Ask HN: Are Linux HowTos dead? Why?

The linked site isn't where most HOWTOs were kept; they resided at https://tldp.org/

Are they dead? Not completely dead, but there is only one or two people still maintaining the site + HOWTOs (they organize on GitHub and a mailing list currently). But there's simply nobody volunteering to maintain them or write new articles.

Why is that? Because popular things get action and unpopular things don't. HOWTOs aren't popular.

Why is that? Because blogs and the invention of "Q&A sites" made them unnecessary. Before you would use a HOWTO to teach yourself everything about a specific piece of software or technical thing. Now you don't really need to learn an entire thing. You can just google the one question you have, get the answer, and move on. You still don't know how 98% of that thing works, but you fixed your problem. Since this solves most people's problems, they don't see value in taking the time to write an entire HOWTO, which may take weeks to months for a really good HOWTO. Similarly, users don't look for them because nobody's writing them. There is no incentive anymore.

SEO is part of the reason that traffic began being moved towards blogs and Q&A sites. But SEO alone didn't bring about the cultural shift towards snippets of answers. It was simply a new generation that learned tech outside of the old OSS community, and developed their own ways of learning. Just like the old OSS community created their own way of learning different than their previous generation.

> Before you would use a HOWTO to teach yourself everything about a specific piece of software or technical thing.'

The whole point of HOWTO's as a documentation format is that they don't try to do this; rather, they address specific user needs. They're somewhat like the "next step" beyond simple Q&A and FAQ collections, and this is where much of their value could still be found.

It was one guy maintaining it. I spoke with him a few years ago about it. I really wish people who do volunteer work would post their wallet address. Even if no one gives at the time when people miss your work it can be a good way of letting them know how much.

>SEO is part of the reason that traffic began being moved towards blogs and Q&A sites.

Exactly mirroring the decline in the quality of google results.

You want something specific with say 4 words in the query string, convincing google not to drop one of those in basically every result it returns giving you nothing but SEO spam is a lot more difficult than it should be. Tried googling for something you read last week following a link and now can't find again despite it being about something really esoteric?

So yeah you can work around some of the goog breakage with :allintext or putting a + in front of ever term or putting every term in "quotes" but yeah, at the stage where you're the audience for howtos are you doing that? Dunno.

Who is old enough to remember when google was /liked/? Seems like a long time ago now, huh?

I’m that old. Google was everyone’s favorite thing they’d ever seen on the Internet. Ditto GMail, Maps, etc.

Want to really blow people’s minds? The same was true for Facebook.

These products didn’t get big enough to have power to abuse by being unpopular. People loved these products.

I don’t recall it that way. Everyone in tech was impressed by Google circa 1998 - it was the first search engine that really didn’t suck in some way. Facebook never had that level of tech cred, although it wasn’t as reviled early on as it is now.

I remember the love for Google products, scoring an invite to gmail. I don’t remember a distinct love for facebook. It was much cleaner than MySpace, and kind of exclusive because I needed my university email, but not love.

I'm old enough to remember when google was well loved. And I'm really not that old.

I'm old enough to still love Google. I think it is the wonder of our age.

Hell, I'm old enough to remember when altavista.digitial.com was the most amazing thing I'd seen. But the worse Google results today are as good as the average results from that or any other pre-Google search engine.

In the homeassistant ecosystem most tutorials are youtube videos. I don't like it to much, but it works mostly. I guess thats the new way of doing it.

Ugh video tutorials are the worst. Nothing like wasting 5 minutes watching a video that turns out to not even have the info you’re looking for.

The only thing worse than video tutorials are what they do to your YouTube recommendations.

I absolutely hate both with a passion, but having mentored a fair number of junior developers I respect that some people benefit from them.

It's the same thing with podcasts and audiobooks- I'm the type of person who is fully tuned in, or I am tuned out. Listening to either while driving is completely pointless for me.

Long live text!

Agreed. But not only that, it is far easier to refer to an earlier point if it's in print, whether dead tree or electronic.

To be honest the the HOWTO system was clearly starting to die even by the early 2000s, before blogs and "answer sites" (as distinct from web forums more generally) really took off as sources of Linux advice.

Google/copy/paste/repeat is the quick and mindless replacement for read/understand/write/debug.

AI for machines and humans alike - and it works much better than you might expect.

Not an answer to why it is dead, but the Archlinux wiki, which offers decent replacement for the functional part of the traditional HOWTOs, have been around since 2005 and have established itself as a decent info source for Linux users at large, not just the arch initiates.

