10 ms·

Rabbit Holes: The Secret to Technical Expertise

276 points by dailymorn 7y ago

jaequery 7y ago
Yeah this is just overkill. I haven’t edited /etc/motd in a while so this was news to me but this is just code creep and makes me sad.
- hedora 7y ago
  Same here. My reaction at each step is “how do you delete this subsystem?” and “what poor person wrote this?” Still, nice learning exercise for the author.
  - pjc50 7y ago
    Exactly. I'm sure somebody wanted it at some point somewhere, but now it's on every system. No wonder people keep starting new minimal Linux distros.
- edejong 7y ago
  That. They could’ve just supplied a default .profile / .bashrc. It’s these kinds of bells and whistles that causes most security bugs due to additional attack surface. Is it the root user fetching the news with wget/curl from a remote source?
- enriquto 7y ago
  All of systemd is the same shitfuckery. I move away from debian to slackware and voidlinux (when using BSD is not possible) just to avoid it.
mustntmumble 7y ago
As one who simply tinkers with Linux but is not a guru, I found this article quite inspiring! I had no idea just how deep that sort of rabbit hole can go. I feel inspired to try to delve behind the scenes for myself... The author has made my day :-)
cle 7y ago
The trick is knowing how deep to go, which depends a lot on your goals. If you just want to learn for your own curiosity, rabbit hole forever! If you want to build useful, applicable knowledge, there’a a peak of usefulness after which there are diminishing returns. So being self-aware enough to know when to pull the plug on your rabbit hole is pretty important (professionally). Also certain kinds of rabbit holes are way more useful than others. If you’re rabbit holing into some obscure, arcane system, and the stuff you’re learning isn’t challenging you fundamentally, you’re probably wasting your time (I call these “rat holes”). If your rabbit hole will yield new fundamental insights into something important...now that’s a good rabbit hole!
- Robotbeat 7y ago
  The best depth before you hit the knee in the curve is undergraduate-level textbooks. Sometimes you have to go deeper, but that's the level that is presented in a form that's easy to grasp and can be quickly applied without using too arcane language and abstractions. Also: Handbooks. For instance, the Art of Electronics. And I recently picked up a mechanical engineering handbook, and the breadth of the topics covered at a level I can quickly understand is very refreshing. (I'm a physicist, but I mostly do mechanical engineering and technician work, with a bit of aerospace and electrical engineering and material science.) And combined with youtube and Wikipedia, you can learn a lot very quickly.
  - tasogare 7y ago
    > The best depth before you hit the knee in the curve is undergraduate-level textbooks. That matches my experience as well. Having graduated from two fields at university, the most foundational knowledge came in the first two years. The third one is a deepening of the first two but can be skipped if only basis are needed. Then master degree is just playing with higher level concepts and linguo.
  - voltagex_ 7y ago
    >Art of Electronics What would you suggest before The Art of Electronics?
HenryBemis 7y ago
Shortcuts is my personal mortal enemy. I talk to people that "really want to learn topic X" but: a) don't want to read any books, b) don't want to watch any training videos (from nuggets in the old days, to youtube tutorials now). They just want a quick and dirty solution to everything, always. That comes with a price. People want shortcuts. I don't like or use shortcuts 99% of the time. My experience is that when you go looking for X to serve a need today, you discover Y and Z and guess what.. Y and Z will come very handy tomorrow, when in a call/meeting about ABC, miraculously Y and Z are similar and you speak up a d impress everyone and get the job/contract.
- kregasaurusrex 7y ago
  Knowledge that compounds in such a way becomes useful in the long term. The Y and Z mentioned may be tangential in a current situation, yet as a technology stack evolves they can become integral parts to more complete and better fitting solution.
- stevekemp 7y ago
  I frequently get emails from people which follow a predictable pattern: "I'm doing a PhD in .. I saw your blog post .. can you point me at a tutorial for advanced-subject?" I often wonder why people don't do research, and what would happen if tutorials just disappeared. People would have to do their own research and learning. Maybe things would take longer, but I think the people would benefit from it.
  - ackbar03 7y ago
    I'm not sure if I can speak for everyone but sometimes I do this in the hopes that it could lead to a connection with the author or potentially deeper insights and discussions on the topic with someone more knowledgeable. Usually I go off and try to figure out what it was I was asking in the first place anyways
    - stevekemp 7y ago
      I'm happy to answer questions, and sometimes I can point to non-obvious resources. But mostly I feel that the approach would/should be different for trying to discuss things than "Please help me do my homework"
sharadov 7y ago
I love this kind of stuff, it's the most pure kind of learning, no end goal but satisfying your innate OCD. I've gone down these rabbit holes in tech and alternative fields - politics, philosophy, cooking. It's crazy, but it lead me to some life long hobbies.
LogicX 7y ago
Or positioned another way, this is Yak Shaving: Any seemingly pointless activity which is actually necessary to solve a problem which solves a problem which, several levels of recursion later, solves the real problem you're working on. http://yak-shaving.urbanup.com/1164318 http://yak-shaving.urbanup.com/1164318
mkl 7y ago
The metaphor critique paragraph seems to be missing the fact that name comes from Alice in Wonderland, where the protagonist falls "through a rabbit hole into a fantasy world": https://en.wikipedia.org/wiki/Alice's_Adventures_in_Wonderland https://en.wikipedia.org/wiki/Alice's_Adventures_in_Wonderla... I.e. it makes complete sense in context.
acangiano 7y ago
Reminds me of how, as a kid, I tried to solve Fermat’s Last Theorem. I didn’t have a chance in hell, of course, but my attempts enabled me to learn a ton of advanced math and even finding on my own some results that turned out to be already well known (e.g., Euler’s formula). It's one of the purest forms of learning but as an adult, I no longer have the time or freedom to do it to the same extent. Still, the basic approach is worth keeping. You just need to limit yourself a little so that you don't end up too deep in the rabbit hole, spending a ton of time on things that you really don't need to know.
- codingslave 7y ago
  I had a similar approach with the following equation: What does the infinite series of 1/(k^2) converge to for k=1 to infinity. When I was in high school, I spent weeks going down this rabbit hole of mathematics, failing to solve it. Years later I discovered that this equation is a special case of the Riemann Zeta Function, which was also first studied by Euler. His proof is really elegant. The funny thing is that with some math equations, sometimes you intuitively can see that there is something important or far reaching about an equation, even if you arent skilled enough in mathematics to discover it. Some kind of curiosity draws you in
mkorfmann 7y ago
KARMA TANK INCOMING
simosx 7y ago
There was drama about the MOTD two years ago. Here is the summary from LWN, https://lwn.net/Articles/726902/ https://lwn.net/Articles/726902/
dusted 7y ago
So, blur the hostname in the SSH command, login to machine, where the hostname is ip-172.31.48.7.. Two things are funny here: First, the failed attempt, and second, the idea that anyone cares what the IP of your raspberry pi is.. If anyone cares to know, mine has 192.168.30.12.
- reportgunner 7y ago
  haha, perhaps there was some profanity that had to be blurred out
lordnacho 7y ago
This is the reason why people who don't respect coding as a discipline end up producing terrible code. In my professional life I've run into a lot of people who will say "the idea is already there, it just needs to be coded up". They will then come across common coding issues like keeping things in version control, adding tests, network config, and language specific things. They decide they don't have time for this particular rabbit hole, because of course they already know what they want. The good people will over time end up investigating everything, even if it doesn't seem relevant to your LOB application. And they produce much better code because of it.
- la_barba 7y ago
  Its not that people don't respect it, its that they don't appreciate it. And the reason they don't appreciate it is that those things are invisible to the end user. How much extra money would you pay for an electrician or a mechanic who does the equivalent of what you're suggesting with code? 1.5x? 2x? Most people I know always complain that their garage overcharges them or that they got ripped off by some contractor or the other...
- jjeaff 7y ago
