/g/ - Technology
install openbsd
[Make a Post]90% of websites work fine if you send no headers besides Host. Lots keep working even without that. I think it's kind of funny that hakase is blocking people based on their headers. The proxies sending them could remove the headers without any loss of functionality. The problem, as per usual, is normalfags who don't give a fuck about anonymity or privacy if it means a .01% chance of improving functionality.
(Another funny thing: hakase can't set http-headers from his code because of the way its set up, so instead he sets headers using html meta tags)
>>1253
Endlessly searching for the perfect, anonymous protocol is pointless when 99% of websites are corrupted from within, i.e. willingly give information to outside groups or government or even give them control. The humans running the sites are the biggest security risk by far for most sites.
>>1253
Why is that funny? If someone is going to go through the trouble of using a proxy and configuring it to remove headers, they could more easily just, you know, use Tor like the site is intended for.
>>1254
It's nice to be able to use nanochan from a normal browser that you can use your bank from. Plus, I don't trust any iphone apps not to be botnetted. But the point is that the clearnet proxies (onion.sh, onion.to) could be configured not to send any special headers. I'm not sure why they aren't.
>>1256
If someone ever makes another alternate/hidden internet, there needs to be some sort of standard content hashing, so nothing needs to be re-uploaded. Just post the SHA-whatever of the image you wanted to post, and it will be presented through the efforts of the network. Combine that with image recognition and I bet you could really lower the bandwidth and storage requirements of most of the internet.
>>1256
>It's nice to be able to use nanochan from a normal browser.
You are not welcome here. Go to another imageboard which is not founded upon the principles of anonymity and non-botnet computing.
>muh iphone
yeah, fuck off normalscum
>could be configured not to send any special headers
And that has not happened. Unless it does, we will be free of cucknetters for the time being.
>>1257
Interesting idea, but then noone would want to store images or worse video files because they are going to have to pay for the server fees for every single usage of that image in the world, but without any of the traffic or income. I think it's still a valid idea, but that will have to be worked on.
>>1257
Content-addressable networks are not a new concept. IPFS and the Dat projects are pretty active on that front right now. Content-addressable networks have some serious privacy challenges to solve, though.
>>1262
>IPFS
Last time I used it, IPFS was a slow bloated piece of shit - had to wait half a second to display the --help message. And that's when it's written in a compiled language.
The fact that ipfs had support for multiple hash functions available was also a masive design flaw, allowing for significant data duplication among people who preferred different hash functions. They should just fucking use SHA-3 or something similarly future-proof and fucking stick with it.
Verdict: 3/10 slow as fuck pajeet shit with bad design
>>1263
What kind of potato machine did you try that on?
$ time ./ipfs --help >/dev/null
real 0m0.092s
user 0m0.068s
sys 0m0.029s
And that is on a Celeron based notebook with shitty flash.
>>1264
>0.1 second to print a help message
Still totally unacceptable.
Similarly complex software written in a real programming language like C does not take 0.1 fucking seconds to print a few lines of text to the fucking screen god damn it. IPFS bloatshit takes about 100x longer.
$ time git help >/dev/null
0m00.00s real 0m00.01s user 0m00.00s system
>>1265
That was an Athlon XP 2000+ btw. If I had tried to run ipfs on that computer, it would probably have taken half a second or longer.
>>1265
C supremacists are having a chilling effect on the diverse rustaceans on this board. It's creating a toxic environment.
>>1257
Yes, the same VirusTotal does:
- hash the file before uploading
- check whether it already exist
- if not upload the file.
>IPFS
Above is such a simple thing to implement IPFS would be over-kill, of course.
File: f82983ab5769300948d0eb97f7b604a2ee4966b1362b295e0e4daa8c01a7bd81.png (dl) (611.14 KiB)

>>1252
correct. luckily this imageboard is one of the few places that works without cookies,"referer"[sic], or JS.
>>1252
HTTP, Browsers and SSL are all problematic for anonymity. Only solution is a maintainable fork of, say, Firefox.
>>1549
Better yet, Tor Browser because they have already done quite a few things to avoid fingerprintability but still:
- Reset NoScript to allow Javascript every time you restart
- Do not have uBlock Origin or uMatrix installed
- Send HTTP referrers
- Have fingerprintable SSL through intermediate SSL certificates
- Have many tickets which are neglected but are quite concerning:
https://trac.torproject.org/projects/tor/query?status=!closed&component=Applications%2FTor+Browser
>>1550
>Have fingerprintable SSL through intermediate SSL certificates
Apparantly not since the 'cert8.db' file is stored in memory and not on disk, this is cleared upon New Identity but there's still a mention of it in 'torbutton.js' which says it doesn't clear it:
https://trac.torproject.org/projects/tor/ticket/21559
>>1261
No, p2p networks are feasible. See Napster, torrent, etc before they were killed for ((("""piracy"""))) reasons. All we're waiting for is an anonymous network like Freenet to become big.
>>1262
>Content-addressable networks are not a new concept.
exactly my response when the IPFS meme started
those are terrible numbers you fuck. stop posting
>>1264
those are terrible numbers you fuck. stop posting
and even with a pentium 2 those would be terrible numbers
[Catalog][Overboard][Update]
[Reply]24 replies
HTTP is a piece of cancer because its structure and design is against anonymity (User-Agent, cookies). Lots of identifiable metadata is transmitted through the headers.
Imageboards should have their own dedicated protocol for data transfer (perhaps a heavily stripped down version of HTTP).