What is Digg?35 Comments
- AllenS, on 10/11/2007, -2/+49"It doesn't really work yet."
And it never will... - scuba7183, on 10/11/2007, -0/+13hahaha noobs
- satansbanjo, on 10/11/2007, -0/+11I own 52, 1653 and tan(90)
- geminitojanus, on 10/11/2007, -1/+11Don't bag the process, bag the implementation. Audible Magic's apparently really sucks, or doesn't get run at the correct time (at upload vs. a daemon process). The software works and has worked for years, but apparently Audible Magic's implementation isn't as hot as they claim it is, especially if it can't tell an /exact same file/, something that could and often is accomplished by simply hashing the damned thing.
AM basically just threw their reputation in the trash can (not that they had a really viable rep to begin with). I've seen better 10k line open source implementations. - kindrobot, on 10/11/2007, -1/+10"the person who BOUGHT IT so they can be held responsible."
We wouldn't want to let the few who actually purchase entertainment go free to do more of that nasty purchasing. Besides, what if the trace leads to some guy who had his dvd/cd collection stolen? Nothing says fair play like forcing some innocent person through a legal process because the process is broken or designed badly.
Or let's say the same man gets all of his CDs and DVDs stolen and he downloads replacements? Is that theft or spitting in the face of .. wait for it.. someone who actually stole something? Oh how I wish I could download a new 1000 dollar bike that was stolen from me a few years ago. It was really nice, collapsing frame, very lightweight. The manufacturer doesn't even make it anymore. My point is, there's no replacement for that bike. When someone stole it from me, there was no way to get another one without paying a collector's price, which I could never afford. I would sympathize with the music/film/tv industries if they were selling bikes and some Swede hi-jacked a truck, making off with their inventory. But the nice thing for them is that they don't sell bikes, they sell intangibles like "rights" and "access".
As the means the industry used to control, like quality production values, fall into the laps of everyone, they are scared to death. Think of it like fight or flight. They've decided to fight when the right choice would be to flee, re-group and recognize that it wasn't something they needed to fear in the first place. What used to be selling out is now a new model and it needs to be seen as the new revenue stream for major artists.. licensing. The film industry may be a different animal, but they seem to be doing fairly well when they release product people actually want to see. TV needs to address overall visibility and they need to find a way to measure it much more effectively than they do now. (examples: Jericho, Firefly, Veronica Mars, which all have intense, large fanbases when you factor in niche and demographic domination, downloads, Tivo and fan participation in online off-shoots including fan sites/forums)
Fingerprinting purchased items violates more than privacy, it violates pure common sense. Nothing will fix the broken business model, it has to be replaced. Lead, follow or get out of the way, because it's all changing whether or not the old boys like it.
These changes will NOT kill indie music, it will simply change the stakes. Will music lose out because the people choosing to create it do so for different reasons than they used to? Someone asked me once if I thought an album like Dark Side of the Moon could be made in today's music business. I responded that it would NOT, but that it also didn't matter. Dark Side of the Moon was already made. And the next great big leap "forward" in music would most likely come not from a budget and a major label, but on ingenious invention and basement technologies. All you have to do is look back to the history of the press and it's control to see proof that fears such as these are unfounded. Quality is not repressed by a lack of money, it's usually repressed by too much of it and lazy acceptance of the norm. The kids will work it out. - Wootery, on 10/11/2007, -0/+6Anyone know where I could buy 09 F9?
- ipoddj, on 10/11/2007, -1/+6Never say never!
- Aliarse, on 10/11/2007, -1/+6If its made by humans, it can be broken by humans.
- strangewill, on 10/11/2007, -1/+6Me, along with anyone else with half a brain, called this.
Hope your ***** company with ***** project managers goes under. :D - ivanisavich, on 10/11/2007, -3/+7Well if it does work someday, I'll just upload all of my videos tilted 15 degrees to the right so the fingerprinter won't know what hit it.
GREAT SUCCESS! - geminitojanus, on 10/11/2007, -0/+4It's not for you to buy, it's server-side not client-side, to prevent the users from pirating along businesses channels.
That doesn't mean there won't ever be a BitTorrent or any other P2P system, it just prevents people from using certain servers to do certain things (like preventing people from uploading network content to YouTube or Myspace Videos, etc). - xyqxyq, on 10/11/2007, -2/+6I own 10 digits of pi, from digit 4000 to 4010.