  As someone coding for my own business, I often find myself deep down a rabbit hole and have to get out and force myself to just use that ugly patch. It's a balancing act, because it hurts me to just leave something without completing it "the right way" but I have found that if I try to do everything "the right way", I would get so little done. I try to keep this in mind when reviewing or working on other's code. Some things will anger me at first glance but the problem is, they were probably just prioritizing different things at the time than I would have prioritized right now.
pjc50 7y ago
I'm not sure about "technical expertise", as most of the stuff you learn while doing this is highly specific, but perhaps the real expertise is having the toolset and determination to do this at all. To keep taking screws out of the system and removing the covers until you find out where the system makes a particular decision. Fortunately in this case you can go to the source; if you can't, or as a first attempt, strace will usually tell you where a particular program gets its information from. Harder to use here as you'd need to setup the login process itself with strace and there's privilege boundary considerations. One observation I would make is how many layers of "plugin architecture" have been added here. Probably the original SysV login program just printed /etc/motd. Now we have PAM, and pam_motd itself also runs a bunch of tiny plugins, some of which also refer across to systemd. I should probably write up some of my own deep-dive debugging stories, if I can still remember enough detail. The "why does structured exception handling sometimes not work on Windows CE" was a particularly time consuming one which involved disassembling bits of Windows kernel.
jimnotgym 7y ago
> * Keen to learn Istio? It's included in the single-package MicroK8s. But this is a rabbit hole! His rabbit hole is about removing a rabbit hole.
Tepix 7y ago
The "right" solution to get a quiet login is to touch ~/.hushlogin in your home directory. It's documented in the login(1) man page. There's also a global /etc/hushlogins file you can create to take care of all users.
tombert 7y ago
Rabbit holes are why I grew to love Haskell and pure-functional programming in general. For better or worse, Haskell (Agda, Idris...) doesn't let you compromise very often. You could argue that this is terrible for productivity, and you might be right, but at the same time, means that a lot of people have done a lot of research on how to do things the "right" way in Haskell, and any project I've done with the language has led me to learning a whole slew of new things about type theory, category theory, and PL theory in general, which I can then take and apply to "productive" languages.
- shanemhansen 7y ago
  This aligns precisely with a memorable interview I gave to a PhD student who wanted to do an internship at a startup. "The thing I love about Haskell is that even the simplest things become research projects" However I understand where you are coming from. There's something to be said for environments that encourage learning.
  - tombert 7y ago
    Every now and then I get mildly embarrassed because someone has managed to summarize my long-windedness to a beautiful and elegant sentence. The line "The thing I love about Haskell is that even the simplest things become research projects" has done that. I'm going to steal that.
drawkbox 7y ago
We use platforms and layers that are at the level of branches and leaves on trees, supported by abstracted away structures hiding complexity that make them simpler to use as a platform higher up the tree (or down). A researching traveler can go down the branch, to the trunk with lots of time rings and layers, and then further down there are many root systems and rabbit holes. All sorts of functionality is abstracted away in layers down there, all the way to machine code and binary and everything before the epoch 1/1/1970. Epic programming and technology pioneer legends created it, they took complexity and made a simple platform layer for others to build on.
peterwwillis 7y ago
It seems like the difference between a complicated, difficult system and a simple, easy system is whether it changes over time. I feel like a Linux kernel in a NetBSD userland might be the best sweet spot, OS-wise, but I haven't used NBSD in a long time.
prometheus76 7y ago
I'm going to go meta with my comment and say that this article is a fantastic object lesson as to why linux faces extremely low odds of ever being adapted as a widespread replacement for windows or OS X. How many people (outside of your coworkers) do you know who would or could follow a path similar to the one laid out in this article? I'm speaking as someone who has dabbled with linux since 1998. At that time, and now, there is a huge, confused ball of conflict between linux distributions, various levels of customization to those distributions, and documentation available on the web. Here's a quick, recent example from my own experience. I decided I wanted to learn more about Emacs. I work in Windows at work, so I downloaded portacle. Great. Works great. Then you start watching youtube videos on how to customize Emacs, but most youtube videos are using Emacs in Linux. So you then have to find where portacle puts all the stuff that is in a completely different place than all the Emacs documentation and videos say it is. Great. And I haven't found anyone who talks about the benefits/costs and gotchas about whether to edit the init.el file, or the .emacs file, but now we're getting into the weeds. So after that frustrating rabbit hole, I think, "I'll just install a linux distro in a VM." Next thought: "Which distro should I get?" Search engines are no help when it comes to finding the "best" linux distro because all I can find is superficial opinion pieces on the matter. So back when I worked at an ISP in the early 00s, they used Debian, so I download Debian. Try to install emacs in Debian, and I start getting weird errors about some certain package not being current enough, and apparently someone isn't maintaining/updating the install script for Debian. That's three days lost getting to that conclusion. So I wipe that VM and download Ubuntu, because it seems to be the most widely used distro, and seems like a good distro for "normies" like me. Install emacs, no problem. Can customize by following youtube videos, no problem. Enjoying emacs now. All of this comes at a high price, however. I've been messing around with trying to use emacs for a few months now, on my side time at work and at home, and it's only been in the last couple of weeks that I've been able to use the package in a way in line with the available training materials. Normies have no kind of patience for this kind of ridiculousness. Techies justify these "rabbit holes" as enthralling and fun, and yeah, it feels good to finally figure some of this kind of stuff out, but overall, it's a massive distraction from actually USING the tools, rather than just futzing around, trying to get the damn tools to work in the first place.
- 7y ago
cr0sh 7y ago
Rabbit holes and their implications aside... As a software engineer, a part of me has died because of the hardcoded strings (in the pam_motd.c patch). Why aren't these defined as constants in the header file? Hopefully there's a good reason...
esotericn 7y ago
This article describes pretty much the entire reason I got in to software development, why I'm still doing it, why I'm here today, why I run Linux on my main machine, etc. I have an undergraduate degree in Physics for largely the same reason. I'm increasingly convinced that this is just a mindset you have or you don't. For me, the world is a place to tinker about in, learn, build stuff, break stuff, etc. I'm currently tinkering about with my homelab setup, tweaking the security, setting up redundancy, etc. Why? 'cos I can, and when you hire me you get all of that. It's certainly served me well so far!
cr0sh 7y ago
Most of us have fell down these rabbit holes of learning before; the author does a great job of illustrating the concept, showing the solving of an "ultimate question" about how/why something works the way it does. I don't know if it counts the same, but I once went on a weird journey to solve a particular coding problem that I was certain was already answered. I don't recall the details, but I do recall thinking to myself "This seems like a common problem; a quick google should bring up a solution..." - so, one google search later, and: https://xkcd.com/979/ https://xkcd.com/979/ At that point, I vowed to solve the problem - again - and post the solution (of course the fact that I can't recall what the problem was negates me being able to show you what I mean - my apologies - sigh). I quickly found that to solve the problem, I first had to learn how to code X, which in turn meant learning about Y and Z, and also coding A, B and C. I ended up going down these rabbit holes; by turns learning and coding partial solutions, by others finding example code I could adapt to my needs (not always in the language I was working with, either). I spent more than a few hours on this task, but I eventually did come up with a solution. And as I recall, I did post that solution to the thread where I encountered XKCD-979 - so future people wouldn't be inconvenienced.
squirrelicus 7y ago
> Unless you got into tech solely for the money (you monster), it's stuff like this that we nerds live for. I take offense to that! :P The rabbit holes of my past are half of why my paycheck is where it is. (The other half is interpersonal skills)

