Paternal Formats

post by abramdemski · 2019-06-09T01:26:27.911Z · score: 54 (20 votes) · LW · GW · 30 comments

In How to Make a Complete Map of Every Thought You Think, Lion Kimbro introduces the idea of coercive vs uncoercive formats for information. [Re-named to "Paternal vs Navigable" here, at the suggestion of MakoYass [LW · GW].]

Paternal:

Navigable:

I initially wrote the above lists as two paragraphs. Turning them into bulleted lists made them feel less paternal. The rest of this post is paternal, in that I don't separate out different topics in an easily distinguishable way.

I'm not going to claim that "less paternal" is always better. Paternal formats are often more fun to read, for example. On this spectrum, HPMOR would be more paternal than the sequences. Furthermore, you might want to present information in a paternal format if getting information in the wrong order is likely to make a person actively misunderstand a concept rather than only fail to understand. (Mere prerequisite concepts are not really a good excuse -- a wiki-style format, where prerequisite concepts are obsessively hyperlinked, works fine.)

A 'sequence' is a somewhat paternal format -- a set of posts intended to be read in a particular order. However, sequences on LW1.0 were less paternal than sequences as they exist on LW2.0. LW1.0 sequences (mostly?) listed the posts in a sequence with summaries of each post,, while LW2.0 sequences only have a summary of the entire sequence at the top, w/o a short description of each post.

A paternal format can also be better for attention management. Navigable formats can create a tendency for attention to jump all over the place and not spend enough time on any one topic.

Paternal formats seem workable for small and medium-sized chunks of information, but become unwieldy for large amounts of information. Imagine if an encyclopedia were written to be read in a sequential order.

Navigable formats often take more effort to create. Drawing a good diagram tends to take more effort than writing the information in text (though, obviously, this depends on a lot of factors). Maybe the ideal should be to move information from more-paternal to less-paternal formats: it is good to initially create a lot of content in very linear formats which have to be read from start to finish, but it is also good to eventually move things to a wiki or something like that.

Another idea is that information should just be available in as broad a variety of formats as possible.

Paternal formats seem to be better for convincing people of a target position, whereas navigable formats seem better for presenting all of the information. This would make paternal formats a symmetric weapon. This isn't entirely true -- advertisements often seem to use information-delivery strategies which I would ordinarily characterize as less paternal (such as bulleted lists, text elements that don't need to be read in a particular order, images). Maybe that's because they need to work with a low attention span.

It is possible that rigorous argument and careful analysis fits better in linear, paternal formats. So, it isn't clear which formatting style is better for healthy epistemics.

Maybe communities which share information in a manner heavily bias toward paternal formats have a greater risk of being convinced by long and overly clever arguments, whereas communities which share information heavily biased toward navigable formats suffer from overly quick impressions formed via merely glancing at a million things.

But, let's not jump to the conclusion that a healthy balance is sufficient. The ideal would be to get the advantages of both.

30 comments

Comments sorted by top scores.

comment by MakoYass · 2019-06-09T05:41:48.897Z · score: 39 (14 votes) · LW · GW

I like the distinction but I wont support the choice of names for a second. You've said that you're not making a value judgement, but in action, you are imposing one. You can't use that word.

I think a better word for the "coercive" end of the continuum is "paternal". This has some of the negative connotation, but only in the right ways, sometimes paternalism is obnoxious, but we need to learn that there are also times when paternalism is appropriate.

(A lot of people think coercion is inherently bad in all situations, and they're right. A lot of people also think paternalism is inherenly bad in all situations, but they're wrong.)

A teacher should always be paternal. We should be as uncomfortable with that as we are uncomfortable with the idea of schools. A little bit uncomfortable. Not too much.

The other end of the continuum seems harder to name, but coining the word "unpaternal" can't do any harm. "Navigable" maybe.

