"Word’s intellectual model is effectively timeless: you paint the text with its attributes. WordPerfect’s is active and progressive: you change a setting, continue typing, and then change some other setting"
Umm... those are the same basic model.
WordPerfect's model is simply more primitive markup. The only difference is, unlike a structural document editor, the markup isn't containers, they're just control directives. In essence you "paint" a region of text with a directive by changing a setting, then changing it back.
Word fails for a very simple reason: they've shoehorned structural editing functions, with styles and sections and so forth, into an editor that traditionally worked like WordPerfect. The result is a complex mishmash of ad hoc and structural document markup, and you end up with confusing rules in order to reconcile the models.
Personally, I prefer LyX: powered by LaTeX, it's purely structural, while providing a rich editing environment with a visual approximation of the output, where the final document is typeset to the desired format. The result is simple and freeing: I concern myself with the content and the structure, let LyX manage the underlying markup, and let the typesetter decide how to lay it out.
> Word fails for a very simple reason: they've shoehorned structural editing functions, with styles and sections and so forth, into an editor that traditionally worked like WordPerfect
In that regard, Pages is (was? didn't really use the new one) really awesome. Although it allows for arbitrary text settings to be applied percharacter, its UI actively encourages use of styles thanks to the now defunct drawer and various UI nudges and hints. As a consequence you were actively defining structure of your text, and intuitively understood that appearance is only applied as a side effect. Combined with styles being assigned F1-F12 shortcut keys and you had a wonderful piece of software that allowed my then-student girlfriend to type lectures and have them structured and typeset properly on the fly. It really highlights how the whole style-thing management in Word is absolutely horrendous.
Consider the layout that he applied to the citation sentence, where Word waves a magic wand and leaves some of them italic and some not. The simplicity of WordPerfect and LyX does not do that.
The underlying problem is that Word tries to be smart and tries to guess what you are trying to achieve. He argues that most of the times it is wrong. But even if it were right some of the time, it's very annoying when wrong, very hard to correct, and indistinguishable from magic.
And I don't want to have some unpredictable tool run on the text that I painstakingly created.
His comparison though, and the similarity to how WordPerfect managed things, is precisely why I switched to LyX as well. It's not perfect, but it's a huge step better at getting out of my way and just letting me tell the computer what I want to do, and letting it figure it out from there. I get to "just write," and in the end I still get something that honestly looks more professional than the things I did in Word and LibreOffice, with a fraction of the work and far less fighting the model.
The author is writing for a non-technical audience so he uses terms like "active & progressive" to describe WordPerfect and "Platonism" to describe Microsoft Word.
For the HN crowd, I think it's clearer to explain the difference between WordPerfect vs MSWord as "stream oriented" vs "object containers". See links describing and contrasting the mental models: [1],[2]
In other words, WordPerfect formatting was more like HTML inserting <b> and </b> tags to toggle things like bold font on & off. That's why WordPerfect's "reveal codes" was prominently assigned to a function key.
MSWord formatting is more analogous to CSS where "styles" and "templates" are more heavily relied on. There is more conceptual separation between text and its formatting.
So yes, if you carry WordPefect's mental model into MS Word, it will seem like the software is fighting you.
> MSWord formatting is more analogous to CSS where "styles" and "templates" are more heavily relied on.
... and this model scales much better. Once you assign a style to 250 different headers and you want to change its font, is one change versus 250. This is never addressed by what is, at times, a very erudite UI critique, but unfortunately smells like the usual sour grapes of somebody who refuses to learn new tools.
>, but unfortunately smells like the usual sour grapes of somebody who refuses to learn new tools.
To be more charitable, I'd say it's just an issue of the MS Word's mental model not being taught explicity. If you read typical training guides for MS Word, it describes the use of "styles and templates" as "best practices" or "saving time". Instead, the tutorial books should use much stronger language: "You must master styles and templates or you will be perplexed as to why MS Word doesn't do what you want it to do."
