The benefits of learning in public
gilesthomas.com
gilesthomas.com
[1] https://www.evalapply.org/index.html#standing-invitation
I've done this for a few books, two video games, and a composer. One of the smaller indie studios wrote me back a nice email about how much it meant to them.
About two years ago I wanted to replay a game demo I remembered from an old (early 00') computer magazine demo disc. The actual game was nowhere to be found anywhere on the internet, but I did locate the developer and sent him an email.
Turns out the game was never actually released - its only public existence was as a demo game alongside a bunch of other games and software. He still had the installer for the game, which was small enough to fit as an attachment in the reply he gave.
He didn't say, but I got the distinct feeling that I might have made his day asking about that game he made more than 20 years ago.
Super niche, mostly irrelevant to all but a vanishingly small number of people, and yet I had proof that someone I admired found it useful.
It's been years, and it still makes me smile when I think of it.
You never know what impact you might make on others.
These signs of appreciation are, themselves, truly appreciated.
¹ https://librivox.org/ - if you'd like to hear him then try The Prisoner of Zenda: https://librivox.org/the-prisoner-of-zenda-by-anthony-hope/
I made it a habit of showing gratitude to content creators, as well as to open source maintainers.
This idea seems similar. I have some accounts on social media that is being used for read-only. That's unfortunate reality.
HN does not do reply notifications, so please send me a link - my contact is in my profile :)
I sent the author a quick thank you, explaining how it helped me in my hour of need. Exactly as others have said here, it goes a long way to making the effort of blogging worthwhile!
Total party skills, arguably the most balanced astute commentary of current events… definitely that I’ve come across. Hidden behind a game master facade.
Here's my most recent one, about using a Tailscale exit node to proxy scraping traffic from GitHub Actions: https://til.simonwillison.net/tailscale/tailscale-github-act...
> Simon Willison is the master of this and even has a subdomain devoted to his
[1] https://www.reddit.com/r/NixOS/comments/1dahr3g/steamos_base...
https://www.michael1e.com/how-to-replace-the-eufy-spaceview-...
Still don’t know how to respond when i get fucked by LLM authoritarians (Grok, ChatGPT, etc.).
They don’t give the traffic back / incentives or even cite us as the source. ¯\_(ツ)_/¯
So much of what I write has proven itself useful to me again even years after publication. Adding search functionality to my site and including my microblog posts there has extended this further.
But I've searched for things and found my own old blog posts on occasion.
I gave a talk about the "why" which turned into a long blog post: https://nicolasbouliane.com/projects/all-about-berlin
I really wish more people did this. I make it a point to show my gratitude when I encounter similar resources. One helped me cycle across Korea. Another helped me fix a very specific device.
It's a shame that Google, social media and AI strip mine the work of these helpful people and rob them of their reward.
Posting is actually delayed because I'm experimenting with how to do this more sustainably than I have in the past. Which means generating dollars somehow, and also making it harder for "AI" crawlers and services to rip off everything. :(
Could I ask for a link?
Recently I have been playing around with the Microchip PolarFire SoC[1] and as I already know, there ain't __that__ much resources available when it comes to FPGA design. But oh man, microchip is a whole other level than Xilinx/AMD or anything Lattice/open source tools related. I fought so many battles to get things done and I finally have some kind of a workflow going and understand the chip. I really like it by now (if anyone wants to dive into hardware-software-codesign, it's a nice budget chip to do it with). But after every battle I fight, I think to myself: "God damn, I should write this one down on a blog somewhere. If I can only help one person in the future, even if it's myself, then that's a huge win". But the lazy (and kinda scared) me won until now.
What if I don't write good enough (because english is not my mother tongue)? What if all I write is obvious to everyone else and I am the only one fighting? What if a friend will see this somehow and laugh at me?
It doesn't matter. I should do it, because I feel like it. Thank you for this post, OP.
- [1] https://www.microchip.com/en-us/development-tool/mpfs-disco-...
Edit: My last few words reminded me of CHarles Bukowski's "Roll the dice". Maybe I should remind myself of that a bit more often: https://hellopoetry.com/poem/68266/roll-the-dice/
For instance, the article on how to open a browser full-screen from IE5 was a roaring success. https://brajeshwar.com/2002/ie-50-full-screen-from-itself/
But then the new issue I have is I end up cluttering my blog with these lower effort posts that get in the way of the longer pieces. How do others solve this? Like separate feeds or something?
Here’s an example: https://varun.ch/posts/macos-keyboard/ I don’t think this should be among by “posts” but do others have a “firehose” section or something, for random thoughts and tips?
[1] I don't know what the best term is to describe theses kinds of posts that distinguishes them from long form blog posts. What do people do on Mastodon--toot? I'm just going with mini-post for now.
1. You will learn more, even when you think you know everything there is to know about the topic.
2. You can point to a thing you personally created. It's nice to be discussing a technical topic and be able to say, "Hey, I actually wrote an article about that, I'll send you the link."
3. AI might slurp up some of what you write, but it's a blurry filter over information. Your unique writing, details, and perspective are your own.
Finally, it can be fun or rewarding in the same way drawing, painting, or any other creative act can be. If it's not, then channel your effort into something else.
Wait long enough and there will be situations where you did something in the past that works really well, but you can't remember anymore what you did or why you did it. An AI doesn't really help you then.
