Torrenting is DEPRECATED

you only have to find a collision for one chunk

Every single torrent has to be completely rehashed again.

for all you fags that don't know how bittorrent works: en.wikipedia.org/wiki/BitTorrent#Creating_and_publishing_torrents

Consider this: all the protocols you use today will be "shattered" 10 or 20 years from now. And some might already be, you just aren't told about it.

I am not understanding why that's such a hard solution.

fuck man, how are we going to do that?

And after we switch to SHA-256, in ten years, are we going to have to rehash all the torrents with new quantum encryption?

>schneier.com/blog/archives/2005/02/cryptanalysis_o.html
it it too late. abandon ship.

Sha256 is quantum secure.

user, educate yourself.

...

...

Not really, there's still some barrier to entry and my thousands of anime torrents don't really have a any value in targeting.

...

no thanks

not written in an official meme language™

shit, I didn't know that. Then what's the point of building a quantum computer? Can't we all just move to quantum-secure crypto right now?

Asymmetric post-quantum cryptography is not a solved problem. You'd have to drop public/private key systems.

If you want to research existing/new quantum secure algorithms, implement them in existing protocols & software, and then convince everyone to upgrade to a protection against a threat that doesn't really exist in practice, go ahead.

why is every fucking board on Holla Forums full of ipfs retards frothing from the mouth? ive *still* never seen them say a single valid thing after at least a year since they started shilling

go back to 4chan faggot

*unsheaths go-ipfs*

if you want to circlejerk over your 1337 t3c|-| $|>>>/reddit/

Some of us are here to learn by posting zero-effort spoonfeeding questions on topics we don't really care about

>duckduckgo.com/?q=bittorrent sha-1&t=h_&ia=qa
oh wow that was so hard and took so long.

Torrents aren't really affected, afaik it's a tree of hashes. You can't going to realistically do anything to torrents.

It's interesting but it's just the hash treadmill. By the time anything will be practical applications seriously affected will have moved onto the next digest.

sha256 is already obsolete today, those who are smart use sha4096 instead

exactly. that makes it even easier

If wikipedia is to be believed each chunk of the file is checksumed and is stored in the torrent file/DHT. Every time a new chunk is downloaded it's hashed and checked against the hash in the torrent file.

Or you're part of a big group, like people who torrent game of thrones or the latest movie or something.

Imo everything should have at least two hashes, if torrents had both MD5 and SHA1 hash of each block this wouldn't be a big problem. Personally I think Blake2 + SHA3 would be a nice combination. Wouldn't even hurt performance that much.

yes it would.

my dick is quantum. simultaneously in your waifu and in your mom.

Blake2 and SHA3 combined are less cycles per byte than SHA1.

case in point. did you guys like go to 4chan and get rekt by them when you tried to shill ipfs or something?

Skein. Get on my level

Skein, Groestl and JH can have 1024-bit hash.

What is there to say about it? You don't need people to SAY something valid when you can test the validity of the client yourself first hand, there's nothing more to it either, you want to send and retrieve files to and from peers, it does that. What more is there to say about it? This isn't like all those past projects that only had whitepapers
I don't doubt it but if nobody is implementing it then it's practically useless, IPFS has actual clients you can use right now, it has transitional gateways that work right now, there's nothing to talk about because it's already implemented, the only things there are to talk about it the future features which seem nice and the optimizations but all that stuff is dull. People don't even talk about that kind of stuff with BitTorrent.

Absolutely false when we're talking about software implementations. Hardware accelerated SHA3 is supposedly extremely fast but I've yet to see any real benchmarks.
Blake2b is a suitable replacement for torrents performance and security wise but the hashes are over 3 times longer which means the torrent files/initial DHTs would also be over 3 times larger. Terabyte size torrents already have torrent files over 50MBs so over tripling it is pretty inconvenient.

That would make things even worse to defend against attacks like this. You do understand why algorithms like bcrypt exist?

What happened to that 'we bear raaaaaaaaa' whatever shit where they 'made the future of torrent' where a file would be divided in multiple pieces and both would be downloaded separately then combined by the end user, supposedly granting legal immunity to pirates

No, it doesn't.

You mean freenet? Nobody used it because it was slow back in the day and required shitty Java. It grants "plausible deniability" because everything you host is encrypted so you have no idea what's actually on your drive at any moment.
Once IPFS implements private swarms and the datastore code, allowing it to point to added data instead of making a copy of it, it will replace torrents anyways. It's already used heavily in many major Ethereum smart contract services.

sha512 is faster and more secure than sha256 and yet still isn't used in TLS HMACs. Really makes you think.

can I use it anonymously yet
if not fuck off

yes. github.com/ipfs/notes/issues/37

Do you understand that you have no idea what you're talking about? Hashing speed has nothing to do with the vulnerability found in SHA1. Google exploited a bug in the algorithm that made it easier to replicate a result, they didn't brute force it. It's still pretty much impossible right now to brute force SHA1.
Second of all if bcrypt is so much better then why doesn't everything use it? Because it's a password hashing algorithm not a general purpose cryptographic hash algorithm. Bcrypt uses a random salt so your hash of "password" is different from my hash of "password". You're comparing apples to oranges.

doesn't downloading large things over tor fuck up the network?

jesus christ. this board is fucking awful. most of you faggots have no fucking clue about anything. i regret that i created this thread. please kill yourselves.

Can you use torrents anonymously yet?
No? Why are you using them?

That's what they want you to think :^)
Besides i2p is built to better handle things like file sharing. Once Monero integrates it every miner and wallet will be a full i2p node and will catch up to Tor level popularity very quickly.

I barely come here anymore because of the constant bombardment of shit threads that are made. The mods should've really been more assertive about all the bullshit. You think it's worth making a /tech2.0/ with mods that enforce the rules?

Just for clarification, does torrenting over tor harm the network because
1. Downloading large files over tor overloads the network
2. Making many connections to many peers overloads the network
3. Some other reason I didn't mention
Or any combination of the list.
According to this link, it seems to be mainly number 2. Also, there's the problem that you're not seeding through tor.
tor.stackexchange.com/questions/11695/is-it-possible-to-use-bittorrent-over-tor-without-harming-the-network/11712
I guess I should ask is there anything different with ipfs that would change its affect on the tor network.

Imo it should be encouraged. If a larger number of the people who use VPN services for torrenting would use Tor there would be more people contributing hardware to is since people like their torrents fast but even more importantly it would make traffic analysis much harder.

Pretty much what you said. Tor clients have to connect to a limited number of volunteer onion nodes so if everyone opens 50 connections it'll get out of hand fast. In i2p on the other hand every client participates in the routing of other peers so it's much more decentralized and better for complex applications beyond just websites and email. Unlike Tor, which only routs TCP, i2p can also rout UDP. i2p actually has an official bittorrent client already.
IPFS is worse then torrenting on Tor because of all the additional DHT broadcasts it makes. Like I said, i2p is the way to go. Everyone asks about Tor because that's what most people know about but they don't really understand it. i2p is smaller but in the next 5 to 10 years I'm betting it will surpass Tor.

What? Tor clients don't contribute anything to the network. In fact the more clients there are the slower the network is given a constant amount of nodes.

TOR and i2p serve two different niches. i2p is not built to access the clearnet.

Why do you think I meant the clients? There would be more reason for people to set up tor nodes is waht I meant.

We should definitely do something about current state of imageboards. Firstly we should move to alternative network, internet is too friendly to normalfags. Maybe i2p or we can go full dn42 style creating one big VPN with use of BGP, everyone who would like to join would be required to run their own router and configure it properly. That would basicaly solve a lot of quality problems, at least on Holla Forums. Result would be almost 99.9% normalfag-free community network. Cjdns could also work. If anyone is interested we should definitely do something. We have the technology to move away from this shit, just a little bit of effort is required.

I apologize for this user was the very first one who shilled ipfs into tech.

i love i2p but it's fucking dead. also vpn with torrent works fine.

ipfs needs i2p support if it's going to ever work because filesharing over tor will kill the network.

gowo-whapf's-this

...

Have you seen just how amazingly long it took for websites to leave SHA1 and RC4 behind? Literally Google and Mozilla had to interfere with the internet and have their browsers stop supporting it since last year, and if you have an old browser (which is pretty likely because normies fucking hate software updates, literally one of the top hates against Windows 10) you're still going to be able to connect to a website using SSL with a SHA1 certificate.

Also, I'm a megacorporation sysadmin, and you'd be surprised at not only how much of a pain in the ass it is to leave SHA1 behind on some softwares (for example, the very widely used IBM WebSphere Application Server), but also how amazingly difficult it is in some companies to get permission to do this shit. My company fortunately grants an exceptional license to fiddle with production servers in the name of security, but I'm pretty sure other companies might not think the same.

I'm kinda glad Moore's law is dead, because I don't see SHA-1 being completely phased out by 2040

cia niggers don't want you to have sha512 nigger cattle.

...

For what purpose? shit died in 2010

Its better to buy shit and download it anyway

...

github.com/nneonneo/sha1collider

This guy gets it.
Except for cjdns part, cjdns community is currently hipsterland, and code itself is kinda buggy atm.
Take a look at AnoNet, they allow peerings over pretty much anything (tor, i2p, clearnet (if you want to)).
Peerings over tor are actually usable both latency and bandwidth-wise.
I haven't looked at dn42 much personally, but it seems that they prefer peerings over clearnet and more nearby nodes.

we are already setting up bgp over tor/i2p/whatever, we got an irc for that sorta:
irc://6mk5za2izxm5ubu7bhzw3io7x5h6yjnlc7iccmn2ilbwptceaiwq.b32.i2p/
or
irc://psii2pdloxelodts.onion/

channel is #overchan

we also have a usenet network fork with imageboard ui but everyone here has shat on it or stuff.

also on mainline cjdns network at irc://reseed.i2p.rocks/

replaced by what? i won't pay for some bullshit "services" like spotify or netflix

So what are we using for our shit now?
sha256?

one time pad ^____^

BLAKE2.
or SHA3 if you don't mind it being slower.

We don't need to make it harder to use we just need to decouple mods from boards. Once we get this you can "choose your mod". You can have Autistic Andy or GNU/Linux Gabe moderating your posts and filtering out the shit threads for free.

We need decentralized imageboards where a faggot admin can't give mod powers to his reddit-tier buddies. Where the first person to claim a specific board doesn't get to stay Emperor for Life (along with his inability to effectively mod).

That's what I'm exactly thinking of.
I gonna base it on NNTP, however, all my current work is in draft stage.
I already started working on implementation though.
Basically, every board would be like its own CA, with BO being essentially root key.
All stuff would be moderated by signed messages from BO or moderators chosen by BO, and populated thru bit modified NNTP (with forced node signing and verification at every hop).
Every node owner would be allowed to pick any boards they want, board name would include public key, therefore it would be possible to add boards with same name, but different key simultaneously, thus node owner would be able to provide multiple versions of moderations for particular board.
Basically all moderation would be based on solid cryptography and decentralized.
I'm planning using whatever mentioned at testing stages, though, to avoid unneded publicity untill it's solid.
If you want more info or help me with it feel free to join irc other user mentioned.

...

The type of people who watch culturally enriched TV series typically run NSA malware as their OS already.

Well, that was how it was solved in the ed2k protocol many years ago.

ed2k use MD4 hashes combined with file size for identifying files. The hash is a hash of the concatenated string of the hashes of chunks of 9.28MiB.

AICH (Advanced Intelligent Corruption Handler) was added to links. For corruption handling and together with the fact that filesize and hash are used together as identifier, it mitigates many problems with collisions.

In AICH each chunk is divided into 53 parts which are hashed with SHA1 to form a hashset of the chunk. The concatenated string of these hashes are then hashed to a verifying hash. All this is put into a tree with the root hash being a hash of the concatenation of the two verifying hashes below it in the tree. This hash is a part of the link or can be exchanged between clients, if it's missing.

Moderation should never involve deletion.
Moderation actions should just be actions on the clients' views. If the client frontend is written to allow users to choose who's moderation decisions impact their view (or if any impact their view), then moderation effectively becomes an anarchic practice. The 'selected' moderators are only special as their moderation actions get stored with the board, but theoretically I don't see why there couldn't be user-moderators who host their actions elsewhere and clients can select them as "action-sources" (so long as the view rendering occurs client-side, for server security reasons).

muh security
design a convoluted obscure system nobody can figure out how to break, thats how you win

I've seen that idea before, and I like it enough that I'd want to try it, but it has a few issues.

Bans become impossible without leaking persistent identities. Bans are not strictly necessary, but they can be very useful.

If you (soft-)delete a post, what do you do with the replies and other posts that follow from that post? Do you also implicitly delete all the replies to the post? How do you detect a reply that doesn't want to get flagged as a reply, and how do you deal with posts that reply to multiple other posts?

Bans are always impossible with true user tech.

Only on 64-bit processors. It's WAAAAYYYYY slower on 32-bit machines.

Torrenting over Tor is a bad idea because BT clients are REEEEEAAAAALLLLLLYYYYY promiscuous with your IP address. IOW, use Tor, show up in the swarm anyway.

There's a difference between anonymity from mods and anonymity from users. The first doesn't affect discussion. It's a privacy feature, not a discussion format feature.

As someone who uses Tor to torrent over, my client is specifically configured to torrent over TCP connections only, and over a SOCKS proxy (to Tor). The only thing the torrent client can access not over TCP/Tor is DHT for search-only (no announce) operations, as 95% of trackers are UDP-only.

There's still the vulnerability that $TOR_EXIT with $PORTID1 matches $DHT_CLIENT with same $PORTID1. Differing torrent and DHT ports would fix this small hole, but the media companies are unlikely to try and pop users of this method when there's lower-hanging fruit.

WRONG

biterrant.io/

sha-1 was deprecated fucking years ago. fucking retards.

Wrong. Now get out.

Nigga who cares? Nowadays even your watch is 64 bits.

Wrong. Now get out.
anti sage

The redditors just willingly out themselves, wew.

kill yourself
anti sage

user pls

kill yourself

Surely it'd be harder with BEP30?

Really makes me think you don't even know what a fucking TCP MSS is.

You are actually a retard

no u

bump

bump

bump

sage