Shallow learning has its uses - there’s no need to avoid it unless it’s displacing real comprehension (which you won’t get from a link aggregator or discussion board).
Mainly by checking up on something if it’s that interesting. Something about an Emacs trick, something about parsing, something about type systems, so on and so forth. Besides that, reading the comments. Other peoples’ two cents on something tends to get me thinking about the benefits or drawbacks of whatever the article had. Heck, maybe even learn something new there.
Shallow learning is actually usually my goal. I’m not working with WebAssembly at my job…yet. So I don’t want to spend my time getting deeply familiar with it. But I do want a shallow understanding of what it’s for and when I should start deep learning it (if ever).
When I actually need to learn something, setting a small hello world project where I can explore the technology is my next step. From there I search for my questions as they come up.
Lobste.rs fills this need nicely for me. It has a way better signal to noise ratio than Reddit or HackerNews. It’s the only tech site I regularly visit even when I’m outside work hours.
I generally look for technologies that are conceptually different than what I’ve seen before. So having learned Ruby, I didn’t bother looking into Python too deeply because they were conceptually the same. Just knowing it existed was good enough (for web app dev). But I picked up Python, since I was doing conceptually different things in machine learning.
However, when React came along, it was conceptually different than Backbone, jQuery, Angular, so it warranted a closer look. I’ve found that looking for conceptually novel ideas behind the technology works as a pretty good indicator to look deeper.
I examine the premises of the article and what it claims. I inspect those claims to see if I can verify them. I test their durability by mapping them to other domains of reasoning, or building new scaffolding on top of their mid-level results.
I also try to make commitments to fully read papers if they are sufficiently interesting. I take the concepts described in the papers, and I imagine them in as much detail as I can.
Anyone have a good way of filtering Lobsters posts by popularity/comments/upvotes through RSS?
I like Lobsters more than HN, but I miss the abliity to filter posts (e.g. only posts with 10+ upvotes).
Coupling Lobsters mailing list feature with an advanced MUA (or sieve)
you can at least filter for topics you’re interested and ignore
everything else.
Hot take: I don’t. Conditionals are conditionals, loops are loops, recursion is recursion. Running a binary is still the same in AWS, Docker, or k8s, just with more fanfare. etc. etc. etc. A lot of the fundamentals have been the same for a very long time; it’s just the plumbing and fanfare surrounding it that changes. I think it makes little sense investing time in the fanfare unless there’s a specific reason to.
I’ll go meta on you and question why it is important to keep up with the latest tech trends. Followed to its logical conclusion, this results in your career to being defined by which way the winds blow on Hacker News. Do you want that?
I certainly don’t. I’ll take the harder, lonelier path to avoid it.
Trends in tech seem to mainly be driven by marketing, marketing is driven by money, money is driven by businesses.
I’d rather keep up with delivering value and getting paid than work on becoming an embedded salesman for some YC company.
(more concretely: learn systems, learn problem domains, go work at places that need that, pick up the tech they use, switch gigs and learn a new stack if you think you need to)
I find my ACM membership and the associated learning library to be very helpful. There are RSS feeds available for certain subject areas so you can keep an eye on papers as they are published.
I listen to quite a few podcasts most of which are interviews with creators, founders, developers of various tech. Not only is it useful for learning about about new things, but the interview style dialogue provides a lot more foundation and motivation between the tech. Contrast this with most product or tech websites which can be difficult to assess the utility.
Addressing the “shallow learning” question, if something seems relevant to any of my work I tend to build small prototypes in a domain I am working on to assess its utility.
A recent embedded episode I recall was on GoTime called “Hardware hacking with TinyGo and Gopherbot” https://changelog.com/gotime/84. I am not in this space so I am not aware any podcasts dedicated to this topic (although I am sure there are a couple at least).
I keep an eye on Lobsters and HN. But, I also don’t worry keeping up with all the latest tech trends right now. Instead, I try to work on fundamentals, like managing my attention, understanding the tech I’m working with at $JOB well, and learning longer-lasting ideas.
I try to learn longer lasting, or unusual technologies on the side. I haven’t had to score a lot of points with recruiters in the past, but developer co-workers seem pretty consistently impressed with PISC, for example.
