I don't think that's coincidence.

August 31, 2023 2:08 PM

"I propose it has consequences to democracy how available to the public are their laws, their policies, their judicial decisions, their holy texts, and even their academic papers about literature. The W3C spent decades insisting in utter folly that they were right in their conception of how ordered lists work, in the face of the whole world, and in doing so they frustrated the transparency on which democracy rests."

Over on Dreamwidth, Siderea gives us a blistering dive into the profound social consequences that can arise from one misguided technical decision: the HTML ordered list.

posted by mhoye (51 comments total)

correct.
posted by bombastic lowercase pronouncements at 2:23 PM on August 31, 2023 [8 favorites]

Interesting idea, and I absolutely agree with the writer from an html perspective (not sure the downfall of democracy can be placed here; I'd see it more as a symptom of the original problem that a lot of people think the law and politics have nothing to do with them, they are "not political" etc. and probably the W3C folks' completely ignoring legal documents as a use case was more a symptom of that, but I don't know).
posted by joannemerriam at 2:28 PM on August 31, 2023 [3 favorites]

In the following essay I will elucidate the de facto link between the deprecation of the HTML <blink> tag and the rise of crypto-fascism as in accordance with the ascendancy of Elon Musk…
posted by blue_beetle at 2:28 PM on August 31, 2023 [26 favorites]

In the following essay I will draw a direct link from the deprecation of the <center> tag in HTML5 to the rise of political polarization and a widespread perception that "the center cannot hold"...
posted by allegedly at 2:34 PM on August 31, 2023 [18 favorites]

folks you know how much i hate to make bombastic lowercase pronouncements, but this time i gotta:

the article is correct. and must be engaged with in a way that respects its correctness.
posted by bombastic lowercase pronouncements at 2:51 PM on August 31, 2023 [13 favorites]

The author wants a description list, with the fixed labels as the item descriptions. But also, in ordered lists this is a matter of setting/changing counters, which can be done fairly simply for the starting number in an ordered list in html. In legal document applications, you already have to know enough to set the type and style of the enumeration items (Roman numerals, regular numbers, letters, whether they’re in parentheses or have a period after them, etc.), and it’s not any more difficult. Now, skipping a number is a different thing - LaTeX deals with that better by making the counter used accessible to the user writing a document. There’s probably a way to do it in html as well, but that’s probably more advanced? But in my experience with legal documents, if an item that was originally included in a numbered list gets struck from the law or document, the number and a note with the date repealed is left in the legal document. Having a new item added mid-list would require a little more work, since the legal practice seems to be to add something like “7, 8, 9” -> “7, 8, 8A, 9”, so the new list item doesn’t follow the same naming scheme. Implementing with a described list solves this inconvenience, however.
posted by eviemath at 2:53 PM on August 31, 2023 [3 favorites]

That's just what the radical centrists want us to think.
posted by jpziller at 2:53 PM on August 31, 2023 [5 favorites]

In short:

1. Legal documents are better thought of as described lists, not ordered/enumerated lists, but
2. You should not be removing items in a list from a legal document without leaving an artifact noting the deletion.

posted by eviemath at 2:56 PM on August 31, 2023 [6 favorites]

o boy.

Yes... the author makes a case about their issue with <ol> for law or political docs.

I would counter that for a legal document, the whole thing - numbering, block-quotes, indents - IS content, so an online legal document should NOT make any use of the ordered lists HTML tag. Legal docs shouldn't use auto anything, if it dynamically generates content.

The idea of a definition-like tag makes sense. I would also propose a tag named pre-ordered list:
<pol>
    <poli order=[0,1,2, etc] label="[1,2,3 a,b,c i,ii,iii ...whatever]" > content </poli>
    ... and so on

(label attribute is mandatory; order is optional)

... and the item label should come along with any copy-paste operation.

Maybe legal docs require their own markup tag subset.
posted by Artful Codger at 2:59 PM on August 31, 2023 [6 favorites]

Not that I’ll defend html or css, but because a vaguely named feature doesn’t do what you want it to, it doesn’t mean it’s wrong or that it’ll end civilization. There’s a reason naming things is the hardest thing to do when programming.