If you want recognition, then writing tech articles is a difficult path to help you gain status or become known.
My guess is that you just dislike helping the enemy. I'm often more driven for negative reasons than I am for positive ones.
Not at all. That's like saying it feels pointless to write a blog if there's search. Like gwern says, lots of people are probably going to be looking for information through LLMs in the future. If you don't write what you're thinking, you'll likely be invisible or illegible to the future. Might as well write it down.
Anyone who has seen my HN history knows that I'm not averse to writing out large quantities of text spouting out my dumb opinions on things, and I don't think I'm a terrible writer or anything. I think at least some of my longer HN comments could be converted into blog posts with just a little effort, but for whatever reason it has never really stuck with me.
I guess part of it is that HN is a lot more conversational, or at least when I write a comment here, I'm writing it either a) in response to someone or b) with the expectation that someone might respond to me. For whatever reason, having that bit of social interaction (with the relatively technical and educated audience of HN) is enough to lower the barrier for me to write something.
I think if I already had a blog that was relatively popular, I'd also find it easier to get into the mood to write a more organized blog post, if for no other reason then there would be an automatic minimum-interest in whatever I write and then it would feel less like I'm shouting into the void.
This of course does not mean that I couldn't fairly easily edit them to provide such context and "blog-ify" them, but that requires a bit more effort that I generally want.
I wrote a post like this once - an article I wish someone else had written and posted. It has hundreds of views - a steady dozen or so a week. It's too bad I don't have more content like this where I figured out a thing which was largely undocumented.
There are tons of things i have (accumulated) to say, but something is stopping me. Probably the current not-easiness of it (plain html isn't that fun), and the need of investing yet-another-little-of-myself to make-that-easier (software? it's what i breathe), and some kind of tiredness of it..
There, i said it. Maybe one day.. i will.
have fun
This was one of my favorites: https://stackoverflow.com/questions/2757107/developing-for-a...
It was a checklist and people would just add on to the checklist as things changed.
Alas, it's not practical to do this anymore, because you'd just get downvoted to -13 under the current administration. Don't build atop other people's platforms, I guess.
1. First, I have classic essays that I wrote by hand. I'll ask an LLM for thematic flow or grammar checking, but it's my thoughts. It's more sporatic, but they're things I'm compelled to talk about and think deeply about. Example is on how visual programming is stuck on nodes-and-wires. https://interjectedfuture.com/visual-programming-is-stuck-on...
2. Next, I have posts that I write on a schedule. These are what I call lab notes, which I'll post every monday no matter what. These are easy to write because I just recount what I've been doing the past week and what challenges and wins there are. This exercises the muscle of posting something. Example is this past week on type checking https://interjectedfuture.com/lab-note-60-writing-words-and-...
3. I do write TILs, though these don't occur for me nearly as much at the moment, due to the type of work I'm doing. It's like my own stackoverflow, I guess. Example is on the common footgun of useEffect. https://interjectedfuture.com/today-i-learned/til-message-ha...
4. Lastly, I have posts where I had conversations with LLMs in depth on a niche topic that I think others would find interesting. If it's a back and forth, at the end of the conversation, I'll have it write a blog post based on what I thought the salient points were. If it's deep research, I'll just post that. I mark it clearly at the top it was LLM generated. The generation can be good enough that it's worth the post for humans to read. But also, in the hopes that the next model might pick it up, as an indication of what's good an interesting. It's basically one single sample eval based on my taste of what's good. Example is how algebraic effects are handled across Koka, Eff, OCaml, and Unison: https://interjectedfuture.com/algebraic-handler-lookup-in-ko...
All's to say to people, there's different modalities of blogging. If you pick just one, you might feel pressure to write when you have nothing to write. But if you don't write you don't get in the habit, which is why I do the lab notes. It makes me post even if I don't have anything to say, or the time to say it.
- Radar: just cool categorized articles, libraries, apps, etc. that I found during the week and organized in weekly drops.
- Guides: In my mind, those were closer to TIL than real guides, though I couldn't figure out how to name them. I love TIL for that.
- Playground: Forcing myself to share prototypes, snippets, etc.
I also have tons of LLM chats that could become content. Instead of trying to rewrite them, I will just share them as they are (while being transparent).
The main idea, as you said, is to find a modality that works well for me and force myself to write more.
Every time I follow this sort of advice, whatever I write inevitably becomes immensely popular and I end up hearing from a lot of people that whine loudly about my furry blog having furry art on it.
(Most recent example: https://news.ycombinator.com/item?id=43105421)
It's exhausting and if you don't reply on news sites, they'll send you social media messages and emails too.
And if you'd like a poignant example of a time I followed this sort of advice, I wrote this one after I failed to understand why some applications were using HKDF different than I intuited it was meant to be used, and a formal methods guy had to explain it to me because I was asking very loudly. It has since been cited by the Python cryptography library docs: https://soatok.blog/2021/11/17/understanding-hkdf/
If you're doing it for intrinsically valuable reasons (i.e., for yourself), wonderful!
But if you're doing it to try to overcome writer's block or establish themselves as a recognized name in a community, there are downsides that people don't tend to talk about much because they might sound ungrateful (especially if they financially benefit from their popularity). I, individually, have no such monetization incentives at play.
I'm free all w--