One other thing that might affect things a bit: I’ve learned enough new/different frameworks and types of programs that picking up a new way to do something isn’t a very daunting task (it could still be time consuming, that’s a different axis). So if all of the jobs are using a New JS framework, and that’s what I need to learn to get a job, then I have the internal toolset I need to spin up on it over a few months and go in that direction. If I need to dive into Pascal for Inno Setup, that isn’t so crazy. After building PISC, and trying to understand Factor code, most other code seems sane by comparison.
Annual “State of X” surveys can be helpful. I just read one about front-end dev that’s on this very site right now and I learned about the existence of some libraries and technologies I was unaware of. Very shallow learning, but some of those I will probably come back and investigate later.
When I see a question like this I have to ask: In general, or in a specific language/community/specialty?
I’m largely a Python guy (Yes I know I can hear the legions of C/C++ ascetics groaning in the peanut gallery :) doing systems engineering / infrastructure as code work, and so I keep pretty up to date in those areas.
And of course, Lobsters :) But the thing I value most about this place is not the daily grind of new tech but the brilliant out of the way things that people post here like the various BSD goings on and amazing hacks :)
Lobsters and HN are pretty good for seeing very high level trends. They’re both a little biased in the type of stuff that will be posted (HN skews heavily towards web tech, for example), but they’re generally not too bad.
For more specific things, like a specific language or library, I subscribe to RSS feeds for people or organizations involved with them.
There’s this one site called Lobsters. Good community, seems pretty up-to-date when it comes to trends. Would definitely recommend.
Yup, pretty good…
…but what do you do to prevent “shallow learning”?
ie. Just skimming a interesting article without going into depth of why and how and without doing it for real yourself…
Shallow learning has its uses - there’s no need to avoid it unless it’s displacing real comprehension (which you won’t get from a link aggregator or discussion board).
The older I get the more I regret shallow learning and value the bits of deep learning I have done…
In retrospect it seldom seems to have been a good decision.
In the flow of tasty tidbits in places like lobste.rs it easy to stay shallow.
One way I’m trying go deep is actively search for a relatively limited set of topics I’m trying to go deep on… read those and engage and comment….
If I make a fool of myself, someone who does know will tell me I’m an idiot and I can ask them what I need to read to be less of an idiot…
….in this way I slowly get deeper and deeper into the topic.
Mainly by checking up on something if it’s that interesting. Something about an Emacs trick, something about parsing, something about type systems, so on and so forth. Besides that, reading the comments. Other peoples’ two cents on something tends to get me thinking about the benefits or drawbacks of whatever the article had. Heck, maybe even learn something new there.
Shallow learning is actually usually my goal. I’m not working with WebAssembly at my job…yet. So I don’t want to spend my time getting deeply familiar with it. But I do want a shallow understanding of what it’s for and when I should start deep learning it (if ever).
When I actually need to learn something, setting a small hello world project where I can explore the technology is my next step. From there I search for my questions as they come up.
Lobste.rs fills this need nicely for me. It has a way better signal to noise ratio than Reddit or HackerNews. It’s the only tech site I regularly visit even when I’m outside work hours.
I generally look for technologies that are conceptually different than what I’ve seen before. So having learned Ruby, I didn’t bother looking into Python too deeply because they were conceptually the same. Just knowing it existed was good enough (for web app dev). But I picked up Python, since I was doing conceptually different things in machine learning.
However, when React came along, it was conceptually different than Backbone, jQuery, Angular, so it warranted a closer look. I’ve found that looking for conceptually novel ideas behind the technology works as a pretty good indicator to look deeper.
Mainly by taking on the technologies that interest me on a deep level, and trying to build something using them.
I’ve done this for stack based programming languages, Nim, and so on.
Deep learning is a lot harder than shallow learning, I find, however, so it happens outside of work less.
I assume that the article is wrong.
I examine the premises of the article and what it claims. I inspect those claims to see if I can verify them. I test their durability by mapping them to other domains of reasoning, or building new scaffolding on top of their mid-level results.
I also try to make commitments to fully read papers if they are sufficiently interesting. I take the concepts described in the papers, and I imagine them in as much detail as I can.
Hmm. This is a pretty good answer… https://lobste.rs/s/55rese/leveling_up_skill_6_commit_tracing
Anyone have a good way of filtering Lobsters posts by popularity/comments/upvotes through RSS? I like Lobsters more than HN, but I miss the abliity to filter posts (e.g. only posts with 10+ upvotes).