Probably it is an answer. People gravitate to a distro offering them docs in their preferred style. And some of these people contribute to docs later in their preferred style. This process becomes self-enforcing at some point: the style is refined, the knowledge base grows, the community matures... It doesn't happen overnight, so it didn't happened in 90s.

And one more reason for a late arrival: in 90s there were less people on Linux, they were not able to create all kinds of knowledge bases, so they resorted to a more achievable project of creating HOWTOs, which was meant to be good enough for everyone.

Second ArchWik it's quite good and not tied to archlinux

Yeah, that and the Gentoo wiki have gotten me past most major hurdles.

I think another answer is that much of what is covered by the HowTos is now built into the distros themselves; there's no need to learn X Y and Z if you want to do T, as you just select the "LAMP STACK" option in the distro installer.

Yeah - I just set up Gentoo for the first time, and the wiki is _excellent_.

I had this preconception that Gentoo was ‘complicated’, but the excellent documentation has made everything easy (and I’m coming round to the opinion that Gentoo is actually easier because how it works is so ‘transparent’).

I found Gentoo much easier if you knew you were going to stray from the beaten path (e.g, custom kernel, weird library choices, etc).

It's been surprisingly stable for me considering everything I throw at it, and much of what I learned has helped diagnose issues with other distros.

I've found the Gentoo wiki very helpful even back when I was running no Gentoo systems.

The state of Linux docs as seen through Google is appalling. Searing for things like "how to set up rsnapshot backups" are full of garbage articles that are 800 words like whose only actual content is "apt install rsnapshot". Maybe this is a Google problem.

It doesn't help the Ubuntu docs have fallen into total disrepair. Most searches for information there find 6 year old stuff that's no longer relevant. Major sections of manuals have apparently been abandoned.

The bright spot is Arch Linux; that wiki is amazing. I now am adding Arch to most of my search queries even though I have no intention of ever using Arch.

Archlinux relies on a very small set of distro-specific tools, everything else is standard tooling and code pulled from upstream without any extra patches. It makes sense their docs can apply to the rest of the Linux world. The Archlinux wiki is amazing.

Doesn't help every flavour of Linux might have a different way of doing things.

Heck, the mess that is systemd and changing your hostfiles or nameserver on newer versions on Ubuntu is baffling with every version having its own quirk. I spent 30-45 mins on something that took 10 seconds in another flavour of Linux the same day.

Windows commands are still largely the same the were 10-20 years ago.

I can't agree more. I'd pay for a printed version of the Arch wiki. I've used the Arch wiki to fix Windows systems before.

If it’s simply offline you want, you can install https://www.kiwix.org/ and use it to download, search, and read, an archive of the entire Arch Wiki.

Bad SEO I suppose. When I search for something, I often get a page by some "expert" in which the information is wrong, commands don't work and so on. I wish I could figure out how to eliminate them from my search results. The worst part is that it is usually impossible to contact the author and point out the errors. Like they would care.

I think there is a law in India mandating everyone to open a blog and explain linux to people. This must be done in the first 3 months of learning linux.

This is certainly true for RaspberryPi tutorials about interfacing with thermometers or other sensors. All the DHT-11 tutorials seem like minor mods of some long lost original, except with new spelling mistakes, and inaccurate pinouts.

This would seem to confirm that advertising ultimately corrupts ad-supported media

The YouTube videos are better because they need it to work on screen.

Videos are slow, and it's hard to re watch tricky pieces or compare with other sources.

I wish someone in the ML community would take this on: create a document from video, maybe using the appropriate frames of the video as media in the document. Give people something to glance through instead of having to watch a whole video. Bonus points for making the document interactive with links to appropriate sections of the video as well as links out to the world for products/definitions/etc. Double-bonus points for the ability to squelch awful humor and filler words.

Until Google decides to take down one for this or that reason. I would rather use a personal blog (backed up at archive.org just in case) with all the necessary text information, plus links to a video hosted on a decentralized video hosting platform, then maybe publish the video, or part of it, on YouTube as well to attract audience.

Unfortunately you’re probably right. SEO optimized pages have basically taken over most search results, even when they have glaring mistakes :(

If we could get the search results from a couple years ago, without the SEO.

Let's be honest Google is having a problem dealing with actual intelligence writing bad stuff as well as surgical intelligence rehashing things. Just because half the internet writes a thing doesn't make it true. And again just because the airport writes it doesn't mean it's tested. Figuring that out by raw ML and no context of the material is probably not going to end well. I think the only solution is will be to keep an updated truth curated by experts humans.

