As an example, a Shakespearean sonnet (iambic pentameter) will have a TESTFRM file consisting of 14 lines of .-/.-/.-/.-/.-/. Each line in TESTFRM must end with an asterisk. After editing the TESTFRM and TESTPOE files, you can run the program by entering its name, SCANPOEM. The computer will "read" the poem, printing in uppercase the appropriately stressed syllables.
Note that the program is a prototype version of the algorithm. It will not handle text with capital letters, apostrophes, or punctuation, so be careful not to include these features in TEST.POE. When using this demonstration program, you will undoubtedly find that some words are not properly syllabified.](https://i0.wp.com/obm.corcoles.net/wp-content/uploads/2026/01/image-25.png?resize=840%2C430&ssl=1)
Pero el colmo del friquismo, en serio, es un artículo entero dedicado a la sesudísima (solo hago un poco de broma, aquí) cuestión de si vale la pena aprender a teclear en un teclado Dvorak (#TLDR, los autores opinan que sí, si te puedes permitir el lujo de escribir siempre en un teclado Dvorak). Que el primer firmante de la pieza sea profesor emérito… de física, dedicado a la astronomía forense, es solo la guinda del pastel.
¿Había dicho yo que volveríamos al tema Amiga a la que nos dieran una oportunidad? Sí, ¿verdad? Aquí, los orígenes británicos de AmigaDOS:
Y aún una página más con contenido Amiga, aunque aquí no sea el contenido lo que quiero destacar, sino el continente. Estamos en 1986, y el mundo comienza a conectarse digitalmente. Byte, de hecho, tiene su propio servicio online, BIX (el Byte Information Exchange), que se había puesto en marcha en junio (a seis dólares de la época la hora de conexión)… pero la audiencia era tan corta (dice la Wikipedia que en el 87 llegaron a 17,000 usuarios) que la revista le daba bombo al servicio destacando un «Best of BIX» en sus páginas. Igual sí hemos cambiado un poco, en estos cuarenta años…
![Best of BIX
AMIGA
Commodore's introduction of the Amiga has produced a flurry of activity among professional developers and personal computer users within the Amiga conference. The summary this month includes discussion on cables, monitors, printers, and software fixes. One of the hottest topics in the Amiga conference is on the subject of improving the performance of the Amiga by removing the 68000 and replacing it with a 68010 or 68020.
68010/68020 Upgrade
amiga/amiga68000 #22
An Amiga conference member asked if he could just drop a 68010 into the 68000 socket. This would give a 10 to 80 percent boost in performance! He had one, just sitting up to its bottom in black foam, on the shelf. But there were all these warnings about what would happen to his warranty if he opened the case.
amiga/amiga68000 #26, from rickross [Richard Ross, Eidetic Imaging]
M68010 works! A 68010 plugs directly into the Amiga and no problems were detected in the operation of the system software. Also, for everyone like me who has been trying to judge from the BYTE review photos, the microprocessor is socketed. The performance increase gained by the switch is not phenomenal, and no benchmarks are available, but it did run perceptibly faster. The M68020 has also been tried and seems to work as well.
amiga/amiga68000 #32
A BIX user provides the following:
The company that markets the 68020 piggyback board is Computer System Associates Inc., 7564 Trade St., San Diego, CA 92121, (619) 566-3911. The prices are:
Board only $ 575
Board plus 68020 975
Board plus 68020 and 68881 1480
For more information, contact Patricia Chouinard at the address above. I believe that 68000/68010 supervisor code that handles exceptions and certain other privileged functions will have to be modified. User code should work as is.
amiga/tech.talk #39
An Amiga owner describes his adventure in opening his computer and replacing the CPU:
You just got your Amiga and it's already the slow boy on the block, right? You can plug a 68010 into an Amiga (there goes my warranty) and it does go faster My Sieve benchmark is down to 5.8 seconds from 6.1.
Note: Your warranty will most likely be dead after you do this. Also, there is a lot of RFI shielding inside the Amiga. You get to undo a lot of screws, bend a couple of tabs, and pray a lot. If you aren't a tech type, don't even think about doing this yourself. The 68000 is socketed, but it is partially under the micro-disk drive, so you have to lift it from one end and kind of levitate out the other end (use of your CHI helps). Also, you only take out the screws in the deep wells on the bottom (five in all). Then there are four places where the top grabs the base at the four corners (there were already marks on mine from where it was put together, I guess). Once you have the top off there is a big surprise waiting for you... Another big surprise is that big RFI shield. Yes, it is a $#%+& to get off! There are screws on three sides and two tabs of metal to untwist. Once the shielding is out of the way, your first sight is of the WCS [writable control store] daughterboard. The custom chips and two parallel I/O chips are made with MOS technology.
The CPU is made by Motorola. The main board looks pretty much like the BYTE review photos. The boot ROMs are 27256s! This gives a 32K-byte by 16- bit boot ROM! What are you guys hiding in there? I could put a BASIC interpreter in that much space!