I observe that the distinction seems to be mostly about how well the author understands what the reader needs to read. If the author understands much better than the reader, the civically appropriate format will always be paternalistic. The unit of media unpaternalism seems to be consumption-decisions per second.

comment by abramdemski · 2019-06-09T21:19:11.590Z · score: 9 (3 votes) · LW · GW

I like your suggestion well enough that I might edit the post. (I'll let it sit a bit to see whether I change my mind.)

comment by gwern · 2019-06-09T02:40:39.312Z · score: 28 (14 votes) · LW · GW

This name seems unnecessarily intrinsically prejudicial. Perhaps 'legible' vs 'illegible' would be better, or use McLuhan's 'hot' vs 'cool' mediums.

comment by cousin_it · 2019-06-09T20:59:23.262Z · score: 6 (3 votes) · LW · GW

Or linear vs open-world, as in video games.

comment by MakoYass · 2019-06-11T03:37:20.691Z · score: 7 (4 votes) · LW · GW

"open world" in games mostly refers to shams. In every instance I've seen, the choice is between "whatever forwards the plot" (no choice) and "something random" (false choice). The "something random" gives the player too little information about the choices for them to really be choices in the bayesian sense. You usually only get a vague outline of a distant object and when you arrive it's usually not what you were expecting. What information you do get is too shallow by the standards of any good game; there's no way to get really skilled at wielding it.

(And the reason genuine choice is rarely present is you end up needing to make multiple interleaved games, which is a huge design challenge that multiplies the points of failure, complicates marketing, and is very expensive if providing one experience for all will do the job just as well.)

This shamness could absolutely be transferred to educational documents. University felt this way to me; you can pay to stay on the path, or you can stray, and straying is generally fruitless, in part due to the efforts of the maintainers of the path, which unjustly reinforces the path.

comment by Said Achmiz (SaidAchmiz) · 2019-06-12T20:36:57.283Z · score: 3 (2 votes) · LW · GW

“open world” in games mostly refers to shams. In every instance I’ve seen, the choice is between “whatever forwards the plot” (no choice) and “something random” (false choice).

Escape Velocity

comment by Raemon · 2019-06-12T20:40:00.454Z · score: 3 (1 votes) · LW · GW

Agreed on Escape Velocity. Minecraft was my more recent go-to example for "actual open world". But I think I agree with the upthread point that open world games often aren't as open as you'd like.

comment by Said Achmiz (SaidAchmiz) · 2019-06-12T22:09:49.424Z · score: 8 (4 votes) · LW · GW

But I think I agree with the upthread point that open world games often aren’t as open as you’d like.

For sure. (An example of the phenomenon: S.T.A.L.K.E.R.: Shadow of Chernobyl, which has procedurally generated ‘sidequests’ which reward resources but have absolutely no relationship to the plot whatsoever, and also a single, almost perfectly linear set of plot missions. It’s a great game, and it’s certainly “open world” in the sense that you can (mostly) go wherever you like, whenever you like, but nevertheless it’s a plot railroad, period.)

But whenever someone makes a generalization and says that they’ve never seen counterexamples, and I know that there are counterexamples, then I think it’s critically important to (a) recall and make salient their existence (lest we mentally elide the generalization into a universalization), and (b) consider what features of the counterexamples allow them to be such—and what the pattern of those features tells us about the general trend.

comment by MakoYass · 2019-06-13T06:40:45.422Z · score: 1 (1 votes) · LW · GW

Looking at this.. I think I can definitely imagine a good open world game. It'd feel a little bit like a metroidvania- fun and engaging traversal, a world that you get to know, that encourages you to revisit old locations frequently- but not in any strict order, and more self-organised. I just haven't seen that yet.