Compared to the 90s where spam was spam and three internet was more for the tech savvy it's changed.

Full-Internet search is dead. Collect and share bookmarks.

So we basically need to go back to the days when Yahoo! was simply a directory of links to cool websites. Curated by humans.

But decentralised on the edge network (top edge of browser, below the url bar)

Would be nice, but today's Yahoo would be SEO-steamrolled too I'm sure.

IMHO what is needed is an open, FOSS, royalty-free database-like format for URLs and descriptions NOT intimately connected with the presentation layer (HTML) like RSS.

Open Graph is kinda the start of what to collect per URL.

Then we need browsers to be able to add/delete/update on this database, let you search through it locally, and allow import/export of portions or the entirety of this database.

This is in direct contradiction to today's walled gardens, so it's not ever going to come from FAANG and friends.

Do not mention he old magic to me, I was there when it was written!

Bad OSE (optimizations on the part of the search engine), you mean.

I used to write articles like this for my blog. There's a reason they really don't exist anymore:

First is that documentation has gotten much better. A lot of times a repository will have examples alongside it. Man pages, help menus, and entire statically generated documentation sites have all advanced this quite a bit - and much of what they do share space with what I used to write about. It's worth mentioning that a lot of times, I was writing to a very specific end goal. Documentation sites will usually what you through what I did and then some.

Second is that Linux userlands have diverged a lot. There's not a ton of standardization around userspace tooling, so it makes writing an article (that needs to be updated) an up-hill battle.

Third, Linode and DigitalOcean use these kinds of articles as PR. They're high quality, often versioned, and help users understand broader contexts as well.

So, are they dead? Yes, in a sense. They inspired a lot though, so in that way I think they still live on.

In addition to userland being non-uniform there’s also outrageously quick development so everything is outdated within a few months. As such, maintenance burdens for documentation is super high. There’s also always the ArchWiki.

Personally I believe Linux has matured to a pretty decent usability level for the average user. Ubuntu, Opensuse, Linux Mint and some of the like have taken a lot of the hard work out of managing and installing a linux OS. It used to be that you would slave hours of getting your specific hardware to work with a particular flavor of Linux. Now you have drivers that are plentiful and up to date, bugs that get fixed fast, and large communities at near enterprise levels maintaining the updates. Frankly a pretty good time to be a Linux user.

While the Linux installers and hardware support have improved more than any other part of Linux distros hardware support still lags well behind actual hardware availability.

That you belive that the installers have advanced to "average user" level really misses the point though. The rest of the os is as disjointed and messy as ever.

Are there any, I don’t know, archives of what it used to be like to install linux? My first linux experience was Ubuntu maybe a decade so ago and it installed fine.

What’s not immediately obvious from this is that it was not uncommon for an installation to require about 30 floppies. Floppies of course are not 100% reliable and in my experience there’s a pretty good chance that at least one of the floppies will be corrupt. For me, this meant a trip into the campus computer lab to download and write a fresh copy of the needed disk.

When you decided that you wanted X11, you would need to follow the readme with the understanding that getting it wrong could fry your monitor.

This also reminds me of a Softlanding Linux System (SLS) boot floppy disk set that was our gateway into Linux. It served the use case that people would use Live ISO images now. You booted a PC off floppy and used a single floppy for the root system image, and could use the typically available second floppy drive to save some user files. This let you preview Linux on a system without the leap to wipe and reformat the HDD.

Seeing multiple virtual consoles with login prompts on your first PC was a revelation if you had been amazed by the multi-user systems in school labs and assumed PCs were inherently single user.

Maybe Slackware could be used that way too, but I never tried. By the time we started downloading Slackware floppies, we planned to wipe a system and do an HDD based install to get all those extra packages that would never fit in a single floppy.

Looks to me the same as it is now. Figure out where to install, partition block devices, format filesystems, copy initial rootfs content somehow, configure basics like fstab, networking, setup users/passwords, install bootloader, reboot. Done.

Slackware just has very detailed documentation. :)

Linux community documentation falls into a few categories:

1. Copypasta that appears on many different websites. Usually the first hits in Google. Usually terrible.

2. Curated, quality material on sites like Linode or Digital Ocean.

3. Blogs from people who actually work with the stuff and have ran into interesting problems/solutions. Can vary but usually pretty good.

