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

Except the 2.0 does have a most popular apps section.

My main wish is if radicle had a way to make issues online, without installing the software. Runing a piece of software is a high barrier of entry to make a bug report, which the entire reason I use codeberg instead.

That's a downside of all decentralised software, isn't it? If there's a convenient access point, that access point is also a point of centralisation. To be distributed, you have to be running the software yourself. The big problem is that the software always ends up being inconvenient. People have no problem using bittorrent because the software is actually usable.

> If there's a convenient access point, that access point is also a point of centralisation

I don't think that's true. Bitcoin and Ethereum are very decentralized but most casual users don't run a node and do everything with a gateway. ATProto/Bluesky is the same.

What they have in common is shared global state, using a blockchain or PLC. That means everyone gets a full view of the network (outside of censoring relays) and user account are portable. You can go with a commercial provider for convenience but always have the option of self hosting if they raise prices or censor you.

That option is more important than the share or users who self host right now.


Ethereum is an interesting case for this, because there's a default node address in Metamask, and that's obviously the one that most people are using. It also costs money, forcing enshittification.

Not sure about bitcoin - I run my own node for that (as I said, running the software). If you have a light client I think you have to have your own server?

Look at IPFS though. Extremely slow to access files but since everyone used ipfs.io gateway which cached all files ever accessed, nobody noticed the underlying decentralised system actually sucked.


We have a prototype demonstrating writes from the web:

https://radicle.zulipchat.com/#narrow/channel/369278-Explore...

This is definitely on our radar.


They have public repositories on GitHub.

https://github.com/radicle-dev


Like onion/ipfs/many others, I'd expect gateways to pop up if it grows relatively popular. If LLM scrapers don't destroy them immediately, at least.

My main wish is a way to search the repositories on a node. Or a way to have tags. The whole network is like a blackbox of projects unless you can get an outside link.

If the node runs an HTTP API (radicle-httpd) backed by an indexer (radicle-search), you can.

Try https://radicle.network/explore and set index.radicle.garden as your search seed.



Thanks this is exactly what I was needing.

What is the point? As long as you (the service) hash passwords securely, and the users use random passwords with a password manager, you are fine. A password is far easier to type onto another computer or a diffrent web browser.

Can't you just share the TOTP secret key?

TOTP is unfamiliar or hard to use for most people, so they use SMS. Most sites don't support TOTP either.

Even if you use TOTP, it's not designed to be shared, for example look up what hoops you need to jump through to export a single TOTP code in Google Authenticator. And they used to not even have that option; they told you to set up multiple TOTP codes on each website instead. Even on 1password I had to look up a tutorial on how to import a TOTP code cause the menu is in a very non-obvious and deep spot.


I've shared with non-tech people a few accesses with TOTP codes and it's just a WhatsApp message away "hey dad, enter 12345 when asked".

Code is different from key. The secret key is how he can get new codes without you having to deal with it every time he logs in.

I belive that passwords with TOTP are far better than passkeys.

This is confirming my belies that Google is trying to block OEMs that don't pay them to be part of GMS, with the goal of eventually being the only android phone maker.

This will create a social problem of people discrediting images by a niche camera vendor or by a bootloader unlocked phone or from budget cameras or or from niche camera manufactuers from film cameras or from cameras that are old, while still allowing for advanced telecine attacks.

I would add cloudflare, but set it to cache only mode *without* the bot blocking features.


No point, they are all unique requests.


I think he means, get cloudflare to cache your content, so the traffic never reaches your servers to begin with. Assuming your sites content is cacheable by cloudflare. I agree it's a sad state of affairs if you have to rely on a 3rd party..

Maybe I'm not understanding how many requests at a time bots are sending to kernel.org (or how larger kernel is), but couldn't they have a local cache system too, where all it has to do it serve up dumb .html pages, needing next to no compute cycles.


It is server-rendered cgit pages, there are potentially quadrillions of unique. They are not cacheable.


Maybe you're a little hard of hearing. NO POINT CACHING, THEY ARE ALL UNIQUE REQUESTS.


Please remove everything starting from `is` or `si` from youtube links. The stuff after it is just for tracking.


But why don't they just git clone?


These crawlers (in contrast to e.g. googlebot and similar better behaved crawlers) are not very smart: they seem to make very little effort to avoiding crawling useless deep trees of generated pages. About the only thing they seem to put a lot of effort into is avoiding blocking.

(I'll note that while these are generally attributed to AI data gathering because of the timing of when they took off, it's not actually obvious who's running these bots. The big players all have crawlers that identify themselves and are reasonably well behaved, but I don't know if anyone has managed to positively attribute these other ones to any particular group)


> it's not actually obvious who's running these bots

This is fascinating to me. It's a large enough phenomenon that it's affecting the entire Internet and yet nobody seems to know yet who's actually doing it. Which isn't surprising, of course, it's hard to trace back to a source through all these proxies and it's probably a bunch of distinct groups anyways, but still!

Personally I have to wonder how much of it is "scrapers for training data" vs just tool-use LLMs. Even if you use chatgpt in thinking mode you can clearly see it searching and visiting a bunch of different websites to answer a question, presumably faster than any human would. That's got to add up. It's got me wondering why everyone seemingly discounts that as an option


ChatGPT uses a distinct bot type for ChatGPT-initiated web search versus model training:

GPTBot → model improvement / training crawl

OAI-SearchBot → search index / ChatGPT Search

ChatGPT-User → fetch initiated by a user's ChatGPT interaction

OAI-AdsBot → advertising landing-page crawl


Especially since if any crawler IP is in your country, you can get a subpoena to find out who it is.


Meta is an exception. They intentionally ignore robots.txt rules.

(No second chances, I blocked their whole IP range.)


Everyone ignores robots.txt because every robots.txt blocks everyone.


I allow Google, Bing, Yahoo, Kagi, and Apple on my personal blog... who else do I need to allow? Claude, OpenAI, Alibaba, and perplexity can go to hell.


Think about incentives, what benefit does Anthropic obtain by going to hell?


Why do we think that only “good guys” are training LLMs? I imagine organized crime is getting in on the game too.


Because they're crawling a billion webpages, only a tiny fraction of which can be git cloned, and configuring a special case just for that tiny fraction isn't worth the effort (of the crawlers).


Because the crawlers dont care, they are the internets parasites. Their creators care nothing for people or systems downstream of their greed.


vibe coded crawlers run by morally bankrupt trend chasers aren't going to be the most well engineered systems you come across.


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

Search: