This Month it was 5 articles (that I remember I read) 4 smaller ones and one big one.
Doom over DNS
Why?
Because it’s just funny, I don’t know anymore how I found it, But just seining DOOM run on anywhere is just interesting.
Description
The Blog Post about how to Store and running shareware DOOM entirely from DNS records.
My Thoughts
This was intriguing, I didn’t know that DNS records just accept any text as a TXT record. I just used a DNS record for this Website to connect the Domain to my deployment, so I knew a little about what I could do here. I thought it was funny that he first tried to store Shrek in the DNS record and then tried to store AND run DOOM, which, I thought, was harder. That being said, after reading, it was the amount of data that was the issue. For me, the next idea would be Bad Apple, but Doom also works, so if you run either of them, it’s a good test.
Well, someone did it,
https://github.com/nicewrld/badapplens
https://www.youtube.com/watch?v=AJ2Q12vYojY
I thought it was more than just storing, but it’s just a boring protocol that does its job. You can use it for anything you want.
How to attract AI bots to your open-source project
https://nesbitt.io/2026/03/21/how-to-attract-ai-bots-to-your-open-source-project.html
Why?
I saw Prime’s video on it and read it for myself.
Description
A blog where the title does a perfect job of summarising it, but with a little more nuance. This is a satire, so it has 12 points you shouldn’t follow to get the least amount of AI bots, from simple ones to ones that surprised me. It’s like an open-source antipattern against AI.
My Thoughts
“Write Vague issue” and “Add a CONTRIBUTING.md that welcomes AI” are obvious ones.
However, the point “Remove type annotations and tests”
Surprised me, I thought Test would help it what to exactly and so bring more but no. The Author stands that with big tests coverage and a fully typed codebase, it knows so much that it knows that there’s nothing to add to codebase, when only looking at the Codebase. Other ones about in end about how Vague different stuff could be ann AI friendly stuff. Still, one last thing to note about the “Include a node modules directory” point. The writer says that the AI finds it more interesting because there is more code and room for it to do things. So the idea of getting PRs for things that come in packages seems silly to me.
# A comment system on top of ATProto and Bluesky
https://snorre.io/blog/2023-08-19-atproto-bluesky-comment-system/
Why?
Wanted to it myself and looked it up if someone else already did it.
Description
A blog where its Title does a Prefect Jobs of explaining it. Told his journey why he did it and then went into technical detail how it was made.
My Thoughts
I looked for it to see if anyone had already done it, so I wouldn’t have need to do, need to reinvent the wheel. Sure, someone had already done it, but he had specified a Sold JS component in his Astro website. So it will definitely be helpful as start point and reference, but ill make it myself but also more agnostic and compatible with everything JS-related. I wouldn’t really change anything because his goals were similar to mine. But I will add my own implementation of an AT protocol for comments only, so it will work like Disqus but made with the AT protocol. But I will make it self-hostable as well if they want it, in addition to just using Bluesky as a account.
So I suppose I’ll improve the Wheel in my own way.
Package Managers are Evil
https://www.gingerbill.org/article/2025/09/08/package-managers-are-evil/
Why?
I was going through the Odin creator blog posts and this caught my interest
Description
A Blog where Bill argues that packages manager are evil because its spread unverified code, and it would better if that you should directly but the package in your Codebase.
My Thoughts
I really liked it; however, I’m unsure if I would so if myself and no longer use them. I would as a Full-stack dev it would be difficult because that’s how Frameworks are built and that’s how my works functions, and it’s like so much work reading through all the packages I would use. It would certainly help choose only the really necessary stuff, But then updated the package would be such pain just too much work, but my Program would have a way smaller chance of having a supply chain attack. But he writes in the sentence in the brackets the flowing (for programming languages) and on that, I cannot argue I never made a language. However, I would say on that well you should write your implementation of that, it’s foundational enough to warrant to do that.
A Social Filesystem
Why?
Saw it on Bluesky and thought that reminded me of Steph’s Blog (Obsidian creator) File over app and I wonder is it that but for Social media and gave it a shot.
Description
A look at the AT protocol(ATP) from a file system perspective.
My Thoughts
This is gonna a long one, 7k words article, unlike the other that were about 2k words. I’m gonna writing a lot about it because I find it fascinating. I’m going through each chapter each
Why Files Are Awesome
This part is saying the same thing as in File over app but explains it from different perspective and so used different examples. But I agree with that a lot, that’s why I also hate Notion with a Passion. Yeah, you can export the information as a CVS, but you lose so much in the Workflow, progress, and the automation a very walled garden. The move from Notion to Obsidian was very rough. Having a API between File and App is so great. So you have the Information alone and can decide how you can access and how it looks. Like I like how I can use my IDE for editing this Website and for Writing articles like this I use Obsidian. Another Metaphor I don’t like what notion does is. When I paint my painting, I want to use what paintbrush I want and not be liked locked into one because I started one painting with a paintbrush that only allows to use that one.
Or another Widespread example is the simple Mail Transfer Protocol (SMTP) it doesn’t matter what you use Gmail, Outlook, Yahoo Mail, iCloud Mail and Proton mail what I use. You can still communication with each other. But what about if you talk from Someone with Discord to Teams, okay, not the best example. Lets say Teams to someone to slack, that doesn’t work they can’t communicate with each other because they don’t use the same Protocol to send messages? They would need to use an open-source Protocol like SMTP and that exists that the Matrix Protocol, you can use any software that supported the same Standard.