276 points by dailymorn 7y ago

Yeah this is just overkill. I haven’t edited /etc/motd in a while so this was news to me but this is just code creep and makes me sad.

hedora 7y ago
Same here. My reaction at each step is “how do you delete this subsystem?” and “what poor person wrote this?” Still, nice learning exercise for the author.
- pjc50 7y ago
  Exactly. I'm sure somebody wanted it at some point somewhere, but now it's on every system. No wonder people keep starting new minimal Linux distros.
edejong 7y ago
That. They could’ve just supplied a default .profile / .bashrc. It’s these kinds of bells and whistles that causes most security bugs due to additional attack surface. Is it the root user fetching the news with wget/curl from a remote source?
enriquto 7y ago
All of systemd is the same shitfuckery. I move away from debian to slackware and voidlinux (when using BSD is not possible) just to avoid it.

mustntmumble 7y ago

As one who simply tinkers with Linux but is not a guru, I found this article quite inspiring! I had no idea just how deep that sort of rabbit hole can go. I feel inspired to try to delve behind the scenes for myself... The author has made my day :-)

cle 7y ago

The trick is knowing how deep to go, which depends a lot on your goals. If you just want to learn for your own curiosity, rabbit hole forever! If you want to build useful, applicable knowledge, there’a a peak of usefulness after which there are diminishing returns. So being self-aware enough to know when to pull the plug on your rabbit hole is pretty important (professionally). Also certain kinds of rabbit holes are way more useful than others. If you’re rabbit holing into some obscure, arcane system, and the stuff you’re learning isn’t challenging you fundamentally, you’re probably wasting your time (I call these “rat holes”). If your rabbit hole will yield new fundamental insights into something important...now that’s a good rabbit hole!

Robotbeat 7y ago
The best depth before you hit the knee in the curve is undergraduate-level textbooks. Sometimes you have to go deeper, but that's the level that is presented in a form that's easy to grasp and can be quickly applied without using too arcane language and abstractions. Also: Handbooks. For instance, the Art of Electronics. And I recently picked up a mechanical engineering handbook, and the breadth of the topics covered at a level I can quickly understand is very refreshing. (I'm a physicist, but I mostly do mechanical engineering and technician work, with a bit of aerospace and electrical engineering and material science.) And combined with youtube and Wikipedia, you can learn a lot very quickly.
- tasogare 7y ago
  > The best depth before you hit the knee in the curve is undergraduate-level textbooks. That matches my experience as well. Having graduated from two fields at university, the most foundational knowledge came in the first two years. The third one is a deepening of the first two but can be skipped if only basis are needed. Then master degree is just playing with higher level concepts and linguo.
- voltagex_ 7y ago
  >Art of Electronics What would you suggest before The Art of Electronics?

HenryBemis 7y ago

Shortcuts is my personal mortal enemy. I talk to people that "really want to learn topic X" but: a) don't want to read any books, b) don't want to watch any training videos (from nuggets in the old days, to youtube tutorials now). They just want a quick and dirty solution to everything, always. That comes with a price. People want shortcuts. I don't like or use shortcuts 99% of the time. My experience is that when you go looking for X to serve a need today, you discover Y and Z and guess what.. Y and Z will come very handy tomorrow, when in a call/meeting about ABC, miraculously Y and Z are similar and you speak up a d impress everyone and get the job/contract.