4. Distro docs like Arch wiki

Imho, the reason for it is that now we have much more sources of information. How tos were a concentrated pieces of information distributed with Linux distributions or via slow mediums… 20+ years ago I was involved 8th Russian translation of many how to’s and they were used because there were no other sources of information, but now you can just translate almost any page to language that you need

Wow. This post reminded me that I wrote a HOWTO 20+ years ago:

https://tldp.org/HOWTO/Emacs-Beginner-HOWTO.html

It's still sorta relevant, I guess...

What a blast from the past.

People don't read anymore these days.

Which explains why so much of modern software sucks.

Eh, I’d argue the opposite: developers expect the end-user to rely on Stack Overflow or perusing GitHub issues to figure out how to use their product, instead of making easy-to-use products.

What I miss more is the “rtfm.mit.edu” FTP site, which was a repository for all the FAQ’s of all the Usenet newsgroups. I don’t even know if it was mirrored anywhere.

(I know about faqs.org, but those are frequently wildly out of date compared with those on rtfm.mit.edu, which seemed to be continously updated.)

Nice, thanks! Now, assuming that at least some FAQ’s kept updating even after 2015, where do these new FAQ’s get collected?

EDIT: I just found <http://www-ftp.lip6.fr/pub/doc/faqs/> (and <ftp://ftp.lip6.fr/pub/doc/faqs;type=d>) which seems to be exactly what I am looking for! It has updates until about 2018, which seems reasonable.

Except for a few programs like ss or nftables nothing has changed. you can still use shred, despite the usage of journaled filesystems, cp works like 20 years ago and nothing has changed with ln and i make still errors with the target and the source. so the old howtos are still valid.

The whole networking userspace is changed.

There's no ifconfig, netstat, route, iptables, ipchains, I think there was one more. pumpd is dead, isc-dhcpd is probably dead (replaced by systemd-dhcpd).

It's not really recognizable.

> (replaced by systemd-dhcpd)

Not everybody uses systemd. There are still a few distros that deliberately don't use it or at least offer their users a choice. I personally have two Linux systems on which precisely nothing has been replaced by any part of systemd.

ALSA, sysV init, xf86config ramdac settings... and a million other things, not just networking

`nftables` may have replaced `iptables` but I don't know it from my current Debian router - my complex `iptables` firewall and routing script runs great.

`ipchains` was replaced by `iptables` in the early 2000's I think.

Also I got `isc-dhcp-server` running and kicking just fine. And straight `bind9`. I know `systemd` does both of these but nothing's forcing you to replace it-at least on Debian anyway.

Modern debian uses an shim to ease migration to nftables.

What is it with ln? 20+ years using it and I still have to read the man page every single time. Whenever I try to do without I always end up with red links

Yes, same here! Even after reading the docs I still get it around the wrong way.

I wonder why it's so hard to internalize?

I switched video cards recently. The right command was "mhwd -Ia" or something.

Then, I tried pairing some Bluetooth headphones with a USB dongle that has regressed default firmware. Not only had everything changed, I hadn't even heard of the deprecated programs. The only cli tools I used from memory were "sudo", " dmesg", "grep" and "ln/rm/etc".

> you can still use shred, despite the usage of journaled filesystems

The docs say otherwise. Journaled filesystems and SSDs are both noted.

https://www.gnu.org/software/coreutils/manual/html_node/shre...

What i meant was: is doesn't crash the pc. And there are still unjournaled filesystems out there

This is misleading, plenty has changed and Linux backwards compatibility is horrendous on a good day.

ln works just like cp.

Use the -r flag if you're on GNU for easier relative link creation.

Stackoverflow et al probably win the search results. The landscape is also too diverse to write a howto that would be very broadly applicable.

I think the simple answer is that distributions got a whole bunch of polish and automation.

I install Ubuntu these days by building a boot drive using a gui tool, then just clicking my way through the installers. Getting online involves ... typing in my password. In the late 90s I had to figure out AT commands and set up ppp and so on, and getting X11 up and running required config voodoo.

I miss some of it. Not that much though.

While formal HOWTOs might not get much attention, it's pretty clear Linux tutorials of all types are both being made and consumed in a variety of media. People often bring up how much value they get out of the articles DigitalOcean puts together, for instance. Those aren't HOWTOs necessarily, but does it matter?

There's also the fact that people just consume that kind of content differently nowadays. While it's hard for me to understand, people younger than myself seem to get a lot out of guided video tutorials over text.

Everything got to complex.. way to many variables for even the simplest of jobs these days.