HTML is not a DSL to write laws.
posted by WaterAndPixels at 3:03 PM on August 31, 2023 [9 favorites]

Artful Codger, it’s easier to update when a new item is added mid-list if the descriptors are in the same place as the content items rather than all up front and then you have to count things out without having the compiled document in front of you. The dl list type already exists, meanwhile, and works exactly as needed.
posted by eviemath at 3:05 PM on August 31, 2023 [3 favorites]

the real correct solution is separation of content vs appearance vs a third thing, maybe "functionality"

You mean Javascript?

Which of course hits on a bigger, though not undiscussed, issue with the Web - the tendency of the “functionality” layer to subsume appearance and content, until there are no documents, only programs.
posted by atoxyl at 3:09 PM on August 31, 2023 [4 favorites]

Mmm, I imagine she meant functionality as in accessibility, not so much in the programming sense?
posted by eviemath at 3:17 PM on August 31, 2023

I have to say I sympathize pretty strongly with the author's premise. I spent what seemed like an hour (but was probably only 15-20 minutes, which was still too long) trying to copy and paste a numbered list from Pages into a text box on a webpage. No matter what I did the numbers wouldn't show up. I finally had to do something like save as RTF, paste into Notes, then copy and paste again into the form I was filling out. I'm sure there is an easier way, but by then I was just trying random things to see if something would work. Why a word processer would work like that is beyond me.
posted by TedW at 3:18 PM on August 31, 2023 [3 favorites]

it’s easier to update when a new item is added mid-list if the descriptors are in the same place as the content items rather than all up front and then you have to count things out without having the compiled document in front of you.

Sorry for not being clearer with my hypothetical. Expansion:

<pol>
    <poli order=0 label="i" > content x </poli>
    <poli order=1 label="ii" > content y </poli>
    <poli order=2 label="iii" > content z </poli>
    ...

</pol>

So yes I had anticipated that descriptor/label was grouped with each item of content.
posted by Artful Codger at 3:20 PM on August 31, 2023 [1 favorite]

TedW, that’s perhaps a slightly different issue, one that “copy text and formatting” instead of just “copy text” was designed to solve.
posted by eviemath at 3:21 PM on August 31, 2023 [1 favorite]

Artful Codger, replace “pol” with “dl” and “poli” with “di” and you have the already-existing described list type in html. (See the link in my first comment.)
posted by eviemath at 3:23 PM on August 31, 2023

Sure, dl would work... but since such legal docs often have definition sections as well, it could be seen as an awkward overload to also use dl for a "legal" list to overcome the article's complaint, and different styling for the separate uses would become mandatory, making it harder to have a minimal "legal" default style that renders sensibly.

There, hair split.
posted by Artful Codger at 3:39 PM on August 31, 2023 [1 favorite]

Mmm, I imagine she meant functionality as in accessibility, not so much in the programming sense?

I meant it in the “doing stuff” sense, in the sense of usefulness beyond simply presenting information. I don’t really know how the author meant it, I was just riffing.
posted by atoxyl at 3:42 PM on August 31, 2023

Now you’re just talking about multilevel lists though.
posted by eviemath at 3:44 PM on August 31, 2023

The definitions section in most legal documents I’ve seen is section 1. Within that, I suppose you might label the items with the terms being defined, but I’ve mostly seen each item numbered, and then the term being defined is in bold within the text. So, not different from any other section of the legal document.
posted by eviemath at 3:47 PM on August 31, 2023

Now, many non-legal documents have a glossary, which is the use case most people would think of for this type of list, where the descriptor is the word being defined rather than a number or letter. Not functionally or format-wise any different.