It's worth noting that the phrase "open world" doesn't occur in the article, heheh.

comment by abramdemski · 2019-06-09T21:15:53.450Z · score: 6 (3 votes) · LW · GW

Maybe serial-access vs random-access, as in computer memory.

comment by cousin_it · 2019-06-10T06:02:29.385Z · score: 8 (2 votes) · LW · GW

How about sequential vs random-access?

comment by Said Achmiz (SaidAchmiz) · 2019-06-10T03:37:02.270Z · score: 2 (1 votes) · LW · GW

It seems to me that a graph-theoretic perspective would be fruitful to take, here…

comment by abramdemski · 2019-06-09T21:14:24.099Z · score: 2 (1 votes) · LW · GW

Yeah.... I thought about this problem while writing, but didn't think of an alternative I liked.

comment by gjm · 2019-06-09T02:42:43.206Z · score: 19 (7 votes) · LW · GW

I think this would be improved by a short paragraph (or less) explaining what "coercive" means in this context. (I made a guess after glancing at the bullet-lists, and then looked at the linked document; my guess was not correct.)

comment by Said Achmiz (SaidAchmiz) · 2019-06-09T05:25:02.365Z · score: 5 (3 votes) · LW · GW

I skimmed the linked document, and didn’t find a clear description of what these terms mean. Maybe I missed it. Could you (or someone) summarize? (Or, at least, give a page reference for where in the PDF the terms are defined?)

comment by gjm · 2019-06-09T14:19:31.203Z · score: 5 (3 votes) · LW · GW

If I am understanding right, something is "coercive" to the extent that it doesn't support a wide variety of ways of interacting with it in order to get whatever benefits it offers. An extreme example would be a document that starts out written in (say) English, but keeps introducing neologisms and new symbols and unorthodox grammatical constructions, so that it ends up written in a language and notation entirely of the author's devising that you can only make sense of by starting at the beginning and working through it in order.

The examples here are all about information sources, for which "interacting with" mostly means "reading", but I think the notion generalizes further. I suspect that the choice of a better term -- I think "coercive" is bad -- will be tangled up with the choice of how far (if at all) to generalize.

comment by abramdemski · 2019-06-09T21:12:14.984Z · score: 3 (2 votes) · LW · GW

I'm curious what your guess was.

comment by gjm · 2019-06-09T23:50:39.361Z · score: 4 (3 votes) · LW · GW

I expected "coercive" to mean something like "attempting to persuade and manipulate as well as inform".

Of course this interpretation couldn't survive a careful reading even of the bullet points, never mind the rest of your post, which is why more or less the next thing I did after making that guess was to look in the document you linked to to find out what it was actually meant to mean.

comment by Pattern · 2019-06-09T04:48:36.739Z · score: 4 (5 votes) · LW · GW

Great post by the way.

a wiki-style format, where prerequisite concepts are obsessively hyperlinked, works fine.
Uncoercive formats can create a tendency for attention to jump all over the place and not spend enough time on any one topic.

There's a concept here (possible what the OP meant by "coercive") which I might call "structure". One of the downsides to wikis (or parts of wikis) is, what if you wanted to read them? Is there a good order? Usually*, no. This limits their usefulness. Wikipedia is like someone who didn't know what a textbook was, reinvented it badly - pages everywhere, connected by string in random places where one page's title is mentioned on another page. The fact that you can only (maybe) find something if you know you're looking for it, and exactly what it's called limits one of the most useful aspects of books of knowledge - the chance to learn things you don't already know.

The Sequences are unusual in this regard (there's an order!) which is why I've read them. (One of the downsides of the medium was that I didn't initially realize that. If I read a physical copy of Lord of The Rings, I'd know I finished it.)

*I'm not aware of any counter-examples.


Minor errata:

in a manner heavily bias[ed] toward
comment by Said Achmiz (SaidAchmiz) · 2019-06-09T05:50:48.323Z · score: 4 (2 votes) · LW · GW

Oh, and by the way—

One of the downsides to wikis (or parts of wikis) is, what if you wanted to read them? Is there a good order? Usually*, no. … *I’m not aware of any counter-examples.

ReadTheSequences.com is, in fact, a wiki.

comment by Pattern · 2019-06-09T16:01:47.734Z · score: 3 (2 votes) · LW · GW
ReadTheSequences.com is, in fact, a wiki.