Practically nobody writes classic HOWTOs anymore.

At best you get amateurs speaking authoritatively on subjects they've barely scratched the surface of in blog posts they can try monetize, at worst you get the same thing but in the form of a youtube video.

I dunno IMHO there's a lot of rose colored glasses looking back at them. I cut my teeth learning Linux in the era of howtos and such of the late 90s/2000s, and it was a real mess. Someone on an IRC channel would tell you to RTFM and point at a guide, you'd read it and slog through pages and pages of bloviation about unix philosophy, incredible minutia about every CLI parameter (half of which don't exist anymore) or whatever and be even more confused than when you started. You'd go back to IRC and someone else would say oh yeah don't read that guide, it's old and not up to date. Here read this guide... and the process starts over again. There were some good guides, but it was really hard to find them and sort out the good from the bad.

I spent my teens installing linux from floppies and Infomagic CDs. Back then the distro would often by default include the howtos package bringing in the whole enchilada of knowledge.

They were generally relevant to the versions of software included, and updated when the rest of the system did.

Things started getting messy as everyone got connected 24x7 and increasingly diverged from what packages distros were shipping, and of course the HOWTOs in general just stopped getting maintained/created until we arrived at today's misery.

Honestly, I still write them. Either for my private documents or office documents.

I’m planning to open my personal trove step by step, on my self hosted server, via a static site. They are Linux/development documents too.

Note: No ads, no tracking, GFDL licensed.

There’s some preliminary stuff I wrote: https://bayindirh.lists.sh/Useful%20technical%20documents

Same. Things are getting so noisy and hard to find that even if good howtos exist somewhere on the internet, finding it can waste hours vs just word searching your own documents.

The linked Evernote pages don't even render here (noscript). Following a long delay it just shows a blank page.

Excellent demonstration of how awful the status quo is vs. opening local files in /usr/share/doc/howto/txt/ with `less`.

In today's OpenCV video, we'll show you how to find the red cup just like the docs.

Make sure to smash that Like and subscribe button.

This video is sponsored by SKYN condoms. Get a 6-pack for just $8.99 that you can truly feel.

And now for the red cup with OpenCV.

Limiting the search to github gists as a first option is a pretty decent way to escape the machine generated algorithmic SEO garbage.

Until HowTo’s get sponsorship, like streaming, that is how it is going to be. Not that enough people RTFM to make it worthwhile. The best way to explain Linux is socially, though.

The best way to learn a subject is from an authoritative source. 'socially' results in a whole bunch of misinformation and wrong practices, like PHP developers chmod'ing 777 the hell out of the system because they don't understand what the fuck they are doing.

Personally, I get almost all my Linux "how-tos" from either Stackoverflow or, for more basic questions, ubuntuusers.de.

The latter one is a German wiki that has a similar premise to linuxhowtos.org, but it seems to be much more active and most entries I was looking at were up-to-date. Kinda weird because I consume tech articles and man pages almost exclusively in English – with the big exception of ubuntuusers.de because there doesn't seem to be a good counterpart within the first few Google/kagi results.

I think at this point the Arch Wiki has supplanted all other sources as the definitive source of Linux howtos, given the size of the community and its values.

The particular site referenced by the OP is not the only source. I spend a fair amount of time writing howtos, mostly about Linux. https://mslinn.com/blog/blogsByDate.html

I'm not dead yet, and neither are howtos!

This to the very extent, is still very valuable for me to a personal level. However I can understand why people would divert away from it, people have grown apathy for comprehensive and deep heap of information and want "quick" low quality "fixes" for their problems. Part of the reason why "googling" is so popular and the index of information is represented in terms of SEO while still questionable on the quality of information and reliability yielded by SEs like google.

Yeah it's been a while since I consulted a leggit HOWTO... I think that HOWTOs get created by volunteers all over the web nowadays.

The site is dead because of lack of maintenance and everyone hosting their own how-to content on their github, podcast, youtube, website and Q&A site. If this was re-imagined, it might be an aggregator that served as a directory to all the howto content...

The best Linux HOWTOs I ever read were all in print magazines about Linux. Those that are still around still do them!

Mostly because Linux is 'good enough' for most use cases out of the box nowadays. Between Ubuntu, Debian and a bunch of other distros there is one out there for everybody with most of the stuff that you'd need a howto for in the past working out of the box. This is a positive development imo.

If you want high quality resources and skill and talent, you need to pay for it. I don't think Linux HowTos was a paid service such as lwn.net