Now, unlike html, Word doesn’t seem to have a described list type, as far as I can find. Lists in Word are a bit of a nightmare, and I would be fully on board had the author been advocating for never using Word for legal documents. Or outlines of any sort where dynamic renumbering would be a problem. Or when you have more than one numbered list in a document. Let alone multi-level lists.
posted by eviemath at 3:53 PM on August 31, 2023

TedW, that’s perhaps a slightly different issue, one that “copy text and formatting” instead of just “copy text” was designed to solve.

As far as I can tell, "copy text and formatting" does not exist in Pages for iPad, at least not in any easily found location. But yes, a different issue, although still part of the problem of functions that work the way the programmers think they should work rather than how the users expect them to work.
posted by TedW at 3:55 PM on August 31, 2023 [1 favorite]

Agreed in reference to copy-paste functionality. I still think the author misunderstands the type of lists that appear in legal documents, with that being a central issue in the fpp link rather than programmer versus layperson misunderstandings, however. But substitute programmer/lawyer as the person with specialized domain knowledge and I can see the similarity in the two issues.
posted by eviemath at 3:59 PM on August 31, 2023

a few thoughts:

1. once again, this article is correct
2. let’s step back a bit from the example of the inadequacy of html to the task of defining legal text to the broader problem that
3. we have here a textbook case of engineer’s disease: the `<ol>` tag correctly describes the structure of the documents the designers are most familiar with, but not others, and the designers carefully avoid acknowledging the non-universality of html as a tool for describing content by implicitly or explicitly asserting to themselves and others that all content is a subset of the types of content they, the designers, are most familiar with.
4. even though that’s not true
5. and furthermore the really troubling thing, the really annoying wrong thing that the author of the article is using the case of html’s inability to correctly mark up legal text to illustrate
6. and let me put in another numbered item just to stress how important this is
7. is that the idea of content and presentation as separable is itself bunk, on the face of it bunk, complete gibberish, simply not true in any meaningful sense
8. and great swathes of the Internet’s infrastructure is predicated on that laughably wrong claim

as such

1. html’s wrongness in the case of legal text is directly downstream from that wider, more significant error
2. and no amount of downstream well-actuallying about elaborate and onerous workarounds for html’s inability to correctly describe legal content will solve the more significant upstream problem, nor stop that more significant upstream problem from continuing to spawn a whole congeries of major downstream problems.

a critique of this article (which is, I cannot stress enough, correct) can’t start with an attempt to critique the example of `<ol>` and the law, because this example is not what the piece is about but is instead just a fascinatingly pyrotechnical illustration of what the piece is about, i. e. the stone cold fact that presentation and content are fundamentally inseparable, that they cannot be cleaved apart from one another no matter how many spec writers and implementers across lo these last three decades have convinced themselves otherwise.

what makes the legal example so good, however, is

1. it’s a real attention-getter, a+ solid rhetoric p.s. solid rhetoric is a good thing
2. and also it nicely demonstrates that a whole lot of engineers and spec writers have managed to hold onto the delusion that content and presentation are separable only by pretending that the presentation norms of their own genre are universal, so universal that they have transcended their status as mere norms and have instead become truths about the nature of textuality itself

it is precisely as if i had somehow deceived the html spec writers of the world into thinking that the way that i personally write metafilter comments is the standard way that content works, and that any problems treating my comments as the universal standard introduces for people who want to use capital letters are not html’s concern. after all, if someone wants to have capital letters they can go through and manually mark up every letter they want capitalized — a time-consuming and error-prone process — and this is all totally 100% fine because, don’t you know, capitalized letters are a presentation thing not a content thing.
posted by bombastic lowercase pronouncements at 4:50 PM on August 31, 2023 [18 favorites]

Nobody show them Markdown's ordered lists.
posted by ChurchHatesTucker at 4:56 PM on August 31, 2023 [3 favorites]

I'm sympathetic with this point of view, but: PEOPLE. We had this whole, like, thing going on for a while, where we were encouraging everyone-- Lawyers! Doctors! Educators!-- to make easily parsable domain specific data. It was called the semantic web. Did it suck? Absolutely. Did it work? Kinda! Could it have been improved and evolved into something that would have let us have a better, content-centered web? Sure, if we'd spent, like, one tenth of one percent of the amount of effort we spent on getting scroll-down animations and replacing HTML tables with, uh, whatever the hell you use now. We didn't, though, because that's not what the people who pay for websites wanted. They wanted print advertising but online, and that's what they got.

The W3C tried to get out of the presentation straitjacket of HTML so many times. No one wanted it. The article can be as correct as the day is long, but it's not like W3C wasn't trying.
posted by phooky at 4:57 PM on August 31, 2023 [9 favorites]

uh the people who decided to spend all that time and effort on getting scrolldown animations and replacing html tables with whatever and designing newer and more efficient ways for you to tell servers that you want dark mode, the people who did that are the exact same fuckin’ people who get to send delegates to W3C meetings. we are in this mess because standards organizations as they exist now are not a tool for making the web more generally open and useful for everyone, but are instead a tool for mutually beneficial collaboration between large browser-producing companies, “mutually beneficial” in that sense that adam smith was getting at when he said that every conversation between businesspeople in the same trade eventually becomes a conspiracy against the public.
posted by bombastic lowercase pronouncements at 5:12 PM on August 31, 2023 [9 favorites]

My point is not that the W3C were the right people doing the right things; my point is that the W3C also created all this other stuff to try to address these problems, and the protocols and standards they produced were ignored. (The legal profession in particular did not give a quarter of a fuck, because they have been trapped in an unhealthy codependent relationship with Microsoft Word for a third of a century.)

oh, so it's bombastic italic lowercase pronouncements now, is it
posted by phooky at 5:23 PM on August 31, 2023 [2 favorites]

(I should also point out that I am bitter because I actually taught myself to use XSLT)
posted by phooky at 5:27 PM on August 31, 2023 [12 favorites]

To this day, law, policy, and similar documents are published in PDF format. Because it is relatively easy and straightforward to typeset nested ordered lists in a word processor.

I’m not quite sure I understand the argument. This must be a solved problem, as in Canada laws, regulations and policy are published in html, xml, and pdf because accessibility is mandated. E.g., the Access to Information Act.
posted by fimbulvetr at 5:51 PM on August 31, 2023 [3 favorites]

I got as far as this before LOLing:

This is self-evidently insane.

posted by Flunkie at 6:01 PM on August 31, 2023

There are three types of lists, not two:

*

unordered lists

2.

ordered lists

(c)

description lists aka labelled lists

Legal documents are not ordered lists. The numbers or letters that label sections are not ordinals. They are labels. Legal documents are described lists aka labeled lists. Trying to use ordered lists to format legal documents doesn’t work as well because it is the wrong tool because legal documents are not ordered lists, not because the tool doesn’t work for ordered lists. HTML has the correct tool for formatting legal documents: description lists. See the link in my first comment for details.


The main example in the fpp link is simply wrong. It is not html’s fault that the author is unaware of the existence of description lists in html and/or doesn’t understand the difference between an ordered list and a description/labeled list where the labels often happen to be numbers or letters.

Furthermore, the thing that the author points out as a problem with trying to implement legal documents as ordered lists instead of labeled lists is not the actual problem. One should never have a dynamic number change due to deleting a section or clause in a legal document: when a section or clause is repealed or deleted, the label for that section or clause should be retained with a note on the date and source of the repeal or deletion.

The problem case, as I mentioned previously, is when a new section or clause needs to be added mid-list. Then the label for the new section or clause will not precisely match the previous labelling scheme, vis: a new section or clause added between section/clause 8 and section/clause 9 will be labeled eg. 8A. This highlights the fact that legal documents are labeled lists, not ordered lists.


None of this is to say that html is perfect. But displaying legal documents in html is, as fimbulvetr notes, a solved task easily accomplished with the correct html tags: those for described/labeled lists.
posted by eviemath at 6:58 PM on August 31, 2023 [11 favorites]

Also, my main critique is that the author has misunderstood the structure of legal documents, not that they are unaware of the proper html tag for this structure. That is a fundamental flaw in the argument the author wants to make about inseparability of content and format/structure. Maybe that argument could be made, but they are not making it with their incorrect example.

Side note: for a clearer and more nuanced take on the general interrelationships between content and format, it’s much more elucidating to read explanations for and debates around more detailed markup languages such as LaTeX. This is an area that a lot of people from a variety of backgrounds across both the sciences and the humanities have put a lot of thought into, in dialog with each other across disciplines even. It’s not something where your uninformed hot take is going to cause major harm in the world, but maybe turn down the dial on the bombast a little bit, nonetheless?
posted by eviemath at 7:12 PM on August 31, 2023 [4 favorites]

Do the entire thing in DIVs and style them all. Need a list? DIVs. Need a table? DIVs. Graphics? No need for SVGs just CSS up a bunch of DIVs.

I call it divML.
posted by Artw at 7:40 PM on August 31, 2023 [4 favorites]

/shudders/

I’ve done that in certain online homework system problem coding environments. It really illustrates the downsides of treating everything as content.
posted by eviemath at 8:00 PM on August 31, 2023

Side note, “ordinals” (as the author of the fpp link mentions) are called that because they come from an ordering.

• The more formal explanation from mathematical logic
• Ordinal numerals

If your list item labels are ordinals, they are definitionally indicating order in the list relative to the other list items - which means that if you remove or add or move around list items, they will now be in a different position, so will now be labeled by a different ordinal reflecting their new position in the ordering. That is, your numbers used as labels are ordinals if and only if they are dynamically renumbered as the list changes.

If that would cause confusion for external references, then you are using numerals as labels, not as ordinals. That’s how you can tell the difference between a labeled list with numerals as labels versus an ordered list with numerals as ordinals.
posted by eviemath at 8:09 PM on August 31, 2023 [8 favorites]

Best of both worlds is the automatically numbered list where each item has a label that isn't part of the display content. Refer back to it by label, and the same process that numbers the list can convert that reference to whatever number it ends up being. That's what I'm used to in LaTeX. If an ordered list in html doesn't support that, maybe it should, but there's nothing inherently wrong with automatic numbering.

The numbers not carrying over when you copy the text is pretty obnoxious though. I understand how they arrived at this, but it seems like the wrong decision.
posted by vibratory manner of working at 8:50 PM on August 31, 2023

Wow.

I'm not a webdev: I do compiled code only. I'm ... kinda shocked, really:

>content vs appearance vs a third thing, maybe "functionality".

and

>content/appearance paradigm

?!?!?

This is ... solved, at least in the mobile domain (and also in the backend domain, afaik!). And the nomenclature is ... disingenuous at least and plain incorrect at best.

We're talking underlying 'data' vs (what is now called/incorporated/transformed as/into the underlying data in a) 'UseCase'. Functionality would be what transforms the data to the display-data.

The nebulous 'data' comes from the data-lake/ocean/backing database. The intermediary programming transforms it into the datastructure which is used to (and is reduced to and limited to the least amount of data needed) to display that data.

Backend API call produces ACDEG from backing data ABCDEFGH. Reciever (be it a webpage or a mobile app) gets ACDEG and reduces it to the bare minumum it needs for display, say, ADG (or even ADGI, with I being a computed value [although I should be produced from the backend]).

And then ADG gets displayed in whatever form wanted.

But OL seems to want to interpret this at runtime. Instead of, say, how you would interpret JSON.

It even seems (as I read the article) that doing ListA-from-JSON->Display-OL-HTML and then Display-OL-HTML->JSONLIST would result in ListA-from-JSON != JSONLIST !!!!!!!!!

WTF??!?!?

Well, TIL: a webpage's OL != a proper linked list or JSON structure AT ALL.