In defense of the WordPerfect "stream" paradigm, it is more intuitive because you don't have that extra layer of complexity with "styles". As evidence, we see the pendulum swinging back to "simple markdown" that lets people just insert things like bullet points at the exact spot they need it. With MS Word or CSS, you'd (ideally) define a paragraph style or a CSS rule.
The 2 paradigms of markdown (streams) vs CSS (styles) will coexist together.
But he has learned the tool. In fact he betrays his real frustration with Word: it's not with the Platonic model itself, but the fact that it betrays the model and uses some hidden logic (the >50% rule) to decide whether or not to actually apply the formatting indicated by the style. I've used Word for many years and have occasionally been stumped by seemingly non-applying styles. Now I know why.
This hidden rule should be the most heavily discussed 'feature' of Word. MS Office docs should trumpet it as loudly as they can. (Or they could just fix it.)
As someone who has laid out entire 2-300 page books in Word and LibreOffice both, what you say is a fine idea in theory, but rarely ever works out in actual application at all with the way those tools implement it.
I think LyX/LaTeX presents a saner hybrid here: using styles, yes, but also declarative markup for where to use those styles in the stream. This is far more analogous to how CSS is actually used (ie. in concert with HTML and being auto-applied to standard HTML tags), than how it ever works out in the mainstream tools available.
Except it doesn't, really, scale well in practice. Even now, writing and maintaining large documents in Word is typically an exercise in frustration. Don't get me wrong, there are many things it does well. Books aren't one of them, and neither is technical content (e.g. math) or other complicate typesetting and structuring.
This may just be an implication of mixing typesetting and writing tasks.
Are there any platforms where writing and maintaing large documents is not an exercise in frustration? LaTeX is great but it's not what most people would consider "easy." From my observations, non-technical people don't get Markdown, reST, or other markup-based formatting at all.
Is maintenance of large documents just an "essentially" complex problem, to use Brooks' classification?
I think it probably is "essentially" complex, but that doesn't mean that all tools are equivalent. Sometimes the issue isn't making something "easy", it's making it possible. Sometimes with Word you end up working your project around the tool because anything else is impractical - but that's exactly backwards from how it should be done.
But those are typesetting concerns, not something authors are traditionally concerned with. The fundamental problem with word processors is that they merged these two separate disciplines. In traditional writing, the author prepares the manuscript which is generally not formatted at all other than basic chapter, section, and paragraph structure. A typesetter would then take that and make it look good in the final published version.
With self-publication more and more readily available, that barrier is getting thinner and thinner, and a lot more writers find themselves doing both like I did.
... and this model scales much better. Once you assign a style to 250 different headers and you want to change its font, is one change versus 250.
I've never worked on a Word document that large, but I've worked on Word documents large enough to demonstrate there's a big difference between this beautiful theory and the way Word works in practice.
I love the user experience description, here. This is an intelligent user of software describing their mental model of what is going on. It doesn't map cleanly to the discussions about software among software developers, because of course it doesn't, but it's also not just plain wrong or misguided.
This is the kind of feedback I would kill for on a product.
Agreed. When I wrote my master's thesis, I started by trying to use Google Docs. That lasted about two hours.... Then I tried LibreOffice. That lasted the rest of the first day. Finally I borrowed my wife's Macbook and switched to Word. It remained an exercise in frustration, even to do simple things like set page numbering in the pre-body content (intro, acknowledgements, ToC, etc) to Roman numerals vs Latin numbers in the body, or correctly setting up headers & footers, or matching footnote styling to the department's style guide. As the author states here, all this complex styling crap is a huge distraction from creating content, but because of how Word works, you can't just work in the old mode and leave the styling for the end. It has to be created inline or you're going to have a really bad time trying to fix it later.
> because of how Word works, you can't just work in the old mode and leave the styling for the end. It has to be created inline or you're going to have a really bad time trying to fix it later.
Why not? I don't doubt you, but I've never understood why it happens.