If you attempt to change your CPU, don't blame me if you muff it! If you don't know about how to make yourself static-free, you could really buy yourself some trouble of the worst kind.
Compatibility: I've run all of the Workbench demos. Everything seems fine, but I'm not making any promises. . .
amiga/tech.talk #41
The adventurous Amiga owner says that yes, his Amiga boots up, squeaks and everything! All the software he has runs and works great. The only potential problem at this point is how many times the MOVE SR.dest op code is used. This is the only active op-code difference. There is a whole host of new goodies, though, some that make a . desire for an MC68881 easier to satisfy.
amiga/tech.talk #43: a comment to 39
Another BIX subscriber replied that the upgrade produced only a 5 percent increase in throughput. Perhaps fortunate, because the descriptions of the hardware here have indicated that bus bandwidth consumption by the 68000 is low enough to allow other custom DMA chips to steal enough cycles to get their work done. It would appear that inserting a 68020 in the socket would require faster bimmers, etc.
amiga/tech.talk #44: a comment to 43
Wouldn't think just putting in a 68020 would affect DMA. Same clock speed. Or does the '20 do something different cycle-wise?
amiga/tech.talk #45: a comment to 44
The author of message 43 replied that the 68020 at the same clock speed will finish an instruction or series of instructions internal to the CPU in less time and start requesting the bus for some ROM or RAM access. He assumed that the DMA chips hold a higher bus priority, so the result will be that the 68020 will often be sitting there in idle awaiting the BUSACK signal. Waste of a 68020. Perhaps that explains why there is only a 5 percent 68010 edge over the 68000.
amiga/tech.talk #46: a comment to 45
Somebody said that the 68000 only uses every other clock cycle (for memory access, that is). The DMA hardware is fast enough to do four accesses during every clock cycle. Most of the DMA accesses the bus during periods when the 68000 doesn't. If the 68020 doesn't have these quiet periods then there could be problems.
amiga/tech.talk #47: a comment to 46
Actually, there is a counterargument to that, which is that the 68020, but not the 68010, has an instruction-only cache, which would mean...](https://i0.wp.com/obm.corcoles.net/wp-content/uploads/2026/01/image-28.png?resize=774%2C1024&ssl=1)
Antes de cerrar la sección, quiero aprovechar para recoger el obituario de Robert Tinney en Ars Technica. ¿Quién es Robert Tinney? El ilustrador de muchas de las portadas de los números de Byte que hemos recogido por aquí, que falleció este primero de febrero. Que su obituario aparezca en Ars da una idea tanto de la relevancia de la revista como del impacto visual del trabajo de Tinney en muchísima gente. Curiosamente, estamos muy cerca de llegar a los números en que la revista dejó de emplear a Tinney para pasar a usar fotos en sus portadas, como podéis comprobar en los archivos de la revista Byte en archive.org, que también podéis usar, si queréis, para avanzaros y comprobar de qué va el número «del mes que viene». Añado que Tinney tenía una tienda, todavía activa (y espero que lo siga estando mucho tiempo), y que ahora mismo estoy peleando muy fuerte conmigo mismo para no comprarme pósters del número de artes digitales de 1982, la de abril del 85, o la de «claves de la educación» de, nada más y nada menos que julio de 1980.
El primero de los episodios se dedica a operar en bolsa por ordenador, algo novedoso en la época. No me ha resultado especialmente interesante, más allá de los cacharritos para recibir información financiera vía radio FM, tanto en forma de cacharrito independiente como de accesorio para tu PC.
El segundo programa del mes va de «software psicológico», desde software para ayudar con determinadas terapias (con la sofisticación de la época, más cercana al programita con el que se juega para renovar el carnet de conducir) a tests de tipos diversos, con sus, inevitablemente, «módulos de inteligencia artificial»… y las mismas preocupaciones y las mismas salidas por la tangente que nos suenan tanto hoy.
(Y en los breves, noticias de la crisis de Commodore, que le debía doscientos millones de dólares a los bancos. La compañía no acabaría muriendo hasta el 94, pero ya comenzaba a oler a chamusquina la cosa.)
El tercer programa del mes se dedicaba al software para astronomía, tanto profesional como amateur (en este último caso, bastante reconocible para cualquiera que haya usado una app de astronomía únicamente… pero cuatro órdenes de magnitud menos potente e interfaces jurásicas). La discusión sobre astronomía «profesional»… lo de siempre: gente alucinando con lo que había avanzado la tecnología en el campo… que ahora nos parece casi de juguete.
(Y en los breves, la muerte de la mítica Osborne… cincuenta y tres millones de dólares de pérdidas de Commodore, por si los doscientos millones de deuda fuesen poca cosa… y la compra de Pixar por Steve Jobs por «varios millones de dólares».)
El 3×22, dedicado al color, lamentablemente, parece que está desaparecido. Como de costumbre, podéis chafardear lo que se viene en marzo tanto en la lista de episodios de la Wikipedia como en la playlist a la que pertenecen los vídeos de YouTube que tenéis aquí arriba.
Y con esto cerramos el mes. Dentro de unas semanas, más.