Aachen

Can confirm in the Netherlands also. I'm not an influencer so I don't know exactly, but a few years ago there was either a crackdown, an expansion of the rules, or at least a big media campaign by the government to make the influences that focus on a Dutch audience aware of the rules surrounding promoted content

I came away feeling like they acknowledged that throughout, save for the promotional bit at the end that I guess is there obligatorily

Very secure but not investing in rolling out this mtproto they went through so much trouble to invent, nor making it usable by implementing device sync or all the other features normal chats have. Why does it smell like fish all of a sudden?

It's not slow though so that's clearly not it. It's just a marketing intensifier here

Hah, literally seconds after the edit window ran out, I thought of an advantage of Signal's, edited it in, and HN rejected it due to post age. It is: Signal lets you add any emoji reaction you want without paying for premium

Calling/ringing works reliably. My gf usually calls via Telegram to notify me and then we call via Signal

Signal can't do live location sharing

    Code blocks (let alone syntax highlighting) was the first thing my friend ran into after switching last week

Many more image features, like you can do yellow highlighting and not only ugly shades of olive, you can put a message above an image instead of below, you can mark an image as spoiler, you can send it uncompressed, much simpler UI for selecting multiple images, etc.

Jumping to a date was one of the first things I ran into

Web client lets you temporarily use someone else's computer without needing to install software

Making a Telegram bot takes 10 minutes if you just want a simple http callback and know roughly how it works. Signal has no support at all. Some third parties have made libraries but we keep having issues at work with our internal Signal bot to the point where I think we just gave up altogether now

Many people use Telegram to check if internet works because when opening the app it does a lightning quick check and it's very robust (I think it has no problem if DNS is down, for example) Looking at the title bar confirming that it works is much faster than trying to do a random web search

You can have memorable/pretty group URLs like telegram.me/OpenStreetMapOrg

Large files can be sent (also because of server-side storage not clogging up your phone, which is a double-edged sword of course)

Scheduled messages send at the right time when you're not online & the desktop app supports scheduled messages

As of like next week you can do moderation in groups in Signal. This has of course been a thing in Telegram since 10+ years. I don't know how we'd have run the OpenStreetMap group on Telegram without being able to delete spam messages

Chats take a lot less space. The Signal authors probably don't have many contacts that they use Signal with because you can view the message markers of 4 chats before having to scroll down. In Telegram, 9 fit in view. Similar on desktop but both numbers are bigger. Messages are also more space efficient in Telegram

Telegram so so so much faster than Signal's web, ahem, electron client

The only advantage of Signal that comes to mind is the intangible property of privacy (and maybe local back-ups, so data availability, but then Signal doesn't provide software to view your backup data so is it really a backup if you can't test it and they can lock you out of the app?). I'm nevertheless switching everything over to Signal but it's often very hard to explain to people why they should give all this up for encryption. My family doesn't use all the advanced features but some techy friends do

No. They don't have the key and can't do encryption in the first place; they're still grand autocomplete engines under the hood. This could only work if the company deliberately builds a mechanism into the backend which runs the decryption function and injects the plaintext somewhere in the context. Which, sure, we can check if they did that, but the whole point is presumably hiding that info so why would they

Fwiw, even just going through your software with one user can give quite a few insights about what's not obvious about it. That's not at all to say you never need more, but very few open source projects do user research in the first place, being passion projects that just scratch the developer's/s' itch. More samples is always better, definitely at n=10, but I'd also not dismiss the results and benefits of doing it!

Thanks for the reply and clarifications. Don't have much to say further besides agreeing with the sentiment

my work be classified as "slop" because a few months ago I subscribed to and started using Claude

Wait I'm not sure I understand the cause and consequence here if you're saying you use an LLM but your articles aren't slop (note that those are rough synonyms). Do you mention in your submissions you subscribed to or used Claude? The mere existence of the subscription and using it for unrelated things won't be why people flag, I can only assume they notice the quality level and/or style

As of a few years ago I can confirm that. Don't need (as stated by GP) to be active for years, at least, depending on how much and what/where/when you post. Being the first reasonable-sounding or insider-knowledge comment on a new article that then becomes popular gets you ~1/7th of the way with 2 minutes of effort per attempt; being the first to post from an RSS feed of a here-popular blog can get you +500 points in a single go. Just posting regularly, not doing anything special to farm, will get you there in a few months

"bad faith"? As in, that the person posts the accusation while actually believing the opposite? I would think the far, far more common reason for posting that incorrectly is that they legitimately thought it was slop. Of course such low-effort wrongful accusations are bad to post for multiple reasons, but I don't think people frequently post such things falsely

AI helped flesh out the idea and write the pages of text and structure the whitepaper.

- LLM for helping with structure: sure. feels to me like more of a motivation problem to commit a first version to paper so you can start editing and moving bits around, and the LLM's role is being a conversational partner who cares and talks to you, but whatever works

- LLM for writing the pages of text: I would almost certainly prefer the version where you just write the key aspects (your insights) and not have to wade through pages of prose to find them

- LLM for fleshing out: hell no. if you want to lecture and bore people to death, this is the way to go

Turning insights into elaborate documents is relevant once this makes it to legal codification (compare "tell people how you use their data" with the relevant articles in GDPR)

re PS: no; unless HN now swallows em dashes, what you posted is https://unicodeplus.com/U+2013, not https://unicodeplus.com/U+2014

Yeah, even if this is very hard to get it perfect, I keep thinking there has to exist a better voting system than what we have. We can do a presumed step in the right direction and evaluate.