kregasaurusrex 7y ago
Knowledge that compounds in such a way becomes useful in the long term. The Y and Z mentioned may be tangential in a current situation, yet as a technology stack evolves they can become integral parts to more complete and better fitting solution.
stevekemp 7y ago
I frequently get emails from people which follow a predictable pattern: "I'm doing a PhD in .. I saw your blog post .. can you point me at a tutorial for advanced-subject?" I often wonder why people don't do research, and what would happen if tutorials just disappeared. People would have to do their own research and learning. Maybe things would take longer, but I think the people would benefit from it.
- ackbar03 7y ago
  I'm not sure if I can speak for everyone but sometimes I do this in the hopes that it could lead to a connection with the author or potentially deeper insights and discussions on the topic with someone more knowledgeable. Usually I go off and try to figure out what it was I was asking in the first place anyways
  - stevekemp 7y ago
    I'm happy to answer questions, and sometimes I can point to non-obvious resources. But mostly I feel that the approach would/should be different for trying to discuss things than "Please help me do my homework"

sharadov 7y ago

I love this kind of stuff, it's the most pure kind of learning, no end goal but satisfying your innate OCD. I've gone down these rabbit holes in tech and alternative fields - politics, philosophy, cooking. It's crazy, but it lead me to some life long hobbies.

LogicX 7y ago

Or positioned another way, this is Yak Shaving: Any seemingly pointless activity which is actually necessary to solve a problem which solves a problem which, several levels of recursion later, solves the real problem you're working on. http://yak-shaving.urbanup.com/1164318 http://yak-shaving.urbanup.com/1164318

mkl 7y ago

The metaphor critique paragraph seems to be missing the fact that name comes from Alice in Wonderland, where the protagonist falls "through a rabbit hole into a fantasy world": https://en.wikipedia.org/wiki/Alice's_Adventures_in_Wonderland https://en.wikipedia.org/wiki/Alice's_Adventures_in_Wonderla... I.e. it makes complete sense in context.

acangiano 7y ago

Reminds me of how, as a kid, I tried to solve Fermat’s Last Theorem. I didn’t have a chance in hell, of course, but my attempts enabled me to learn a ton of advanced math and even finding on my own some results that turned out to be already well known (e.g., Euler’s formula). It's one of the purest forms of learning but as an adult, I no longer have the time or freedom to do it to the same extent. Still, the basic approach is worth keeping. You just need to limit yourself a little so that you don't end up too deep in the rabbit hole, spending a ton of time on things that you really don't need to know.

codingslave 7y ago
I had a similar approach with the following equation: What does the infinite series of 1/(k^2) converge to for k=1 to infinity. When I was in high school, I spent weeks going down this rabbit hole of mathematics, failing to solve it. Years later I discovered that this equation is a special case of the Riemann Zeta Function, which was also first studied by Euler. His proof is really elegant. The funny thing is that with some math equations, sometimes you intuitively can see that there is something important or far reaching about an equation, even if you arent skilled enough in mathematics to discover it. Some kind of curiosity draws you in

mkorfmann 7y ago

KARMA TANK INCOMING

simosx 7y ago

There was drama about the MOTD two years ago. Here is the summary from LWN, https://lwn.net/Articles/726902/ https://lwn.net/Articles/726902/

dusted 7y ago

So, blur the hostname in the SSH command, login to machine, where the hostname is ip-172.31.48.7.. Two things are funny here: First, the failed attempt, and second, the idea that anyone cares what the IP of your raspberry pi is.. If anyone cares to know, mine has 192.168.30.12.

reportgunner 7y ago
haha, perhaps there was some profanity that had to be blurred out

lordnacho 7y ago

This is the reason why people who don't respect coding as a discipline end up producing terrible code. In my professional life I've run into a lot of people who will say "the idea is already there, it just needs to be coded up". They will then come across common coding issues like keeping things in version control, adding tests, network config, and language specific things. They decide they don't have time for this particular rabbit hole, because of course they already know what they want. The good people will over time end up investigating everything, even if it doesn't seem relevant to your LOB application. And they produce much better code because of it.

