Re: yes, they look for stego, as a "Hacker Tool"
At 05:30 AM 8/14/04 +0200, Thomas Shaddack wrote:
On Fri, 13 Aug 2004, Major Variola (ret) wrote:
Even if you map a particular hash into one of a million known-benign values, which takes work, there are multiple orthagonal hash algorithms included on the NIST CD. (Eg good luck finding values that collide in MD5 & SHA-1 & SHA-256 simultaneously!)
Argh. You misunderstood me. I don't want to find hash collisions, to create a false known hash - that is just too difficult. I want to make every file in the machine recognized as "unidentifiable".
No, I understood this. In a later post it was brought up that this is essentially watermarking your content with a unique ID, which can be bad for P2P tracing purposes. So I was suggesting that by using a finite set of 'watermarks' one could avoid essentially embedding a unique label to one's copy of some content, at some cost in Cycles.
The signature busting of MP3s has a disadvantage, though: makes their sharing back to the P2P pool more difficult, and a lot of programs relying on their hash (emule, Kazaa(?),...) instead of their file name will consider them a different file, which causes problems with multisource download (though the problem won't be on your side).
True. But I've found some manual intervention to be required anyway, sometimes you find a few copies of the same content stored as independent files due to slight differences in naming or truncation.
Sorta like the National Forests... resource of many uses... may as well include a mixmaster payload in that worm :-) which also provides some
other overt free benefit like antivirus or anti-helmetic or defrag or
game or bayesian spamfilter or chat or screensaver or anon remailing client or free ringtone :-)
Free ringtones. Good attractant these days. I tend to forget about them as I tend to shun fancy tones - telephones should have a distinctive ring but "distinctive" does not have to mean "orchestral". But apparently there are large sets of people who like it. Weird...
It was disturbing that, as the bottom fell out of telecom, and handsets became commoditized, faceplates and ringtones were highly profitable. Faceplates are at least made of atoms. There are several lessons there, from economic to sociobiological (if there's a difference), none of which are terribly pleasing in my aesthetic. Fortunately the whole PDA vs. cell vs. camera vs GPS vs. smartcard vs MP3 player vs. email-pager etc bat-belt [1] frenzy will resolve in a few years, and perhaps some of the Linux based solutions will not be involuntary citizen-tracking devices and will support privacy of data stored, and in transit, including voice data. And free ring tones :-) All that's needed is one of the hardware-selling companies to start the process, making money off the atoms, and possibly Sharp's Zaurus (?) already has? Perhaps there's a biz model in buying a 3-D color prototyping machine for $40K and setting up a custom faceplate biz for the integrated gizmo of the near future. Hmm, with freedom-enabling software being distributed on the side, it sounds like a Heinlein novel... [1] Batman (tm) wore a belt with too many gizmos. Some widget-fetishist friends/early adopters are similarly afflicted.
On Sat, 14 Aug 2004, Major Variola (ret) wrote:
Argh. You misunderstood me. I don't want to find hash collisions, to create a false known hash - that is just too difficult. I want to make every file in the machine recognized as "unidentifiable".
No, I understood this. In a later post it was brought up that this is essentially watermarking your content with a unique ID, which can be bad for P2P tracing purposes. So I was suggesting that by using a finite set of 'watermarks' one could avoid essentially embedding a unique label to one's copy of some content, at some cost in Cycles.
We can also periodically "reuniquize" the shared files, in some sane period, say every weekend. (That pollutes the shared-files pool with a lot of almost-the-same copies, diminishing the advantage of multisource download. So perhaps is it better to just use encrypted data storage and anonymized P2P network, and keep uniquicity only of the system executables?)
on their hash (emule, Kazaa(?),...) instead of their file name will consider them a different file, which causes problems with multisource download (though the problem won't be on your side).
True. But I've found some manual intervention to be required anyway, sometimes you find a few copies of the same content stored as independent files due to slight differences in naming or truncation.
Yes. However, depending on the system, same files (with the same hash) differing only by name will look as a single file (eg. edonkey or WinMX). Other systems, depending on the file name only (eg. OpenNap), will show files with different names as different, even if identical inside.
It was disturbing that, as the bottom fell out of telecom, and handsets became commoditized, faceplates and ringtones were highly profitable. Faceplates are at least made of atoms. There are several lessons there, from economic to sociobiological (if there's a difference), none of which are terribly pleasing in my aesthetic.
Care to elaborate further, please?
Fortunately the whole PDA vs. cell vs. camera vs GPS vs. smartcard vs MP3 player vs. email-pager etc bat-belt [1] frenzy will resolve in a few years, and perhaps some of the Linux based solutions will not be involuntary citizen-tracking devices and will support privacy of data stored, and in transit, including voice data. And free ring tones :-) All that's needed is one of the hardware-selling companies to start the process, making money off the atoms, and possibly Sharp's Zaurus (?) already has?
Or buy an Enfora Enabler GSM/GPRS module, add a Gumstix module with built-in bluetooth, slap in a suitable display and keyboard, eventually add a GPS receiver, and we're set. All features and security modes we can imagine, and then some. Preventing spatial tracking is difficult though, as we're dependent on the cellular network for staying online. Though if the given area has wifi mesh coverage, it could be easier. (And if the device becomes widely popular, the handsets can serve as mesh nodes themselves - but that's a song of rather far future.)
Perhaps there's a biz model in buying a 3-D color prototyping machine for $40K and setting up a custom faceplate biz for the integrated gizmo of the near future. Hmm, with freedom-enabling software being distributed on the side, it sounds like a Heinlein novel...
Why not? :) Isn't the main purpose of science-fiction (at least its certain kinds) to be the inspiration for the future? On the other hand, perhaps it's cheaper to just get a bulk supply of "blank" faceplates and hire an artist with an airbrush and a talent. It's also possibly easier (and cheaper) to make the parts in more classical way, eg. by casting them from resin. The rapid prototyping machines so far usually don't provide parts that are both nice-looking, accurate, and with suitable mechanical properties at once.
[1] Batman (tm) wore a belt with too many gizmos. Some widget-fetishist friends/early adopters are similarly afflicted.
There is nothing like "too many" gizmos! (Well, you could call such situation "almost enough", but never "too many".)
participants (2)
-
Major Variola (ret) -
Thomas Shaddack