[mjw1007]

When I come across bad documentation (which is often), by far the most common problem I find is that the information I needed just plain isn't present.

Articles like this one really aren't helping. If you wrote a piece of software I'm using that doesn't make you my teacher. It makes you someone offering a contract, and what I need to know is what that contract says.

The first duty of your documentation is to be complete and correct. Unless you've got that sorted, no amount of "putting yourself in the student's place" is going to give adequate results.

[Swizec]

There are 3 types of documentation:

1. Why/what 2. API spec 3. Tutorial

You need all 3. They are distinct, use different styles, and exist for different purposes and audiences.

[kkapelon]

You actually need four https://docs.divio.com/documentation-system/

Explanation (why, what)

reference (API spec)

tutorials

How to guides

[Phlebsy]

There's another more thorough version from the same author, for what it's worth. Just moved on from that company but the overall points are the same https://diataxis.fr/

[baxtr]

Would have been great if they organized this document exactly as they propose. But they don’t? Which is a bit weird to be honest.

[taneliv]

But they sort-of do, no? The left hand side bar has the top-level (bold) headers "Home", "Start Here", "Applying Diátaxis", "Understanding Diátaxis" and "Colophon". The middle ones could well be labeled "Tutorial", "Explanation" and "Reference", but that would be very confusing for the reader!

[ranit]

Right! The good old days when the software used to come in a box with several varied width books containing all these Four, and one named Getting Started.

This same set could be easily "transposed" to the contemporary world of web. With all the proper indexing. Why is this "art" "lost" for most of the software :-( ...

BTW, one Excellent incarnation of this documentation art is on the front page right now:

https://news.ycombinator.com/item?id=43381627

[cogman10]

> Why is this "art" "lost" for most of the software.

Before the internet, the printed book was all you really got. That meant the company distributing the software had to hire technical writers who'd work with the software devs to create all of this, send it to an editor, and ultimately get published.

We no longer live in an era where tech companies hire tech writers. Software documentation lacking is something that can and limp along with jira cases and support services sold rather than trying to put in the upfront effort to fix everything.

Now, for open source software, hate to say it but the docs have always been pretty crap. Certainly some stands out (usually when the business model was around providing services on top of open source software), but nobody is really paying anyone and few people really want to do that sort of free labor.

[pnutjam]

Some places still have very good tech teams, but it still requires working between the teams. (EX from a company I worked with: https://forums.madcapsoftware.com/viewtopic.php?t=30365)

The writers can't do their work without input from the technical side and the time for that is often not avaliable.

I know I've been punished for taking time to push ideas to tech writers. Not only does it slow me down in other places, it often gets swept away as unnecessary changes because upstream Sr techs disagree.

For example, when I modify ssl configs, I alway reference the files with soft-links. This makes it so you don't need to modify the config files and simplifies keeping old and new certs so you can flip back during the overlap period you should be providing to test. I try to avoid editing production files by hand whenever possible because in my experience it introduces the possibility to create errors.

I rewrote some docs extending alot of areas with example commands showing how to test things, with explanations that fleshed out the previous quick and dirty documentation. I also modified the method from copy and replace certs, edit files, restart service; to copy files, replace or create soft-link, restart service.

Upstream approvers trashed the whole thing because the thought it was unnecessary and disagreed with me about manually editing config files.

[bee_rider]

The downside to all that was software releases happened once every year (or longer). Which, was it actually all that bad, but let’s not be completely wooed by the green looking grass on the other side. There were long delays between new features or bug fixes.

[ranit]

Yes, the upside of the contemporary on-line software distribution is clear. No question about this. "The good old days" :-) reference was only about the comprehensive and ordered documentation that often used to come with the physically shipped software medium.

[roenxi]

I suspect all many documented projects died because people couldn't figure out how to use them. By natural selection, anything that survived the good old days is either obvious or well documented. The C programming language, for example.

[Swizec]

> By natural selection, anything that survived the good old days is either obvious or well documented

Or so much better than the competition that people use it in spite of poor documentation. Many things like that also grow a cottage industry of people making documentation and teaching (see React or Rails).

[Swizec]

Thanks! Knew I was missing something. I guess in my mind tutorials and howtos merged :D

[rgun]

What is the difference between "Tutorials" and "How to guides"?

[jeffreygoesto]

It is explained well in the video, well worth watching.

Tutorials are teaching a method, like which ends of the pliers you grab. They do not assume a lot of domain knowledge.

HowTos pick the user up where she stands with a concrete problem and walks you through a possible solution of that specific problem. Like how to use pliers to twist a wire just firmly enough to hold two things together.

[ChrisMarshallNY]

Often referred to, as “recipes.” —Defining the goal, and a specific set of steps, to achieve that singular outcome.

How to calculate the airspeed velocity of an unladen swallow, etc.

[mevric]

+1 for recipes I have provided recipes in my documentation for complex projects, making it a time-saver for both me and my audience.

[rehevkor5]

Read this https://diataxis.fr/tutorials-how-to/

[baxtr]

I like the concept. I will note the same thing I said further down though:

Why don’t they use their proposed system to explain their own system?

It seems like a wasted opportunity. So odd. Or am I missing something?

[stonemetal12]

I like to think of it as learning a foreign language. You need a text book, an English to language dictionary, and those tapes that walk you through a conversation.

[AlienRobot]

Making these 3 from static websites generated from docstrings is a multi-billion dollar industry called LLMs.

[dharmab]

A docstring won't contain the necessary context for "why", and is something I see coding assistants get consistently wrong without human data.

[bee_rider]

OTOH doctrines at least have a slight chance of being updated along with the code.