- To give an outlet for (dis)agreement: We could have two arrows on each post, one for agree/disagree, the other for 'legitimate post that helps the conversation'. Someone posting wrong info rarely helps the conversation, but also often enough it's an opinion matter and you can disagree while leaving the latter arrow alone. The latter determines sort order (most helpful first); the former is displayed as a fun fact on posts that are deemed helpful (author sees exact counts just like now)

- A system that seems to work better: A local tech site uses a system where you assign a quality level to a post. The short labels are (translated) "flamebait", "irrelevant but well-intentioned", "relevant but well-known", "useful contribution", and "exceptionally useful". Each one has a numerical value, ranging from -1 to +3, and it'll take the median of all votes (with a half-vote bias towards +1 I think, but that's an implementation detail). If your median comes out to -1, it's collapsed by default; at +2 and +3, it gets highlighted. Incorporating opinion into your quality assessment is a problem here, too, but then you might get +0 instead of +1 and not (like on HN) a score of -7. Not just does that feel a lot more fair, the system and its UI also just makes more clear to everyone that you're supposed to objectively score the comments on the mentioned guidelines (it shows a description every time you hover over a vote option)

(Not OP.) I can understand that it feels like a punishment though, if you are legitimately trying to help and this is the thanks you get. It's not just a human-to-human notice, but a change of your permission level; you're being stepped down. Of course I have no idea about this specific case and what the user flagged/vouched, and I can see your logic, just that I'd probably also feel bad if it happened to me and the wording seems fitting enough to me

Why wouldn't you be allowed to?

How does it differ from the existing apps that do this?

Oh, okay I didn't know you weren't getting around to all incoming emails, that makes sense then. Previously the responses were fast and reliable and I thought I was being dropped by a silent bot/spam detection algorithm (as is becoming frustratingly common).

It's indeed from two different email addresses from the one on this account. They're not that important so never mind, thanks for checking and for the reply!

Blocking in ipv6 works roughly the same way as in ipv4, just that the scale is different. Instead of blocking something like a company's /24 or an ISP's /16 when they don't respond to abuse messages, you block the company's /48 or the ISP's /32. It'll vary per organisation how large a range they got exactly but you can see that in WHOIS. End users are no longer at a /32 (v4) but at /64 (v6), or some prosumers might have a /29 (v4) and /56 (v6). Same concept, just a different prefix length

Opnsense has a traffic capture feature in the interface diagnostics menu, if you want to spot check what servers the devices are currently talking to.

Should be pretty obvious: client devices and internal services will have no traffic >95% of the time, just NTP for timekeeping, DHCP lease renewal, and associated ARP (running total: two dozen packets if you monitor them for a full 24h), then any system updaters (readily identifiable by the initial DNS requests), and finally of course you'll see the traffic of the service that the device hosts, if any, which can be easily dismissed by not looking at incoming connections (scraping uses outgoing connections)

LI've emailed there in May and June about different topics and gotten no reply to a request to confirm receipt. Is there a backup method for when your algorithm throws people's email away without even informing the sender with a delivery failure notification?

Can't have nice things because someone else is worse

The article says the game should cost 18$ (should be ~15.74€), but it's actually showing as 19.50€ aka 22.29 USD. Does anyone know what's going on there? Is Stream charging more due to refund laws or something?

I can't find a website for the developer, also nothing listed in the Indonesian Wikipedia article. Sadly no way to buy it directly and give them 100% of the money in their currency

I found Damn Interesting because of an orbital mechanics simulation the author coded in javascript as a one-off for an article about iirc cyclers. Crazy amount of effort for what's probably about 10 seconds of eye-candy for the average reader. I found it a really neat implementation and while the articles are a bit long for me, it got me hooked on their podcast. There seem to be few projects as thorough and long-lived as this one.

Why make the assumption they don't understand? I understand how this tech works and I'm very wary about it being rolled out widely, even if I see the legitimate uses, so I can understand their point. It might not have to do with 'bashing things one doesn't understand'

For example mutual TLS enforcement is something that can happen on my phone to prevent me reading what an app is uploading. It prevents me from auditing things, at least as easily as it used to be: add your key to the trusted keys list and see if e.g. a claim about on-device processing is true. It isn't only the corporate world that uses it, or at least, not only on their own devices

Ah yes, very true

I'd say government is your best bet unless you live in a low-income country and can get into one of these teams

We broadly aim to be legal worldwide since it's not very cool if we get outlawed in e.g. Australia and businesses there can't use OpenStreetMap for anything, can't download OsmAnd there, MapComplete is blocked from browsers, etc. Incorporating in the EU was actually a consideration due to Brexit (don't remember the reasons) but it sounds like the benefits didn't outweigh the effort of moving, seeing as it hasn't happened

Interesting though that there's this different (15 years) term where normal copyright is just about forever. That's going to come in handy somewhere for datahoarder me I'm sure :D

Yay! Works for me and looking good :)