The Fediverse has a global name registry. Your instance-name is part of that.
(That is, names are globally unique. They may not always be findable, but for a sufficiently federated profile, searching from a sufficiently federated instance, that shouldn't be a hurdle. And I strongly suspect third-party directories will emerge.)
The problem (other than disastrous security concerns stemming from server admins being able to read everyone's DMs) with Mastadon and similar is that once you move to another instance, your username changes.
With Farcaster, user names are both decentralized and global.
Having migrated a few times (both before and after official support for this existed), I'm aware of the Fediverse's non-portable namespace issues, and some way of resolving that would be useful.
The idea of not having your identity tied to one specific instance is very appealing.
Is there some way to create an account on desktop or web? Or at least view content without an account? It seems to only push some smartphone app, which is a big step down from Mastodon if that's the only option.
I had a look through frontends. Plenty let you sign in/connect some existing account, but I saw none that offered the ability to create an account. The Farcaster page on clients[0] seemingly reflects this by saying you use that app to create an account and then use an alternate frontend "once you've created an account". Given the 2.0 star average due to reviews claiming that the app requires a $12 payment then often doesn't work[1], I'd be hesitant to use it even if it was available on desktop/web or compatible with my phone.
If there's something I'm missing and there is somewhere to create an account, would you be able to link me in the right direction?
While I'm glad that I can read what someone is talking about on X/Twitter without giving away my privacy, I'd much rather see a trend of moving off of the platform altogether (or at least posting to multiple platforms to make the transition off of X more mainstream palatable)
I know this is a joke, but I remember when I was learning LISP I was SOO WORRIED about that and couldn't possibly fathom how I was going to actually be productive with it.
I don't remember now, but it wasn't AT ALL the big issue that I initially anticipated. I just mention it, because your comment made me realize that I also don't understand WHY it wasn't an issue!
> it wasn't AT ALL the big issue that I initially anticipated
LMK if you figure out why! Because whenever I'm refactoring in a bracket-heavy language, it's almost guaranteed that, every so often, the parser starts throwing syntax errors, and I need to stare at something like `)]})})` for a while.
What tooling am I missing that can prevent this from happening? I use auto-formatters and rainbow bracket colorizers whenever possible, and they're great! But in no way prevent this from happening.
It opens the gates of hell and corruption! Since every parenthesis has its own & unique match, you need to find the one missing its other half, and add the correct one to the correct place.
Every parenthesis in Lisp world is unique like a snowflake, and they come in pairs.
Parenthetically, parenthetical entanglement is consistent with duality, so at least our theories are consistent at a grand scale and do not necessitate any unified field theory (at this point).
I restarted my phone because I thought my network was down blindly believing HN can never be down. Then I opened another website and realised this is one of those times that the website was actually down.
Back in the day (90:ies) a colleague at the ISP I was working at (small time dial up business) told me that the one site on the Internet you could count on always being up was playboy.com. Therefore, it was our standard "check if Internet is working" site. In the event that it was a DNS issue, it was also best to memorize its IP.
For some reason, knowing the IP number of playboy.com by heart was not a skill I bragged about a lot :P
These days it's 1.1. Easier both to remember and type than anything else.
I used to have a smartphone plan that throttled to 100kbps after I used my data. At that point, for the rest of the month I would usually live with it because HN and google maps still worked even when every other site wouldn't load.
Same here. It coincided with some MS Teams breakage at work, so I honestly thought the Internet is down - I've rebooted the work computer and asked my wife to check if the Internet is working for her before I realized it's just HN.
Yeah, my phone even said that it couldn't connect "because your iPhone is not connected to the internet", despite being connected just fine. I had to run a globalping to actually convince myself that it was even possible.
It could be impressive, but the website frequently goes down when big news hit.
I think they could move it to a more powerful machine and the remaining issues would be fixed. It wouldn't be as big of a flex as running on an ancient Intel CPU though.
Edit: I got "We're having some trouble serving your request. Sorry!" while trying to send this comment :)
Looking at that status page, unless it's completely wrong, there's outages multiple times every month. They really could just upgrade their instance and get more mileage out of it and it wouldn't require any "refactoring". Site's mostly fine as is
Amazing... yes, but there are also obvious drawbacks. Take a look at any big thread that has manual and clunky pagination on it, or the site being down because a single provider or disk crapped out. Totally acceptable and fine things depending on the scale of what you're trying to achieve, and especially so for a free website... but it can definitely be better (no, I'm not talking about microservices in Kubernetes on the Edge in the Cloud powered by AI).
... replaced the entire board with HN comments. From now on, all decisions will be made by ChatGPT5, posted on a regular "Who wants to lead OpenAI?" thread, and upvotes will decide which ones are implemented.
You're recycling the playbook of one of those crypto schemes. That's ancient history. Modernity asks for several LLM (Mistral, Llama 2, ChatGPT5) deciding which output is best and following that course. Typing parrots all the way down.
Are you doing the actual arc maintenance of HN? From your comment in that thread it seems like you do that + the moderation (how you have the time, I have no idea).
Do you come from a background of Lisps? I just learned my first lisp (Clojure) this past year.
Interesting. I've wanted to learn a Lisp since reading his essays almost 15 years ago, but never had a chance until this year.
Unfortunately it wasn't the best circumstances to learn Clojure - I was taking ownership of a codebase that had lost all its developers, and was fairly complicated. I wonder whether I should try to learn Arc for fun sometime.
I think Arc is fun! But the community is small and only intermittently active (to put it mildly): http://arclanguage.org/forum.
If I were you I'd probably take up Racket or Common Lisp, depending on your tastes. If you prefer PLT style hacking, I'd go with Racket and if you prefer more systems style hacking, I'd go with SBCL.
Also, if you like Emacs, I'd definitely go with Slime + SBCL. There is also a lot of newer-generation Emacs/SBCL tooling that I haven't gotten around to yet.
Arc is closer semantically to Common Lisp than it is to Racket - for example the way Arc uses T and NIL is very CL, not Scheme. Personally I find Arc more usable than CL for a bunch of reasons, such as (1) hashtables are far easier, and (2) it's a Lisp-1 rather than a Lisp-2.
Well, this made me realize that downforeveryoneorjustme.com is useless, because HN was down from multiple servers, but dfeojm happily claimed it was up, only later changed it to down, not sure when, but at least 10 minutes after it started.
Though, for pure power, I prefer https://jsdelivr.com/globalping to do an HTTP ping, easily check from like 100 different locations worldwide live. It also does a great job at helping you know if it's just slow in your area or worldwide or whatever.
I use HackerNews to check if my internet is OK. I'm sure even with this incident it has better uptime than Github, Gitlab and quite a few other services from major cloud providers / companies.
I keep seeing these comments, but in my experience hacker news has minor outages several times per year. That’s not counting massive slowdowns from big stories.
HN is a modern scalable multi-AZ, multi-region, multi-cloud Kubernetes edge worker Docker serverless lambda IoT paradigm SPA web 2.0 blockchain AI architecture (OCI) and certainly doesn't get those kind of reliability numbers from being a carefully watched server under dang's desk.
HN, StackOverflow, Hey.com/Basecamp, Lichess and probably many others that I'm not remembering or aren't famous run mostly or exclusively on real iron.
We don't need buzzword bingo to build reliable, internet-scalable systems, especially when a hiccup occurs so rarely that it's news when it does happen.
I've noticed this happening more frequently. It really ramped up about a month ago with the whole Sam Altman leaving Open AI situation and the articles that spawned. I wonder - is HN going to upgrade its infrastructure because of these outages?
Initially, it seemed like a DoS to us too, but it was not. This was confirmed by upstream provider metrics. No major traffic spikes. It was a combination of non-malicious things. More info later, some of us need sleep.
I won't be surprised if that happened in my country's internet. My country one day blocked Reddit out of nowhere, without clear reason, almost a decade ago. It's still not clear why, to this day.
I wouldn't be using Hackernews if it wasn't for the community and moderation.
This website is terribly bad:
- UX/UI is terrible and not user friendly
- Infrastructure and the code is like COBOL from the 80s
- No notifications when someone replies to your comment
- Profile section is non-existent
- API is just as bad as the website
I could go on with the list, but you get the idea... for a website which receives millions of visitors, it's a shame that it's so bad.
I think many of the things you identify as flaws are actually strengths of this site.
The minimalist text-only gui IS very usable for discussion, and forces that the focus is on the discussion.
I think for instance adding images to posts or comments would make the site worse, as images serve to attract and focus attention away from discussion. But the discussion here is the point!
The minimal profile sections are also a feature. This account is not your digital homepage. It is a place people go for discussion. A bigger focus on your profile would be a step towards use of the site as brand-building which favors attention-whoring over quality discussion.
Again, the focus is on reducing frictions to people to contribute to quality discussion, and avoiding dynamics that might supplant this.
Most of the poor discussions I see on here are when one user tries to steamroll a thread, replying to most every single comment, doubling down on their own points of the discussion. It turns into a personal argument, not a discussion by the community.
So I think a lack of notifications is a great feature. It lets us say our piece and move on, and let the community discussion fall where it will.
Totally agree with this. I thought it was weird when I first joined, but over the years, I've grown to appreciate the lack of it.
Keeping track of your points is a decent indicator that activity has happened though. I just remember roughly where it was when I left, and then I know whether I should check /threads or not.
Or just moving the collapse button next to the vote buttons, so collapsing multiple comment branches becomes easier. Would not change anything the GP mentioned and would probably be super easy to implement.
Easy enough to do it yourself via userstyle (for Stylus) or userscript (for TamperMonkey)? :).
I'm currently using both - userstyle for DIY dark mode, userscript on mobile for font size adjustment (no Stylus available on Firefox for Android yet). Plus uBlock Origin configured to nuke the karma counter.
And though I don't have a relocated collapse control, there are modified light/dark mode HN styles linked in my HN profile, for those interested in either using those directly or as a starting point for your own restyle.
With that logic, why not just provide an API. Everybody just hack together their own client. Being a hacker doesn't mean you never get to benefit from prior work, or never work together on something, or never share your work.
I get the appeal of providing a minimal service like HN, but unless I'm missing something, my proposed change is an absolute no-brainer considering it's effort-reward ratio.
The thought's occurred to me that among the conspicuous omissions from the HTML spec is of a threaded discussion (or more specifically, messages within a threaded context) as a first-order object, permitting arbitrary structuring of that content at the volition of the reader.
Though on reflection: if you're really interested in a technical discussion site, the speedbump of needing to code (or at least install) your own client might make for an interesting form of gating.
To that extent, your suggestion's somewhat growing on me.
This is one of the features of HN that I appreciate most. The lack of notification means that you don't (typically) get a ton of back and forth sniping arguments like you do on sites like reddit.
I find this makes people put more effort into their first comment and keeps the arguments lower.
You get the list of trending articles, you can go read the article itself, you can look at the discussions, and you can also get a comment box to add to those discussions. Oh, and you can up- and downvote articles and comments.
What else do you really need, honestly? It's like reddit, only with actually readable comment threads.
Personally, I'd like some more formatting options:
1. inline code formatting, e.g. Markdown's `place code here` with backticks)
2. direct links, or some sort of footnotes system. I prefer to thoroughly source posts, especially when it's either rare/obscure knowledge or when it's an opinion/viewpoint that goes against the "hive mind" or however else you want to call the "mass opinion" on here - it's easier and more productive to debate on sourced facts than on unsubstantiated claims.
3. an actual quote indicator, just as Reddit does with lines that begin with > - it would be a waaay better way to follow visually in lengthy posts
4. actually working ordered and unordered lists. Markdown's syntax is bad, but anything is an improvement over nothing.
One of the reasons I learned a lot of css and js was actually inspired by me writing userscript to modify HN webpage to be like I want. That might count as a plus for me. The HN page source makes it easy to customize.
And lack of notifications is a good feature. It does help in many cases where you would go into very heated or not very useful discussions.
If only they could raise a seed round and hire a decent team to add some animation on the frontpage and set it up on kubernetes then blog about their unique experience...
I mean, Reddit may suit your needs? I find HN pretty perfect for my needs. The simplicity keeps me coming back. I find all the things you mention to be benefits, personally.
I completely agree. I was just saying for the person I replied to that maybe Reddit was more their thing, which is fine. I still use Reddit for some things. It can be a really great resource at times. But I much prefer to spend my time on HN, overall.
And yes, the community and moderation here are a literal breath of fresh air compared to most of Reddit.