Which is kinda funny because I dealt with android PDF files in 2012 (which was a PITA) ... but it turns out webpages are just as/even more of a PITA :P
posted by MacD at 8:52 PM on August 31, 2023

I also enjoy the semantics of numbered lists in markdown, where you introduce each item by using a number, any number, doesn't matter what, and the software will automatically number them properly for you. So you end up writing like:

 1 item
 1 item2
 1 item3

posted by vibratory manner of working at 8:53 PM on August 31, 2023 [1 favorite]

I would, actually, kind of agree with this guy that OL is not spectacularly useful. On the other hand I don’t actually see it used all that much.
posted by Artw at 9:03 PM on August 31, 2023

We're talking underlying 'data' vs (what is now called/incorporated/transformed as/into the underlying data in a) 'UseCase'. Functionality would be what transforms the data to the display-data.

That is more and more how the web works, but it started as a hypertext document format with predefined layout elements.
posted by atoxyl at 10:54 PM on August 31, 2023

> None of this is to say that html is perfect. But displaying legal documents in html is, as fimbulvetr notes, a solved task easily accomplished with the correct html tags: those for described/labeled lists.

nevertheless, i stand by my unhinged rant!

this has been your bombastic lowercase pronouncement for the day.
posted by bombastic lowercase pronouncements at 10:56 PM on August 31, 2023 [2 favorites]

eviemath: Legal documents are not ordered lists. The numbers or letters that label sections are not ordinals. They are labels. Legal documents are described lists aka labeled lists.

This is the heart of the matter. Moaning about the unsuitability of <ol> for legal docs is a shaky foundation for the ensuing HTML sux! conclusions. And you CAN successfully separate content from presentation; you just have to understand what is properly included in content: organization and identification of the parts of the content is still content - its structure, specifically - and necessary to it. Labeled lines are content; how they get rendered (indent size, font, colour, line spacing, etc) is presentation. Why is that hard?

Using <dl> works, but it's arguably a kludge. Someone's already given the reductio ad absurdum proof for this by correctly stating you can do anything with divs and no other tags. So, at least for legal docs there is a case to be made for a pre-ordered list tag, or a "clause" list tag - a tag type with some semantic connection to the content it contains and organizes.
posted by Artful Codger at 7:35 AM on September 1, 2023 [1 favorite]

HTML’s specific issues are a subset of the larger problem that computers have never been built to serve their users’ needs to store, access, quote, and create content. We laugh at the aesthetics of shitposting and QAnon, at their mishmash of copy/paste formatting and marked up screenshots and inability to cite sources. But this aesthetic is what emerges when you try to save content with the most readily available computer tools.

Even 45 years into the personal computer industry, we haven’t figured out the most basic functionality of an information appliance.
posted by Headfullofair at 7:47 AM on September 1, 2023 [1 favorite]

If you really can’t bear the thought of using anything other than OL for your list with noncontiguous highly formatted “numbers” that contain things like “1.1.1” you can always use CSS to hide the numbering and put in your own, making it semantically an ordered list and content-wise whatever the hell you want.

Most people just use UL for this.
posted by Artw at 8:44 AM on September 1, 2023

As someone who has been tangentially involved in normative standard-making, including at the W3C, I must sorrowfully say that it is all like this. Pages and pages of it. Some people really like it, some people are very good at it: but nothing can ever be truly resolved either to anyone's satisfaction, nor for the greatest good.

Onward, though!
posted by ntk at 1:36 PM on September 1, 2023 [2 favorites]

Obligatory XKCD re:Standards
posted by Artw at 2:39 PM on September 1, 2023

Someone's already given the reductio ad absurdum proof for this by correctly stating you can do anything with divs and no other tags.

*cough* yes… that is what that was, nobody would possibly do that for about a decade.
posted by Artw at 8:02 PM on September 1, 2023

(The legal profession in particular did not give a quarter of a fuck, because they have been trapped in an unhealthy codependent relationship with Microsoft Word for a third of a century.)

Only after they had WordPerfect pried out of their cold, dead, hands.
posted by Karmakaze at 5:47 AM on September 5, 2023 [3 favorites]

WordPerfect handled lists better than Word.
posted by eviemath at 12:57 PM on September 5, 2023 [2 favorites]