la_barba 7y ago
Its not that people don't respect it, its that they don't appreciate it. And the reason they don't appreciate it is that those things are invisible to the end user. How much extra money would you pay for an electrician or a mechanic who does the equivalent of what you're suggesting with code? 1.5x? 2x? Most people I know always complain that their garage overcharges them or that they got ripped off by some contractor or the other...
jjeaff 7y ago
As someone coding for my own business, I often find myself deep down a rabbit hole and have to get out and force myself to just use that ugly patch. It's a balancing act, because it hurts me to just leave something without completing it "the right way" but I have found that if I try to do everything "the right way", I would get so little done. I try to keep this in mind when reviewing or working on other's code. Some things will anger me at first glance but the problem is, they were probably just prioritizing different things at the time than I would have prioritized right now.

pjc50 7y ago

I'm not sure about "technical expertise", as most of the stuff you learn while doing this is highly specific, but perhaps the real expertise is having the toolset and determination to do this at all. To keep taking screws out of the system and removing the covers until you find out where the system makes a particular decision. Fortunately in this case you can go to the source; if you can't, or as a first attempt, strace will usually tell you where a particular program gets its information from. Harder to use here as you'd need to setup the login process itself with strace and there's privilege boundary considerations. One observation I would make is how many layers of "plugin architecture" have been added here. Probably the original SysV login program just printed /etc/motd. Now we have PAM, and pam_motd itself also runs a bunch of tiny plugins, some of which also refer across to systemd. I should probably write up some of my own deep-dive debugging stories, if I can still remember enough detail. The "why does structured exception handling sometimes not work on Windows CE" was a particularly time consuming one which involved disassembling bits of Windows kernel.

jimnotgym 7y ago

> * Keen to learn Istio? It's included in the single-package MicroK8s. But this is a rabbit hole! His rabbit hole is about removing a rabbit hole.

Tepix 7y ago

The "right" solution to get a quiet login is to touch ~/.hushlogin in your home directory. It's documented in the login(1) man page. There's also a global /etc/hushlogins file you can create to take care of all users.

tombert 7y ago

Rabbit holes are why I grew to love Haskell and pure-functional programming in general. For better or worse, Haskell (Agda, Idris...) doesn't let you compromise very often. You could argue that this is terrible for productivity, and you might be right, but at the same time, means that a lot of people have done a lot of research on how to do things the "right" way in Haskell, and any project I've done with the language has led me to learning a whole slew of new things about type theory, category theory, and PL theory in general, which I can then take and apply to "productive" languages.

shanemhansen 7y ago
This aligns precisely with a memorable interview I gave to a PhD student who wanted to do an internship at a startup. "The thing I love about Haskell is that even the simplest things become research projects" However I understand where you are coming from. There's something to be said for environments that encourage learning.
- tombert 7y ago
  Every now and then I get mildly embarrassed because someone has managed to summarize my long-windedness to a beautiful and elegant sentence. The line "The thing I love about Haskell is that even the simplest things become research projects" has done that. I'm going to steal that.

drawkbox 7y ago

We use platforms and layers that are at the level of branches and leaves on trees, supported by abstracted away structures hiding complexity that make them simpler to use as a platform higher up the tree (or down). A researching traveler can go down the branch, to the trunk with lots of time rings and layers, and then further down there are many root systems and rabbit holes. All sorts of functionality is abstracted away in layers down there, all the way to machine code and binary and everything before the epoch 1/1/1970. Epic programming and technology pioneer legends created it, they took complexity and made a simple platform layer for others to build on.

peterwwillis 7y ago

It seems like the difference between a complicated, difficult system and a simple, easy system is whether it changes over time. I feel like a Linux kernel in a NetBSD userland might be the best sweet spot, OS-wise, but I haven't used NBSD in a long time.