In those, both Protocols, you can host your own Data or “Files” more than other programs I mentioned before.
But How would do that with social media? That’s the author wrote about that so.
The Everything Folder
The Idea of every like and follow is file seems so weird, when I first heard it didn’t make sense. But the more I thought about it, the more I liked but understand didn’t increase tho XD. However, I did think that would make then the social media apps just an Aggregator site and so make it so the only applies to a User is UI/UX if they come on the platform.
A Social Filesystem
Not much to say then, yeah there’s a lot on the ATmosphere and his Blog post looking at the AT protocol from a web-centric perspective is also an interesting read.
A Record & Record Keys
In this technical part, there isn’t much to write else than I agree with the choice, and it’s fascinating to see What is part of posts that belongs to you and what doesn’t. A Social media app is really something that people create together and not one platform.
Lexicons
This Problem here that a lexicon is solving here. I’m getting it 100% what problem it is solving, The blog explainer from another Bluesky dev helped but like not sure if I understand it, idk why, but it reminds of DTO’s a standardised way to share data, or post for that matter.
I just hope this doesn’t happen.
However, there’s no perfect standard after all, just your (or your company’s) perfect standard.
Collections
Here Lexicons made more sense for me, it’s like how a post on Social media is structured but like only the record’s part, your part of the Post as the author.
Collections are an essential part of AT protocol(ATP). That what makes sense way more versatile. Because I think that people believe that Bluesky is just a Twitter clone, I mean it is, don’t get me wrong. If the ATP didn’t have Collections, it would just be a twitter clone where you can own your Data and so access the posts in any way you want. What is cool, don’t get me wrong there either. But with them, it is much more.
Without collections, you could syndicate posts like from Twitter and Threads, but with collections, you can but anything on ATmosphere. You could do anything from Tumbler to Instagram to TikTok to YouTube based on it, just make a different collection.
There Is No Lexicon Police
Very cool, simple old concept still works. No Higher Authority, No force keeping it in check. Just technical doesn’t work. My favourite way of filtering stuff.
Links & Identity
Yeah, having like an array of likes to different posts makes sense. How do it with Identity?
Here comes my favourite part, the thing that made look into BlueSky and then ATP as why, well ill come to that later. Now to make Link that never break. It’s easy, just use the path of path from the collections, and then It’s no longer easy anymore. How do we do Identity? It can’t be the normal UID/ID from a central database that communicate of the URL, so it can’t be in BlueSkys case be bsky.app/[UID]/.. That would make the URL locked into their URL. That’s quite the problem.
The First three Attempts
Why reinvent the Wheel if already works? A Domain is unique, exists only once, and it makes you very independent to be on the internet compared to other ways on the internet, and It’s pretty cheap if you don’t buy like a .com TLD. However, interestingly, it isn’t Perfect, the Author said that losing domains is pretty common. I just have one and not for long, so I’m in no position to say if that correct or not. But I don’t see how a Protocol could resolve that Problem.
Attempt 4: Hash as Identity
This gets no way more technical, but when I understand it correctly, ATP also uses the Public-private Key system for verifying, or like signing a commit like the OpenPGP standard. So like with signing a commit and say self-verifiably this is me, that signed operation or that hash is that is your let’s say UID, you can always make sure that connect made post.
I might be saying BS but cool that you make so decentralised just by using Maths. cryptography is so cool but damn sometimes it doesn’t click XD But still not perfect for Author? damm.
Attempt 5: DID as Identity & at:// URI
Ah! Nice the DID is the UID I mentioned before, close enough. So the URL is now finished and will never break. It looks like a lot of gibberish. PERFECT. Even though seeing an ID in the URL is not my favourite(A blog explains it good here). Also, this isn’t really meant for the most people to read like a URL in HTTPS. I mean, I love Dissecting a URL and finding out how a Website is built, it’s just a path after all, so like backtracking that path.
at://did:plc:6wpkkitfdkgthatfvspcfmjo/com.twitter.post/34qye3wows2c5
└─P─┘└─────────── who ──────────────┘ └─ collection ─┘ └── record ─┘
So here we see an ATP URL is made of 4 parts. First: at:// Tells what Protocol to use, simple enough. Second: did:plc:6wpkkitfdkgthatfvspcfmjoGET http://localhost:1313/favicon-16x16.png Tells who is it, the DID what is simply a signed Handle and that handle is based on the given Domain.
Third: com.twitter.post
tells What is it? A Twitter post in that case
Fourth: 34qye3wows2c5
Tells the IDs or better the TIDs So it can get the Content.
Hyperlinks for JSON
Finally done. The post that could have been just one object has now been turned into 5 different kind of records. With links to everything. But I would say it’s way more social, its many records because many people created the post together, very social if i so myself.
A Repository
Finally, We got a name for the “everything folder”. Cool that the example for self-certifying is the same as I thought before with git signing.
Up in the Atmosphere
So cool how you can do something somewhere else, and then you can see the results somewhere else, definitely ones of the biggest reasons why I love coding. There are so many different projects, but they can use stuff from their own or just other Data and add to their stuff, just a cool side of the Internet web.
I want to be up there in the atmosphere or better said, my Project. The Disqus clone based on ATP.
An everything ecosystem lets everything get done.
What a great sentence, its cool what you can do with the ATP.