Which anyone can create an account on, edit, and make new posts/articles? The fact that it looks like a book, rather than a ghastly mess led me to believe otherwise.

Also, it's a tree, and it's obviously self-contained. It...flows. It has a homepage with an introduction and a table of contents which contains tables of contents which contain posts. You read it by reading, scrolling down, clicking (to go down a level), and when you've read that level you go back up and continue reading. (On the bottom level, posts/pages, you don't go any deeper.)

It's linear. There's a clear path through it.

comment by Said Achmiz (SaidAchmiz) · 2019-06-09T20:00:17.169Z · score: 3 (2 votes) · LW · GW

It’s linear. There’s a clear path through it.

It’s linear because I created views on the pages which present them in a linear order—which is my point. The pages are also hyperlinked together in a chaotic manner, as any other wiki is; and of course you can search it, which ditto.

You read it by reading, scrolling down, clicking (to go down a level), and when you’ve read that level you go back up and continue reading.

(You can also use the next-page / previous-page navigation buttons, which is even more linear.)

Which anyone can create an account on, edit, and make new posts/articles? The fact that it looks like a book, rather than a ghastly mess led me to believe otherwise.

The Sequence posts themselves are not publicly editable, for obvious reasons. The Talk pages (see ‘Talk’ link in top left corner) are publicly editable—with no account creation necessary. You can’t create new pages—but that’s only because I’ve got the permissions set that way. A change of configuration—a moment’s work—and that is enabled, too.

comment by Pattern · 2019-06-10T18:34:36.913Z · score: 1 (1 votes) · LW · GW
It’s linear because I created views on the pages which present them in a linear order—which is my point.
The Sequence posts themselves are not publicly editable, for obvious reasons.

Then I don't see a point of disagreement.

In regards to the OP's point, I'd say that not only are "books" a (linear)/simple structure, but physical books may act to coerce such a structure. It's not that I have something against other sorts of structures, just ones lacking clear paths. Are there books which suggest a reading order other than first page to last page? Yes, and and they tell you what it is.

The pages are also hyperlinked together in a chaotic manner, as any other wiki is; and of course you can search it, which ditto.

The level on which is this occurs is important. A hierarchy requires (clearly distinguished) levels above posts/articles to only reference lower levels (and call them as such).*

and of course you can search it, which ditto.

The linear/hierarchical structure of ReadTheSequences.com also allows for another kind of searching. If I read it in order, but forget where I am, I can binary search and see if I remember reading something. If I have (including the end), then I can eliminate it from my search along with everything before it. If I haven't, I can eliminate it from my search along with everything after it.

*This isn't undermined if these higher level pages note the page which contains them (while being explicit at a minimum that it's "a page which links here") I'd say something wikis miss is not having posts/articles contain a list of pages which link to them. (If not in the sense of not having the tech, then in not making it obvious: UI.)

comment by Said Achmiz (SaidAchmiz) · 2019-06-10T19:23:59.277Z · score: 2 (1 votes) · LW · GW

I’d say something wikis miss is not having posts/articles contain a list of pages which link to them. (If not in the sense of not having the tech, then in not making it obvious: UI.)

FYI, this is an artifact of the specific wiki software that you’re likely familiar with (namely, MediaWiki, on which Wikipedia is built). Other, better wiki platforms have easily accessible lists of backlinks (see “Backlinks” at the top right).

comment by Said Achmiz (SaidAchmiz) · 2019-06-09T05:49:07.467Z · score: 4 (2 votes) · LW · GW

One of the downsides to wikis (or parts of wikis) is, what if you wanted to read them? Is there a good order? Usually*, no.

You are conflating multiple different issues here, which must be examined separately or else any conclusions you reach will make no sense.