prometheus76 7y ago

I'm going to go meta with my comment and say that this article is a fantastic object lesson as to why linux faces extremely low odds of ever being adapted as a widespread replacement for windows or OS X. How many people (outside of your coworkers) do you know who would or could follow a path similar to the one laid out in this article? I'm speaking as someone who has dabbled with linux since 1998. At that time, and now, there is a huge, confused ball of conflict between linux distributions, various levels of customization to those distributions, and documentation available on the web. Here's a quick, recent example from my own experience. I decided I wanted to learn more about Emacs. I work in Windows at work, so I downloaded portacle. Great. Works great. Then you start watching youtube videos on how to customize Emacs, but most youtube videos are using Emacs in Linux. So you then have to find where portacle puts all the stuff that is in a completely different place than all the Emacs documentation and videos say it is. Great. And I haven't found anyone who talks about the benefits/costs and gotchas about whether to edit the init.el file, or the .emacs file, but now we're getting into the weeds. So after that frustrating rabbit hole, I think, "I'll just install a linux distro in a VM." Next thought: "Which distro should I get?" Search engines are no help when it comes to finding the "best" linux distro because all I can find is superficial opinion pieces on the matter. So back when I worked at an ISP in the early 00s, they used Debian, so I download Debian. Try to install emacs in Debian, and I start getting weird errors about some certain package not being current enough, and apparently someone isn't maintaining/updating the install script for Debian. That's three days lost getting to that conclusion. So I wipe that VM and download Ubuntu, because it seems to be the most widely used distro, and seems like a good distro for "normies" like me. Install emacs, no problem. Can customize by following youtube videos, no problem. Enjoying emacs now. All of this comes at a high price, however. I've been messing around with trying to use emacs for a few months now, on my side time at work and at home, and it's only been in the last couple of weeks that I've been able to use the package in a way in line with the available training materials. Normies have no kind of patience for this kind of ridiculousness. Techies justify these "rabbit holes" as enthralling and fun, and yeah, it feels good to finally figure some of this kind of stuff out, but overall, it's a massive distraction from actually USING the tools, rather than just futzing around, trying to get the damn tools to work in the first place.

7y ago

cr0sh 7y ago

Rabbit holes and their implications aside... As a software engineer, a part of me has died because of the hardcoded strings (in the pam_motd.c patch). Why aren't these defined as constants in the header file? Hopefully there's a good reason...

esotericn 7y ago

This article describes pretty much the entire reason I got in to software development, why I'm still doing it, why I'm here today, why I run Linux on my main machine, etc. I have an undergraduate degree in Physics for largely the same reason. I'm increasingly convinced that this is just a mindset you have or you don't. For me, the world is a place to tinker about in, learn, build stuff, break stuff, etc. I'm currently tinkering about with my homelab setup, tweaking the security, setting up redundancy, etc. Why? 'cos I can, and when you hire me you get all of that. It's certainly served me well so far!

cr0sh 7y ago

Most of us have fell down these rabbit holes of learning before; the author does a great job of illustrating the concept, showing the solving of an "ultimate question" about how/why something works the way it does. I don't know if it counts the same, but I once went on a weird journey to solve a particular coding problem that I was certain was already answered. I don't recall the details, but I do recall thinking to myself "This seems like a common problem; a quick google should bring up a solution..." - so, one google search later, and: https://xkcd.com/979/ https://xkcd.com/979/ At that point, I vowed to solve the problem - again - and post the solution (of course the fact that I can't recall what the problem was negates me being able to show you what I mean - my apologies - sigh). I quickly found that to solve the problem, I first had to learn how to code X, which in turn meant learning about Y and Z, and also coding A, B and C. I ended up going down these rabbit holes; by turns learning and coding partial solutions, by others finding example code I could adapt to my needs (not always in the language I was working with, either). I spent more than a few hours on this task, but I eventually did come up with a solution. And as I recall, I did post that solution to the thread where I encountered XKCD-979 - so future people wouldn't be inconvenienced.

squirrelicus 7y ago

> Unless you got into tech solely for the money (you monster), it's stuff like this that we nerds live for. I take offense to that! :P The rabbit holes of my past are half of why my paycheck is where it is. (The other half is interpersonal skills)