Coupling Lobsters mailing list feature with an advanced MUA (or sieve) you can at least filter for topics you’re interested and ignore everything else.
Hot take: I don’t. Conditionals are conditionals, loops are loops, recursion is recursion. Running a binary is still the same in AWS, Docker, or k8s, just with more fanfare. etc. etc. etc. A lot of the fundamentals have been the same for a very long time; it’s just the plumbing and fanfare surrounding it that changes. I think it makes little sense investing time in the fanfare unless there’s a specific reason to.
I do want to watch the fanfare, but from a distance, on my own terms. That way I know the correct words when telling the kids to stay off my lawn.
I’ll go meta on you and question why it is important to keep up with the latest tech trends. Followed to its logical conclusion, this results in your career to being defined by which way the winds blow on Hacker News. Do you want that?
I certainly don’t. I’ll take the harder, lonelier path to avoid it.
I do periodic literature surveys. I recently completed one on hyperparameter optimization.
Trends in tech seem to mainly be driven by marketing, marketing is driven by money, money is driven by businesses.
I’d rather keep up with delivering value and getting paid than work on becoming an embedded salesman for some YC company.
(more concretely: learn systems, learn problem domains, go work at places that need that, pick up the tech they use, switch gigs and learn a new stack if you think you need to)
Interesting points made
I find my ACM membership and the associated learning library to be very helpful. There are RSS feeds available for certain subject areas so you can keep an eye on papers as they are published.
I listen to quite a few podcasts most of which are interviews with creators, founders, developers of various tech. Not only is it useful for learning about about new things, but the interview style dialogue provides a lot more foundation and motivation between the tech. Contrast this with most product or tech websites which can be difficult to assess the utility.
Addressing the “shallow learning” question, if something seems relevant to any of my work I tend to build small prototypes in a domain I am working on to assess its utility.
Any podcasts you’d recommend to someone interested in compilers / embedded?
A recent compiler episode which I enjoyed was from the CoRecursive podcast: https://corecursive.com/037-thorsten-ball-compilers/ (there are many good episodes on that podcast).
A recent embedded episode I recall was on GoTime called “Hardware hacking with TinyGo and Gopherbot” https://changelog.com/gotime/84. I am not in this space so I am not aware any podcasts dedicated to this topic (although I am sure there are a couple at least).
I keep an eye on Lobsters and HN. But, I also don’t worry keeping up with all the latest tech trends right now. Instead, I try to work on fundamentals, like managing my attention, understanding the tech I’m working with at $JOB well, and learning longer-lasting ideas.
I try to learn longer lasting, or unusual technologies on the side. I haven’t had to score a lot of points with recruiters in the past, but developer co-workers seem pretty consistently impressed with PISC, for example.
One other thing that might affect things a bit: I’ve learned enough new/different frameworks and types of programs that picking up a new way to do something isn’t a very daunting task (it could still be time consuming, that’s a different axis). So if all of the jobs are using a New JS framework, and that’s what I need to learn to get a job, then I have the internal toolset I need to spin up on it over a few months and go in that direction. If I need to dive into Pascal for Inno Setup, that isn’t so crazy. After building PISC, and trying to understand Factor code, most other code seems sane by comparison.
Annual “State of X” surveys can be helpful. I just read one about front-end dev that’s on this very site right now and I learned about the existence of some libraries and technologies I was unaware of. Very shallow learning, but some of those I will probably come back and investigate later.
When I see a question like this I have to ask: In general, or in a specific language/community/specialty?
I’m largely a Python guy (Yes I know I can hear the legions of C/C++ ascetics groaning in the peanut gallery :) doing systems engineering / infrastructure as code work, and so I keep pretty up to date in those areas.
Specific resources I use:
Python Bytes Podcast
Python Weekly
Python Reddit
Devops Weekly
And of course, Lobsters :) But the thing I value most about this place is not the daily grind of new tech but the brilliant out of the way things that people post here like the various BSD goings on and amazing hacks :)
Lobsters and HN are pretty good for seeing very high level trends. They’re both a little biased in the type of stuff that will be posted (HN skews heavily towards web tech, for example), but they’re generally not too bad.
For more specific things, like a specific language or library, I subscribe to RSS feeds for people or organizations involved with them.