The first issue is that of content type. Many pages on Wikipedia, and on many other wikis, are simply not the kind of thing that you would “read”. They might be lists, or disambiguation pages, or category pages, or reference pages, or summaries of other pages, or meta pages, or media galleries, or blog-type updates, or “latest [whatever]” pages that communicate the status of something, or pages designed to be transcluded as components into other pages, or pages designed to be disassembled by transclusion and viewed elsewhere, or pages that implement some dynamic functionality, or data pages, or logs, or “Talk” pages, or profile pages, etc., etc., etc. Asking “what order should I read these pages in” is completely nonsensical when it comes to pages of any of these types.

The second issue is that of grouping. In what order should you read the following set of Wikipedia pages:

This, too, is a nonsense question. You could read them in any order you like, because they’re on completely disparate, unrelated topics. There just is not any kind of ordering that you can impose on them and say “there, now these three pages form a natural progression; certainly you shouldn’t read the last one first…”.

And this problem isn’t unique to Wikipedia. Even topical wikis often have pages that span a far wider range of subjects than would make sense to arrange into any kind of ordering.

And the third issue is that of views on information. Let us suppose that Wikipedia contains some subset of pages which, together, constitute the contents of a good cookbook (a bunch of recipe pages, some pages about cooking techniques, etc.). But these pages aren’t arranged in any kind of order…

… but what is stopping you from putting them in order? You could create a list of pages, which, if read in order, would make up the cookbook (or whatever). You could even make such a list… as a wiki page. Via transclusion, or such tools as wiki trails, you can assemble the list of pages into a single page, or into a structure which behaves just like an ebook would (or a web book, like ReadTheSequences.com or Butterick’s Practical Typography). And—importantly—each of those pages would still be a wiki page; it would still be browsable in the usual way, could be included in other “books”, etc.!

So, you see, to ask “in what order should I read this wiki” is simply to fundamentally misunderstand the nature of wikis—as well as their tremendous power…

comment by Pattern · 2019-06-09T16:19:48.722Z · score: 1 (1 votes) · LW · GW

Content type: I can read a book by skipping the introduction, table of contents, and go straight to the index/indices.

Grouping: This is a fair point, though most Wikis seem to have a narrower purpose than "encyclopedia". They're usually the encyclopedia of something (and these days Wikipedia supposedly has some limits, though they seem kind of vague).

But if someone separated parts of Wikipedia out into groups, and say identified a subset of pages to be 'Math Wikipedia' or 'Wikipedia Math', or the 'Math Project on Wikipedia' then they might start by identifying all 'Math' pages, putting together a list of 'Math pages', and deciding how important different pages are, and how much work needs to be done on them.

Views on information: Yes. What I see as missing are 2 things: clear groupings*, a reading order within groupings, and flow. Some articles are contradictory because there were fights and so the top of the page has something opposite the middle. Yes some groupings contain others. But when all the organization happens on (topic) articles, then it's a rather messy graph instead of an list which says 'all these things go together'.

*One way of doing this is to have a set of groupings which covers everything (level 1), then, within each of those top level groups, a set of groups which covers everything (level 2), and so on.

Yes, there may be multiple reasonable such sets.

comment by Said Achmiz (SaidAchmiz) · 2019-06-09T20:04:44.353Z · score: 3 (2 votes) · LW · GW

Basically, my point (which, I think, you have now understood, so I am stating it explicitly for the public benefit) is this:

The problems you outline are not downsides of wikis, the technology—they are downsides of Wikipedia, the project (and some, but not all, other wikis that are run in a similar manner). In fact wikis, the technology, are not uniquely bad, but uniquely good at solving these problems (since they so easily enable the creation of arbitrarily many different views on any given set of information-chunks, or any subset thereof)!

comment by Pattern · 2019-06-10T18:41:10.454Z · score: 1 (1 votes) · LW · GW

While they may make it easy to create different views of information chunks, what's the benefit of such pages if no other users can find them? Having an official, well put together* page hierarchy which starts at the homepage and includes all pages is pretty valuable.

