Hacker Newsnew | past | comments | ask | show | jobs | submit | ljm's commentslogin

If I was to drop acid and hallucinate an alien invasion, and then suddenly a xenomorph runs loose around the city while I’m tripping balls, does being right in that one instance mean the rest of my reality is also a hallucination?

Because it seems the point being made multiple times that a perceptual error isn’t a key component of hallucinating, the whole thing is instead just a convincing illusion that could theoretically apply to all perception, not just the psychoactively augmented kind.


If everything is a hallucination then nothing is a hallucination.

Thanks for listening to my Ted Talk.


… I think I’ll just stick with pydantic AI for now


This is quite a disappointing self-limitation given the improvements uv brings to the table. You’re missing out on some good stuff.


I don't think so. It'd be best to fix native tools.


And many of the conversions between metric and imperial align with the Fibonacci sequence on any order of magnitude. 130km/h is roughly 80mph simply because the fibo sequence has 8 and 13.

Obviously not an emergent property but shows how these things were designed.



I don’t think any common conversions fall near there other than miles->km. It’s certainly not the case that the systems were designed to have the golden ratio as conversions between the two.


1 mile = 1,000 [double] paces of 0.8m each = 1,600m

1m = 1e-10 times half-meridian from the North Pole to the equator, via Paris for a croissant, apparently.

So kind of a coincindence... But a very neat one. Meanwhile, ratio of adjacent Fibonacci numbers converves to some expression involving sqrt(5) which is approx 1.6


converges


My 'beef' with Cursor is that the editor is part of the package and you don't really have the same kind of hooks into the agent that you do with Claude Code or similar, which really means you're at the mercy of the Cursor team to prioritise those things on their roadmap. That includes things like the limit of 40 MCP tools that you can only enable globally (and MCP proxies that try to do this dynamically are a bit flakey) - even just using the GitHub MCP blows through that limit because it's all or nothing.

It's good for what it is but I don't love that I have to change to a whole-ass new editor to get access to the agentic additions, when alternatives that require an API key or a more flexible CLI tool can do a better job.


So Amazon becomes a supranational entity.

You should be ashamed of yourself for shilling for this shit. Curtis Yarvin would be proud.


I don’t know about this law specifically, but every other law attacking the internet or encryption has attempted to exempt people in government.

That defeats the point of the legislation since it creates a gaping wide backdoor to exploit official people, who are now the most valuable targets because of that exemption.

Never mind the matter of providing a rule for the people and making the people who made the rule immune to it.


Have they researched how many of these "age check" companies are actually run by Russian intelligence services?


Even having a single employee linked to the Russian intelligence services would be sufficient.


Only 20 years since it started as a hobby. There is programming that I enjoy doing for the fun of it or for experimentation and I wouldn’t use AI for that (most likely because it’d be something that isn’t well known or documented).

If work wants me to use it for the job, then sure why not? That too is something new to learn how to do well, will possibly be important for future career growth, and is exciting in a different way. If anything, I’ve got spare mental compute by the end of the week and might even have energy to do my hobbyist stuff.

Win win for me.


I on the other hand find agentic LLMs mentally draining.

I can't enter a flow state since the workflow boils down to waiting and then getting interrupted, and then waiting again. Often the LLM does the wrong thing and then instead of moving to implement another feature, I'm stuck in a loop where I'm trying to get it to fix poor decisions or errors.

It's possible I get a feature implemented faster thanks to agentic LLM, but the experience of overseeing and directing it is dreadful and pretty much invariably I end up with some sort of tech debt slop.

I much prefer the chat interfaces for incorporating LLMs into my workflow.


A lot of native apps are just wrappers around a JS context with a few bridges into native APIs and they are pure data grabs.

Reddit always asks you to use its native app, for example. Why the fuck would I care so much about Reddit that I want it outside of my browser? Same goes for any other website.


Reddit is one of the cases where a native app makes sense. Some of the 3rd party Reddit apps were great.

But I'll eat my hat before I'll install Reddit's own app. Reddit killing off 3rd party apps is why I post here and not there.


How does an app for Reddit make sense? It’s an image and text platform. There’s no weird hardware apis required.

Native apps make sense when you need to tap in to platform specific features like the Lidar api and such. They don’t make any sense for most websites.


The 3rd party Reddit apps made an effort to be more 'native', and actually used native UI elements to make rendering and interactions faster than the web page could.

WAAAAAY too often the 1st party native app is exactly what the other poster said: a browser context with access to some local native API's in order to hoover more data about the user. It is rare that a first-party app actually has some effort put into it to be a quality app. Is in fact so rare, that the sites that actually put in the effort suffer because folks can't believe that a native app for a site could actually be better or worth it.


I think the parent's point was that an app for reddit only makes sense because they deliberately don't add the features you like to the mobile site. There's no reason those features couldn't work perfectly well in a browser, they just choose not to (and to kill off third party apps).


If Figma runs perfectly well in a web browser, Reddit can do the same. It was built for and evolved almost entirely within the browser, like many other Internet forums. Pure data grab.


Figma shows what it is possible to do in a browser, but the cost of doing so is basically prohibitive. The level of persistence and technical nous needed to stand it up are on par with getting a first-person shooter running at an interactive frame rate on a 286 -- they basically reimplemented a browser within the browser.


I’ve worked on teams that made complex production web apps with a tiny tiny fraction of Reddit’s cash flow.


Reddit runs well IMO if you go to old.reddit.com. The mobile site is borderline useless, presumably intentionally.


According to Reddit's "Staff Platform Engineer (Web Platform Team)":

--- start quote ---

Old Reddit has the advantage of being pretty much static non-interactive content. No video, tiny thumbnails, and barely any JS or styling. Some people like this and some don't, but the end result is a very lean website that performs well out of the box.

https://x.com/jimsimon_/status/1841087335414280571

Suffice to say, I'm on the frontend perf team and we're acutely aware of these problems

https://x.com/jimsimon_/status/1841092341991403974

--- end quote ---

This was in October 2024.

Which is of course a bunch of bullshit when you consider that Reddit's backend returns most data in under 400ms, and it takes Reddit frontend 3+ seconds to render it

It could be that they are just incompetent.


Based on how universally horrible their video playback performance is (even in the iOS app), I can only assume incompetence.


Figma is sort of an Apollo Project among webshit, isn't it? IIRC they did rather extreme amount of R&D to make the webapp performant in spite of the web as a platform. Great that they did, and I hope their insights will keep trickling down to everyone else - but I don't think they're currently an example anyone can actually follow.


It’s great, but it’s not Apollo-level anything. Most games are far more interactively and visually complex. Apples to oranges with UX and interaction problems to solve, but certainly not depth and complexity. I’ve certainly experienced bugs in Reddit’s interface before — there seems to be this idea that they have to be so risk averse that they can’t do anything significant— I’ll bet you a pizza that their official app which implements all the features that people really want to use is made with JS/HTML on the back end anyway.


They’re not indifferent to browsers (less data mineable contexts) so much as actively hostile. For the past few years some things I have to add “-reddit” to my Google searches, because they killed i.reddit.com, which was the only useable, fast, non-complete-shit mobile site they have ever built. Their old. subdomain isn’t really readable in a cell phone.

Their new version is incredibly slow, moves me to sub-pages trying to expand comment threads (very disruptive if I saw something in the Google preview snippet and want to control F to it, but whatever comment that was literally isn’t loaded), and sometimes outright fails to load. now I can’t/wont use it.

So screw reddit, it’s a glorified q&a site, with sub forums run by fedora neckbeards, that’s gotten uppity and chosen to be hostile to users. And for some reason Google hasn’t just downranked it to death. The other day there was a thread complaining that their AI responses are reducing websites clicks. I hope that it is very damaging to reddit.


There are extensions that redirect to old.reddit.com with mobile friendly CSS.


> make rendering and interactions faster than the web page could.

McMaster-Carr begs to differ. Hell even old.reddit is pretty snappy (but deliberately shittily rendered on mobile). Websites can be fast if you don't stuff them with bullshit or degrade then on purpose to drive traffic to the app.


Right -- their website is a great example of a great web app. Their web site is brilliantly organized. But their revenue comes from sales of their products, not harvesting user data, so they have little need to add all the extra jank.

But if they had a native app (do they?) I imagine they would have the wherewithal to build the app natively, with the same stellar navigation of their website, and maybe some native-only features? Imagine if you could use the 3d sensor + camera of an iPhone, and point it at an assembly, and the app would identify the parts it could, and you could order with one click, or integrate with a local ERP or other systems...


Simply because native UI is faster and more functional and better integrated and better thought out than anything webdevs of any company can put together even if they cared to do it well. Sites like Reddit, or platforms like Slack or Discord, are perfect use cases for native clients, because there's a lot of space to make them better and more streamlined than the webapp.

Unfortunately, that only ever happens when some third party gets involved, and rarely survives long - but the experience, however brief, is glorious. See: RIF ("Reddit is Fun") on Android; Ripcord (Slack/Discord client) on Windows.


It’s a website for showing text and pictures. I know I’m being a bit reductive, but there is absolutely no reason why reddits functionality couldn’t be extremely performant in a browser. In fact the Reddit mobile site actually is very snappy, it’s just full of nagware trying to push you on to the app.


They would seamlessly in the background pre-cache all the articles and images coming up in your feed so if you had intermittent connections like on the subway, you could browse nearly[0] unaffected.

[0] Unfortunately, the app I used in the before-time did not implement queuing for submitting comments/posts so that functionality was broken while you were between stations, and videos weren't cached.


Web apps can do that too with service workers, it is just a bit of a pain in the ass.


And if you ever push a service worker with a bug, then you make the browser permanently unable to display the site unless the user knows how to manually remove the service worker. I've seen it happen on Gmail.


Yes it is a pain in the ass, but just for other people who read this, the bug was in the self-update code for the service worker. If you mess up the cache configuration it can make the service worker never look for an updated version of itself.

It is quite subtle thing in a very small part of the service-worker code but it is not like any random update can brick your app.


There isn't even a need for JavaScript for reddit though it does seem to require it. I posted this without JavaScript enabled so it obviously would be fine for reddit too. Using an app for reddit doesn't make any sense to me at all. Banking apps make sense, they are doing some crazy device finger-printing to avoid id theft. But when the goal is to convey information use html and css. If you are taking payments then yeah maybe some JS. If it is a game, try wasm. Apps are for things that need access to hardware that the browser doesn't allow, which these days is a short list.


Reddit is in full in AI data hoarding mode. Try to load their website with fingerprinting entirely mitigated and you'll be greeted by impossible are you a robot validation


>fingerprinting entirely mitigated

This is an interest of mine, but I’m still fingerprinted per a recent comment. May I ask:

How?


Chromium with autogenerate device resolution and client agent id with only simple JavaScript enabled(no rtc or gdi data) is enough to wall you from Reddit login script unless you're already validated by cookie


If you’re on a VPN it straight up blocks you with no captcha until you log in.


> a game, try wasm

On iOS? Waste of time for anything but simple 2d games.


The main reason they make sense is that no matter which version of real reddit you use it's got irritating behaviors. But a browser based better reddit wrapper could easily also make sense.


That is in context of Reddit being Reddit. It kept screwing with its mobile site for years (now it's FUBAR btw), so third-party apps were the only sane way to use it on mobile. Even Reddit’s official app used to be a decent third party app - Alien Blue. Then Reddit bought it and made it pathetic. That’s why people used third party reddit apps.

On desktop, the browser’s always been the best way to use Reddit — as long as old.reddit still works. If you are on a non-Safari browser, there's also RES.

Same goes for many other sites. Like HN — it’s fine on mobile browser unless I bump the font size, then it pretty much breaks. But I’m not installing an HN app for something the mobile usage time share is barely 5–10%.


I used vger.app (a frontend for a Reddit alternative) as a PWA for a few weeks. Then when the native Android app released I switched to it and it felt so much better to use. I can't tell you why, it was just more responsive.


We’ve forgotten what an “app” was


I too switched from Reddit to HN during the API protests of '23. But I always browsed through old.reddit anyway, I never used the third party apps. I'm aware of names like RIF and that everyone said they were great, but what was great about them?


Better features, less ads, smoother experience and in the case of Apollo—the one I used—it just looked much better.


So apart from the ad blocker, that's ... features, smoother, better. What?

Edit: I'm not trying to be rude (it comes naturally). But you just explained "great" as "better, with more". I guess smooth might mean faster, which might be because it isn't doing ads and tracking. It seems to come back to third-party being the crucial difference, and "app" not mattering.


UI/UX is not that tricky. Caring about your users is the hard part usually.

Third party clients could be webapps, too, of course.


Apps can bypass websecurity (CORS fetch), that allows for third party clients for example on video platforms using their internal APIs.

I don't think the reddit clients work this way though.


Caching and loading times mostly.

I'm on old.reddit.com too and I use the mobile app (including the 3rd party ones back when they existed) for one primary reason: Two windows I can quickly switch back and forth on. On my phone I use Reddit to look up things. I can have a Reddit thread on one window and a Google search on the other and go back and forth. In a browser switching tabs back and forth is painful, often reloading pages, losing the spot in the browser, having this url bar and top bar taking up tons of screen space.


I imagine it depends on how you use it. I came to Reddit late and never got into the old interface. I commented a lot on technical subreddits and didn't do much with the doomscrolling ones.

I used Boost. Its ads were not intrusive (and I despise ads) and the UI was written with a small touchscreen in mind. If not for my distaste for phone keyboards, I'd say it was a better experience than the website on a desktop.

Would it be possible for a mobile browser to have a better experience? I don't know. I value my sanity too much to do web development. But Reddit was absolutely determined to make its mobile site unusable and the official Reddit app had a bad reputation (and I wouldn't give those bastards the satisfaction after being nagged so much to install it), so a 3rd party app was the only reasonable solution.


I hold phones sideways, then I forget that it's weird to do. People like apps (and mobile versions of sites) because of human hand anatomy, I guess.


Unfortunately, it's getting harder and harder to hold your phone like this and see any useable amount of information. I'd love to have a 3:2 desktop monitor to replace my 16:10, but even 16:10 isn't available for phones. My current phone is 9:19, and I think I've seen 9:21. The wider the phone is, the larger percentage of precious screen space the system UI uses up when horizontal. I can see half a paragraph on my phone in landscape mode, and if it was any wider I'd never even try. As soon as I want to type something, I have to switch to portrait or I lose my view of what I'm looking at. There are terminal apps I simply cannot use because portrait mode isn't wide enough (unless I shrink the font so small I can't read it) and landscape mode I can't see anything at all. This affects some websites as well, due to their insistence on removing UI elements entirely when the device is too narrow (instead of moving them to the bottom) and others that set a ridiculously large minimum width. I far prefer tethering my phone to my tablet for this reason, which I'm lucky enough to have in 16:10 form factor.


That's because reddit on mobile browser sucks ass (feels like it's intentionally made to suck) even more so than its native app.

I don't think being nativr is what made 3rd party apps great


> Some of the 3rd party Reddit apps were great.

Because they were competently designed. But you could put that same design into a web page and it would work fine.


I use HN+Tildes instead. I left a couple years before the API fiasco because I was sick of the ragebait and toxic culture.


Redreader is still around as a good 3rd party app.


Some of the third party apps were quite good, certainly better than the reddit mobile site, but that's mostly because the reddit mobile site is just so deliberately awful.

There aren't really any major technical reasons why the mobile site couldn't be as good.


Yeah but if you want some restricted content the app is the only way


That was part of the 3rd party app purge. Before then, the 3rd party apps could see NSFW content just fine.


This is so funny. For me, it was as if the "monkey's paw" had played me.

Back in the early 2000s, I loved desktop applications. My thinking was that there's no way a web app could do what a desktop application could. I loathed slow, proprietary, online-requiring, HTML based web apps .

25 years have passed, and now we DO have some "native" device apps... but they are just HTML web elements bubdled in a freaking custom browser.

Edit: anyone remember the "PortableApps" wave? I loved having that in a usb drive.


You never experienced the horror that is XAML. Not HTML, not native control either, it’s a weird middle ground of platform lock-in that you couldn’t escape until recently.

What I miss are the days where one could Win32 call a window up, and it looked like every other. Not sugar for me and none for thee.

I cut my teeth programming GUIs, I still like making GUIs - immediate mode guis, event based guis, animated guis and informational guis. I left front-end web dev when every 6 months there was a new framework, a new new, and everyone dropped everything for it. I understand why React ate the world at the time but it’s gotten to the point where it’s no longer standards driven, its ecosystem driven, and even then it’s leaking.

What I love about these hybrid apps though is that from Apache Cordova (PhoneGap) onwards, they’ve all looked really really good. Proving that a normal user can’t tell the difference. Which makes solo-dev or small-dev dev easier. Go with what you know. No need to learn flutter, or SwiftUI, or Kotlin.


Svelte’s been pretty great in terms of “just use web standards where it makes sense” so far.


The most annoying thing is repeat questions ( reddit, linkedin, facebook, ... ). If I already told the site 10 times that I don't want to use the mobile app, stop asking me. That's even worse than cookie consent banners, at least those stay away


Speaking of apps that are just wrappers around websites – it’s possible to do that in just 50 kb: https://f-droid.org/packages/us.spotco.maps


Your comment got me a bit curious and so I spent time playing around creating a simple Android app using webView for my personal website and got it working, the only permission I added was INTERNET. So what's the next level of awfulness - do I add additional permissions and then additional information can be presented to my website server, or would I actually have to implement an additional path to collect the kind of info these apps are trying for?


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: