Has anyone besides Dr. Boli noticed that computers are not ideal for handling text? For all their power, they have to be pushed and prodded into creating text that is merely adequate typographically. This is mostly because they are doing work they were not designed to do. They did not evolve as typesetting machines, just as humans did not evolve as flying animals. We can fly now, but it takes a great deal of mechanical ingenuity to do it; we can typeset pages on a computer now, but it takes many adaptations on the part of the writer and designer that ought not to be necessary.

Some of these problems are in hardware, and the rest are in software. On the hardware side, we have a keyboard that is inadequate for proper and correct typeset work. On the software side, we have many inadequate adaptations to our inadequate keyboards.

But suppose we were putting together a real writer’s computer—a computer that could, naturally and easily, produce text properly formatted for printing. What characteristics would it have? Dr. Boli has thought about this question for months and scribbled notes as he thought. He invites his readers to do the same and, if they like, scribble their notes in the comments.

1. By default, the end-of-sentence space should be a different character from the between-words space. The desideratum is to put the designer in control of sentence spacing without having to change the text at all: it should be easy to decide that the space between sentences is the same as the space between words, or that it is a two-em space, or whatever the designer desires. We should note that not even the question of the serial comma produces as much anger and dogmatism as the question of whether a sentence should be followed by one or two spaces; current fashion among designers is to use a word space after a sentence, but the fashion is not universal, and arguably not good. Text should be easily portable from one style to the other.
   ☛A possible implementation: a double space could be automatically interpreted (unless the user disables this feature) as an end-of-sentence space. Since so many typists hit two spaces at the end of a sentence anyway, it would fit with already-established habits.
2. There needs to be a simple mechanism for proper catchwords—that is, a way of providing that the first word of each page should also appear at the bottom of the previous page, as was universally done until about 1800. This should be trivially easy for a computer to accomplish, but even TeX, the most outrageously bloated typesetting software in the computing world, cannot easily do it.
3. Why, in the twenty-first century of the Christian era, are there not separate keys for proper quotation marks and apostrophes? Why is there no key for an em dash—one of the most commonly used punctuation marks? Our keyboards are cluttered with characters like the backslash and the tilde (which is not a useful composing tilde for languages like Portuguese and Spanish, but just a sitting-there-naked tilde) that are of use only in dealing with the technical workings of computers, but we have to use clumsy substitutes or awkward key combinations for some of the most basic written symbols.
4. How would we put together the ideal writer’s keyboard? Here is a simple way. Look at a typefounder’s specimen book—any time from the early nineteenth century to the end of metal type will do. What characters make up a complete fount in that book? Those are the characters that need to be accessible directly from the keyboard, with no obscure key combinations needed to unlock them.
5. Perhaps some contrarian can make a Linotype keyboard with a USB connector, and we can all learn the ETAOIN SHRDLU keyboard layout. Dr. Boli has seen pictures mocked up of such a creation, but as far as he knows it has not been produced.
6. It should be easy to add any arbitrary text in the margin and link it with text in the main column, so that it is always beside that text. In other words, computers ought to be very good at setting marginal headings and notes, or running summaries in the margin. This can be done with TeX; but we said it should be easy, which eliminates that choice.
7. It should be very easy to deal with the long S by setting a single switch: “Long S on.” A settings panel could control the rules for different styles and historical periods. The same could be done with U and V, so that it would be possible to switch to Renaissance style, in which V is always initial and U is found in all other positions (“vnto,” “very,” “moue,” “blue”). The point of being able to flip a switch is to be able to flip it back again, so that the same text could be printed with the long S and without, or with Renaissance and modern use of U and V.
8. It should be easy to set a last-paragraph style that would taper to a triangular point, as a proper Renaissance printer would do. Settings would control number of lines to taper and narrowness of the point.
9. Position-dependent styles should work without thinking. We should not have to tell the software that the first paragraph is the first paragraph: it should know that it is because it follows a heading (or other marker that we choose), and then it should apply first-paragraph attributes to that paragraph. If that paragraph moves, or another is inserted before it, then the computer should know that it is no longer the first paragraph. The same is true for last paragraphs, of course.
10. The same letter should not always look exactly the same. There should be subtle random variations, as there are in metal type.
11. Now here is a wild desktop-publishing fantasy. Knowing that most text is displayed on screens or printed on smooth paper, we might nevertheless be able to simulate the appearance of ink on rag paper, adding a healthy irregularity to type. The parameters of the virtual texture of the paper should be adjustable.

These are some of the things Dr. Boli would find desirable in a true typesetting computer. What has he left out? What else do the writers out there miss that should be easy to implement on a computer, but unaccountably has not been implemented?