Perhaps this is a non-sequitur, but if you want something like an IDE for writing, Scrivener does a decent job. Its designed to help you marshal research (links, text, media) as well as to organize your writing. Plus it appears to be quite customizable. People have made document templates for screenwriting, research proposals, comicbook scripts, novel writing, etc. Its my preferred tool for longer prose.
As a longtime employee of big corporation I am still thinking why we are still using Word. Day to day documentation need doesn't require that big set of features. Even core expectations don't get satisfied and always need support from third party plugins. Other products like requirement analysis tools are built on top of Word which leads more buggy and bloat setup.
I think that's because it became an industrial standard that nobody choose to unfollow. I saw some people choose to attach word documents to email instead of pasting actual text to email body. Word will be on stage for a while unless we don't get rid of this mindset.
I've used Word before and I will if I'm required to, but otherwise I prefer to use a text editor and have gotten some strange looks from others for doing that/sending them a document as a plaintext file. I don't understand why people seem to always reach for it even when they're writing short unformatted pieces of text; it's rather overkill for that, and if you've ever seen things like RFCs, it is possible to create some very readably formatted plaintext documents too.
If I'm doing something that needs fancy formatting, then my tool of choice if TeX; but I still focus on the content first, and the formatting afterwards.
I was curious about this myself recently and a conversation with my wife managed to clarify it a bit for me. She's very technically proficient but is not a hacker. She also uses Word for almost everything, which drives me crazy, but I think I understand why now.
To her content and presentation are not separate entities. Font choice, heading size, margins and spacing, and colors are secondary to the text and the document structure, but they're still semantic elements and carry meaning above and beyond the structure they present. Presenting a sentence in red, for example, conveys a different meaning than making the same sentence bold, even if in theory they're simply alternate ways of conveying importance. She judges the extra expressiveness to be worth the extra complexity, even for informal documents and notes to herself.
This is why I loved Markdown when I first discovered it. Even without it, a lot of my plaintext files do things like the * for emphasis, or using dashes to underline headers and so forth. It's been done in software documentation for decades (a lot of the Markdown syntax isn't even original to it); it puzzles me it took so long for someone to sit down and just write something to use it explicitly.
What enforced my appreciation for separation of concerns was software instability. How would she react if Word crashed randomly every 30 minutes (nothing unusual in 95-era). I couldn't stand formatting/layout bugs wasting the time to type. So I naturally delayed formatting until my whole text was in and save safely. Only then I would create a copy with added styles.
Wait until you get the people using spreadsheets as word processors.
I knew someone who used the spreadsheet of Microsoft Works (incompatible with Word or Excel) for almost all work - fax cover sheets; letters to clients; posters and notes.
Hmmm... my experience differs: management/business types will use Excel to organize columnar data. They'd rather use Word for everything, but almost nobody thinks that Word's tables and formatting of text in tables, is sane, or even looks good. So: Word for text, almost always with no styles applied, and Excel for everything else. You want to give someone a subset of your columnar data? Take a screenshot and send a picture.
This does my head in at word. I write the database that the place runs on and the solution to anything to interact with the database is "another Excel to...". They don't seem to appreciate what a backward way of doing thing s this is, when you could have a proper web form, with half of the data filled automatically and proper data validation and error messages.
It makes a certain kind of sense. It's sort of like working on flexible graph paper. It's the same kind of logic that lead to table-based layouts in HTML.
As I discovered over various conversations with people, Excel support for Japanese blows away standard "plaintext" formats. There are a few standards for japanese encoding, which makes CSVs tricky (there's no standard way of specifying a codepage), but XLS and other formats explicitly state what encoding they use.
Yes, if the spreadsheet is actually spreadsheet, that's one thing. (And perhaps that works better than CSVs...)
What being circulated around is those spreadsheets with 1x1 grids used to create a document. No document structures; even TOC would be handwritten instead of auto generated from the heading.
Because if you're not into tools, having one tool that works okay for everything is better than having several, each of which works very well for some specific thing.
>Day to day documentation need doesn't require that big set of features. Even core expectations don't get satisfied and always need support from third party plugins.
Those two statements are somewhat contradictory, assuming "core expectations" are something you need for your "day to day documentation needs".
From my point of view, core expectations are editing document with multiple users, tracking changes, comparing and merging versions and you can list whatever you can do easily with the plain text file. These would be painful when you are doing it with Word.
The problem is, if you try too hard to mimic the real world example (typewriter in this case), you will end up with some constraints or weird bugs. For example, if the indentation were to be applied on per-line basis, what would happen if you later added some text in the previous line? would just the same amount of words be indented, or would the indented line not change at all, and the excess words pushed to a new line? This is exactly why some things in Word (and other word processors for that matter) are done the way they are done.
>This post is about word processors, but I got the idea for it from something W. H. Auden once said about political philosophers. In 1947, talking with his learned young secretary about an anthology he was compiling, The Portable Greek Reader, he mentioned Isocrates, a Greek orator whose simple-seeming ideas about relations between rich and poor cities were sane and practical. Naïve-sounding Isocrates had solved problems for which Plato’s grand theories had no answer. “Isocrates reminds me of John Dewey,” Auden said. “He’s a mediocrity who’s usually right whereas Plato is a man of genius who’s always wrong.” Only a genius could have devised Plato’s theory of the forms—the invisible, intangible “ideas” that give shape to every visible, tangible thing. But the theory of forms is always wrong when applied to political thinking, as every experiment in ideal, utopian politics has proved.
This reminds one of the "worse is better" notion.
Though, I would point that, just like with "worse is better", nothing of the short has been proved or "permanently settled".
It's a common cliche in american political thought to say that "every experiment in ideal, utopian politics" has failed etc, but the reality much more nuanced, and there's much ideology involved in the seemingly "non utopian" politics that's just transparent to their supporters.
It's too bad that most people are hopelessly computer illiterate, so that a separation between presentation and content like it is achieved with latex is out of reach for most. Countless hours wasted on adjusting margins, font sizes and filling out chain letter templates in Word.
I have written quite a lor of work in LaTeX and I am still perplexed by floating images and I can never tell where floats (images, tables) will be or should be, because it always jumps randomly from page to page.
My roommate at college had a thesis with lots of images and tables and he basically gave up on LaTeX because of that.
Also, installing fonts to MS Word is trivial. Installing fonts to LaTeX ... well... not so. You need to use XeLaTeX. Which breaks some other things. I tried to use bibtex with xelatex, and started to do some weird issues, until I found out that the texlive packages in debian/ubuntu repos are outdated, so I had to uninstall them and install it again from the website, and then something else broke.
Meanwhile MS Word just works as it is. And you don't need to learn the weird table syntax. And the weird floating logic. And the difference between TeX, LaTeX, XeLaTeX, BibTex, Texlive.
Regarding your comment about floats in LaTeX, of course you don't tell LaTeX where they go. They float. That's why they are called floats. If you don't want them to float, don't use the float environment. ( Or use the float package and the [H] option, as explained on https://tex.stackexchange.com/questions/8625/force-figure-pl... ).
Agreed. LaTeX is a mess. Its basic idea is fantastic. But its implementation is a brittle, entangled, inconsistent, old mess. Could use a fresh backwards-INcompatible reboot.
Separation between presentation and content is a brilliant concept, but every time I write a document in LaTeX I end up having to do all kinds of fiddly things to make things look right. If your paper doesn't conform to certain constraints you end up dealing with the presentation domain anyway. A very simple example is line breaks in equations: you don't know how much you can fit on a line until you render it, and LaTeX will run your equations off the page instead of wrapping them intelligently.
Don't get me wrong, LaTeX is still the best document processor I know of, and I use it all the time. I'm just saying that the "separation of domains" idea is a bit of a fantasy.
> If your paper doesn't conform to certain constraints you end up dealing with the presentation domain anyway.
After having read some of Edward Tufte's material, this struck me. I think the main point is: "make it legible" and (Tufte) "as informative as possible". If its not maximally aesthetically pleasing, then so be it.
EDIT: I apologize if this is trite, but I think that's really the whole philosophy of (La)TeX.
There is truth to this -- I do sometimes get caught up in irrelevant presentation optimizations. But sometimes the default output is clearly "wrong", or even nearly illegible. In the ideal world I have someone else (i.e. a publisher) to think about this stuff for me, but unfortunately that's not usually the case.
I'm pretty computer literate and I like the idea of separating presentation and content, but I've yet to find a really great implementation of it. HTML/CSS is a mess, and LaTeX is nice but very difficult to use.
Markdown seems to be the best, but isn't exactly wonderful. I wouldn't ever try to convince everybody I know to use Markdown for everything instead of Word!
This is the reason I use Org-mode. It provides simplicity of formatting like Markdown, but provides enough flexibility to export to format of my choice as needed. (Mainly, when I provide documents to others, I just generate PDF via LaTeX, but other formats including MS Word can be created fairly easily via ODT export.)
Latex is very easy to start with. Just learning a dozen commands will allow you to create many documents. What is a pain is customization, but remember that Latex was created so that only experienced designers would be required to customize a document style.
I like textual markup formats and separated styling. But my experience has been that Latex is "unbeatable" when it comes to wasting countless hours fighting your tools. It is great in some ways, but incredibly bad in others. (Random placing of figures, unbounded reruns, incompatible package hell, no stderr / stdout separation, unfixable warnings bloat, ...)
I must have been lucky then, I've written several 50+ page documents with latex that included figures, footnotes, equations and citations and rarely had to do more than select a documentstyle and occassionally align multiline equations. The only real pain point were figure environments, otherwise I never adjusted any of the preset values and ended up with more than acceptable documents.
I'm currently writing my (150 pages so far) doctoral thesis in LaTeX. The "flafter" package helps somewhat with the placement of floats, but generally if you put your float definitions where you refer to them it does the "most sensible" job. It is after all having to place floats at the top and bottom of the page only, or if necessary on a whole separate float-only page, while using no more than \topfraction of each page at the top, \bottomfraction of each page at the bottom, and ensuring there is at least \textfraction of each page left to text. The default values of these variables are a little too restrictive, so one of the first things I do is redefine them, which allows LaTeX to put the floats in more sensible positions.
It's ironic that you present Latex as a way to not waste hours and hours (or should I say days?) on getting formatting right. Of course if you have a simple and/or standard layout and want to keep all settings as they are provided by the environment, then fine. (I should probably add 'and are on a unixy system'). Otherwise - ...
Word has outlasted its original functional spec and is now fighting to retain relevancy and usefulness in a much different world. WYSIWYG used to be hugely valuable, making it easy to avoid wasteful reprints of work that didn't look right, or, worse, using an experiment/print/assess model to experimentally get markup right.
But that problem has been solved for years, it's no longer interesting, and it's not how people use editors today. Today word docs are not used for print media, often they are the primary form of media transmission or are used for collaboration, uses to which word is almost uniquely poorly suited, but because word is there it's used. Word is horrible for collaboration because it's focused on appearance, not content. Editing in word fails the principle of least surprise and it's generally non-deterministic (you can't see the markup "under" your cursor that will take effect when you start typing or make other changes). Too much effort is spent fighting the tool rather than actually communicating or collaborating. This problem can't be fixed while maintaining word's compatibility with it's previous functionality so it's unlikely ms will fix it any time soon.
what a curious piece of writing. it is really interesting to read a liberal arts guy critique software interfaces like a piece of literature. unfortunately, I can't really connect with the critique - too abstract. sometimes, you just need some good screenshots, and captions...
The first thing that comes to mind when I read something like this is, usually the issues are a training problem. He clearly knows how to use wordperfect - therefore it is easier. And if you are an author who never does typesetting, then most of MS word is a waste. If you are producing documents that routinely go directly to customers - MS word is a necessity. PDF the document and ship it out...
I remember using wordperfect, and I remember when word came along - it was like something from another planet (WYSIWYG). I'm sure you could improve word in many ways, but no one seems to be trying. Google Docs is as much of a clone as i could imagine.
I couldn't replicate the behavior he was talking about in Word for Mac 2011. Setting the style on a paragraph, regardless of how much text formatting you've done (I tried approx. 25%,50%, 75%, 90%+) doesn't touch any direct formatting.
If you want to clear the direct formatting, and revert to the paragraph style, you have to manually "Clear formatting" from that Paragraph - which is exactly the behavior I would expect.
The behavior he described is horrible, and I'm happy that it's not present in my version of Word. (Note, you can ask word to display the styles applied to paragraphs in the left side bar, and also ask it to highlight any place direct formatting has been applied. Very useful when doing style management).
Back when I used to have to deal with word processors a lot, my perception of Word (coming to it from WordPerfect) was that it felt like driving a very sophisticated car, except one that had no brakes - you could always ultimately get it to where you wanted to be, but there was no point along the journey where you actually felt like you were in control.
For no particularly good reason, I'm tickled that the author here has such different a metaphor for the same problem.
I still miss Word 5.1. It was simple and straightforward. Wrote a 100k word thesis using it in the mid 90s. On a Mac Classic with 9" greyscale screen (which I still have somewhere).
I love that the NYRB is getting a lot of play on Hacker News recently. It's really a top tier publication and reminds me a lot of the higher quality conversations on this site.
A more precise TL;DR would be "I wrote my document with some parts italicized and then applied margin formatting which removed my italics if less than 50% of the selection was italics, but kept it otherwise." Wordperfect didn't have this "feature," so it's better.
Based on this, I would suggest personally that once you've highlighted all the italics via Find, you could apply a style that's italic (such as "Subtle Emphasis" which looks like non-bold italic by default) which should in turn change the direct formatting to style-based formatting and, in theory, act like WordPerfect did from then on...
A bit strange to call Heading 1 a paragraph style when it's both a paragraph and character style. But then I tried it out with a pure paragraph style and was surprised that it removed direct formatting. Things I never even knew because I never used direct formatting if I could help it. Weird. And it's probably a good idea to continue eschewing direct formatting for that very reason.
I think perhaps you’re confusing WordPerfect with Word. WP, as a DOS tool, was very big on “show formatting” with a syntax much-like HTML. In that mode, you could actually delete formatting with the backspace key while leaving text intact.
A pompous piece by an author who appears to know little of Plato, and less of software design, but nevertheless enjoys concocting absurd analogies like these:
"Word’s 50-percent rule for applying styles is a descendent of the Demiurge, and just as much of a kludge."
We don't have enough information to know if the author knows much about Plato or not. What he says is not necessarily particularly deep, but it's not wrong either.
Generally, I liked the article.
What I don't like is the fact that the obvious advantages of the style-model are not mentioned. Most people don't grasp these advantages, which is why they have much more trouble than is necessary - even in Word - when working on a long text (a thesis, say).
Umm... those are the same basic model.
WordPerfect's model is simply more primitive markup. The only difference is, unlike a structural document editor, the markup isn't containers, they're just control directives. In essence you "paint" a region of text with a directive by changing a setting, then changing it back.
Word fails for a very simple reason: they've shoehorned structural editing functions, with styles and sections and so forth, into an editor that traditionally worked like WordPerfect. The result is a complex mishmash of ad hoc and structural document markup, and you end up with confusing rules in order to reconcile the models.
Personally, I prefer LyX: powered by LaTeX, it's purely structural, while providing a rich editing environment with a visual approximation of the output, where the final document is typeset to the desired format. The result is simple and freeing: I concern myself with the content and the structure, let LyX manage the underlying markup, and let the typesetter decide how to lay it out.