*If the organization system doesn't "cleave reality at the joints" then it's probably not doing it's job.

comment by Said Achmiz (SaidAchmiz) · 2019-06-10T19:30:33.801Z · score: 10 (3 votes) · LW · GW

While they may make it easy to create different views of information chunks, what’s the benefit of such pages if no other users can find them? Having an official, well put together* page hierarchy which starts at the homepage and includes all pages is pretty valuable.

I concur, but again, this is not a problem with wiki technology any more than it would be a problem with book technology if I were to publish a textbook without a table of contents or an index.

An analogy: suppose I say that a knife is uniquely good at cutting things (compared to other tools like hammers, chisels, etc.), and you protest that you have only ever seen knives used to smash things, whereas it’s hammers that you’ve seen used to cut things (with the claw side of a claw hammer, say). That would hardly be a sensible reply, yes? It’s simply that you’re not using knives (or seeing them used) correctly!

In short, you’re saying that if the power of a tool isn’t actually used, then it doesn’t do any good. I agree entirely! The answer is to go ahead and use it, not to discard the tool; knives, books, and wikis are, in fact, quite powerful, even if some foolish people use them to smash things, publish ones without indexes, or fail to create publicly visible index pages.

We ought to learn from the folly of others—not be discouraged by it.

comment by Davidmanheim · 2019-06-12T06:46:28.287Z · score: 1 (1 votes) · LW · GW

Based on the discussions below, it seems clear to me that there are (at least) two continuous dimensions of legibility and coercion, which are often related but conceptually distinct. I think they are positively correlated in most good writing, so they are easily conflated, but clarifying them seems useful.

The first is Legible <--> Illegible, in Venkatesh Rao's terms, as others suggested. This is typically the same as serial-access vs random-access, but has more to do with structure; trees are highly legible, but may not require a particular order. Rough notes from a lecture are illegible (even if they are typed, rather than hand-written,) but usually need to be read in order.

Coercive <--> Non-coercive, mostly in the negative sense people disliked. Most of the time, the level of coercion is fairly low even in what we think of as coercive writing. For example, any writing that pushes a conclusion is attempting to change your mind, hence it is coercive. Structures that review or present evidence are non-coercive.

I think it takes effort to make something legible but non-coercive, and it is either very high effort OR badly structured when they are illegible and non-coercive. And since I've brought up Venkatesh Rao and mentioned two dimensions, I believe I'm morally required to construct a 2x2. I can't upload a drawing in a comment, but I will "take two spectra (or watersheds) relevant to a complex issue, simplify each down to a black/white dichotomy, and label the four quadrants you produce." Given his advice, I'll use a "glossary of example “types” to illustrate diversity and differentiation within the soup of ambiguity."

Paternalistic non-fiction writing is legible but coercive; it assumes it knows best, but allows navigation. The sequences are a good example, well structured textbooks are often a better example. Note that being correct doesn't change the level of coercion! There are plenty of coercive anti-evolution/religious biology "textbooks," but the ones that are teaching actual science are no less coercive.

Unstructured Wikis are illegible and non-coercive; the structure isn't intended to make a point or convince you, but they are also unstructured and makes no effort to present things logically or clearly on a higher level. (Individual articles can be more or less structured or coercive, but the wiki format is not.)

Blueprints, and Diagrams, are legible but non-coercive, since by their structure they only present information, rather than leading to a conclusion. Novels and other fiction are (usually) legible, but are often non-coercive. Sometimes there is an element of coercion, as in fables, Lord of the Flies, HP:MoR, and everything CS Lewis ever wrote - but the main goal is (or should be) to be immersive or entertaining rather than coercive or instructive.

Conversations, and almost any multi-person Forum (including most lesswrong writing) are coercive and illegible. Tl;drs are usually somewhat illegible as well. The structure of conversation is hard to understand, there are posts and comments that are relevant that aren't clearly structured. At the same time, everyone is trying to push their reasoning.