[ChrisMarshallNY]

My biggest peeve, is bad indexing.

The docs are there. The information is there, but finding it, is so difficult, that I have often just said "bugger this for a lark," and written my own implementation.

There's two types of docs: The teaching kind, as the article mentions, and the reference kind, for helping folks that already have a start, to find what they need.

I generally do the second. It can be fairly easily generated from inline headerdoc comments, these days, but we still have the issue of indexing, and I still have my work cut out for me, there.

Here's some stuff I wrote about my approach to documentation: https://littlegreenviper.com/leaving-a-legacy/

[Sparkyte]

Absolutely this! If I have to use a search box to find docs and not the navigation pane you know you're doing something wrong with your docs.

[ChrisMarshallNY]

I’m hoping that LLMs may be helpful, here. Sort of like a “documentation concierge” service. Like going to the reference desk in a local library, and asking The Librarian to help you find some information.

[Sparkyte]

I believe so. I am also hoping it can just pull in repositories and tell me more about the code and history relating to it. There is a lot of potential to this. TLDRs are needed for code, LLMs will be an excellent replacement.

[Sparkyte]

I actually like re-writing documentation, but if I can not reorganize the documentation too I am discouraged from re-writing. I believe a lot of contextual information is lost when people use confluence as a dumping ground and don't plan the structure of the documents.

I've emotionally moved to git for documentation, but I can not get people to follow or transition to better documentation methods once someone is emotionally tied to tribal knowledge and communication.

[mixermachine]

I can confirm this from the producer side. The documentation is often neglected.

I got fed up with the Word docx (with code examples) we sent out to our customers and replaced it with a hosted Docusaurus markdown documentation.

It is a lot of work to transfer the knowledge to a the new place. It is a lot of work to keep the documentation up-to-date. It is also a lot of work to formulate things in a widely understandable way with screenshots and good examples.

Still I would do the same refactoring again. While the documentation is still not perfect, our customers now comment that it truly helps them to get going fast with our product. A search function is also very helpful.

Our QA team does follow the integration instructions in the documentation when a major change is released (so "putting somebody in the student's place" does work for us). It of course takes time.

[fn-mote]

I agree that this article isn't doing something for me.

It barely covers the important points of what _should_ be done, and spends a lot of time lampooning bad approaches to documentation.

I also found the metaphor of "the black triangle" obscured the main point of that part of the argument, which to me was "have some foolproof basic starter setup to get a user up and running quickly".

[madsbuch]

> The first duty of your documentation is to be complete and correct.

It is probably unrealistic to expect documentation to be complete as it is not clear where the border on inclusiveness goes.

Take eg. AWSs node packages. Here they spend words recommending using async / await which IMHO is firmly outside the scope of library docs.

I am curious on what you feel is missing? is it elements concrete to the piece of software you use, or is it pieces that can be deemed expected knowledge from a professional software developer.

[mjw1007]

The former.

[ajnin]

Yes, using the analogy used in the article, programmers are not the students trying to learn how to drive, they are the professional engineers and mechanics that build and maintain the cars. They don't need teaching, they need technical documentation. They already should possess the know-how to transform such documentation into practical applications.

The "be a teacher" motto is how you end up with "documentation" that is a bunch of examples of how to do various things and then no exhaustive and rigorous technical reference or even none at all. And then good luck scouring blog posts and StackOverflow or ask help on IRC (pray it's not on Discord) in the hellscape that has become Internet search if you want to do anything not in the examples.

[cjfd]

The article is quite a bit smarter (if a bit long) than this comment. "The first duty of your documentation is to be complete and correct". The source code satisfies this demand. It is complete and it truly describes the product in all its details. But now we are stuck in 'act 1' of the article. So maybe we need some more helpful hints. O wait, there is an article called 'Teach, Don't Tell' that provides these helpful hints.

[mjw1007]

The bad documentation I come across often shows every sign of having taken the advice from this article (and others like it) to heart.

The trouble is that the advice doesn't include ways to make sure your documentation actually covers everything it has to.

What this article says is that you should have « "API documentation" for every user-facing part of your project. »

That commonly leads to "reference" pages which are very little better than the autogenerated documentation the author dismisses. The main problems tend to be that behaviour that isn't controlled by a specific command or function or configuration setting doesn't get documented at all, and that commands with complex behaviour get described using terms that themselves need further definition which isn't provided anywhere.

[AlienRobot]

My experience is that the main problem with online documentation is that it doesn't include links to separate tutorials, but has plenty of links to more documentation, leading you to wander from page to page without destination trying to find that one page that actually explains what you're trying to do or infer it from fragments of information scattered around the documentation.

For example, in Qt you have a view class, a model class, and a selection model class, and none of their pages tell you how to actually select something programmatically. You have to guess or ask ChatGPT these days.

[sourcedemand]

> The source code satisfies this demand.

If it's available...

Also, it's much harder to parse multiple files than one doc. Been there, done that, aplenty. Yes, I can find the answer, but that isn't sufficiently going to make it easier. Plus, if one admits the docs are bad because "the source answers that type of in-depth question", then are the docs not still bad?

[cjfd]

Sure, if the docs are bad, the docs are bad ;-). Jokes aside, though, I think the nice thing that this article does is pointing out various common ways in which the docs can be bad. My 'favorite' one is having docstrings for documentation. They take quite a bit of time to write, clutter the code with helpful information like 'the method get_heigth, gets the height' and if they are the only thing that is shipped as information use users severely lacking.

Another thing to note is that what this article lists as good docs actually cost a lot of time to create. Perhaps this is why people settle for the less than helpful alternatives. I think, though, that it is better to provide a subset of the good docs rather than useless docs that just annoy people.