- geminitojanus, on 10/11/2007, -0/+4The problem with "screenshots" is size disparity, and encoder noise. You have to resize the largest image to the size of the smaller, then do a compare, then do levels on the comparison to make sure people aren't shifting or putting colors out of phase, or any kind of signal degradation at all.
..actually, rather than explain all of the gruesome details about video/audio fingerprinting, you can just read this paper:
http://ismir2002.ismir.net/proceedings/02-FP04-2.pdf
or read any of the three papers I linked above to find out more about general binary data fingerprinting. - Error601, on 10/11/2007, -0/+4Stop confusing people with science.
- knobtwiddler, on 10/11/2007, -3/+6you can't own numbers.
- glitchbit, on 10/11/2007, -2/+5Audible Magic taking on such a task was stupid move. Did they not see what was coming? When it doesn't work on the high profile sites they become the scape goats and all the blame and lawsuits start heading Audible Magic ways instead of the respective sites the content was loaded on.
And yes hash checks should have been implemented, however the hash changes if the video is re-encoded or cut shorter than the original. So the best solution would be to do two methods.
1) A hash check of the entire clip, and a chunk of the bytes in the center of the clip.
2) The second method should employ a screenshot of a frame or better yet several frames at spaced apart intervals and then have it compared to previously banned clips and if the majority of the frames come to a close match then it is also banned.
The concept is extremely easy to understand and there already exists code that do both of things but not in that manner as far as I know... If that is what some of them like AM is supposed to be doing then they simply aren't doing it right.
Note: Method 1 would should be first as it is much less cpu intensive and Method 2 should come second since it will take longer. - Travelsonic, on 10/11/2007, -1/+4"Why instead of doing all of that, just check the files when they're uploaded. If it's copyright someone else, delete the file and ban any further uploading of it. The process works..."
Nah,
1) the rare instance in which the copyright holder gives prmission to upload the works I.E indie music, or in the instance where the copyright holder is uploading the music he made.
2) Very time consuming.
3) Workarounds... people are smarter than you think. - inactive, on 10/11/2007, -2/+5@ xyqxyq
4000-4010 = 11 numbers - stalefries, on 10/11/2007, -1/+4@xyqxyq: Isn't that 11 digits, actually? I contend you are infringing on my digits!
- geminitojanus, on 10/11/2007, -1/+4Here, have a paper: http://citeseer.ist.psu.edu/lindkvist99fingerprinting.html
And another: http://citeseer.ist.psu.edu/715559.html
Or my personal favorite: http://citeseer.ist.psu.edu/151172.html
The processes are extremely solid, the math has checked out for at least 10 years, and in practice can see widely varying success, depending on who implements it and what thresholds are set. There's a limit to how well it can work that's similar to Error Correction thresholds, but the process is very solid. - moxley, on 10/11/2007, -1/+3Even if they get this stuff to work, who would ever use it? Who would ever buy something with this crap included?
There will ALWAYS be a way to circumvenmt and subvert, - anybody who tells you otherwise is either full of ***** or stupid (at least until we get into deep quantum mechanics, and at that level we all have everything and nothing at the same time anyway).
A system may work for a short amount of time, or may have the effect of causing people who are not very technical not to be able to circumvent as easily, but protection systems will always be broken. - Travelsonic, on 10/11/2007, -1/+2"digital fingerprinting fingerprinting system to see if it blocks copyrighted material. Guess what? It doesn't really work yet."
Of course it doesn't work yet, it's supposed to only filter copyrighted works in which the holder didn't give permission to upload.
Why is this so hard to get RIGHT?
I may know what they mean, but if the world really must know the facts about copyright, then it MUST be worded right, as the majority of people DON'T know what they meant. - MilesLombardi, on 10/11/2007, -0/+1Solution - Martians.
- Atomic1fire, on 10/11/2007, -0/+1Heres what one could do
create a specific ID code of sorts for each perchased dvd/music cd or file and then when it gets uploaded you have a fingerprint to go by
and if someone tried to change or remove it
create a script in the file that would self remove the data there by rendering someones copy useless
of course that would be the easy way out
and someone could just make a mirror of the file and when the original blows the mirror is readably changeable
unless they are connect to a server or some other connection (not always the case)
and the script could still get removed by some hacker/cracker
even with safeguards - Travelsonic, on 10/11/2007, -1/+1"then explain it to the site runner; it's likely they've got some kind of override for uploading protected content from legit sources, yours is no exception."
Another question is raised; why should people who legally uploaded copyrighted stuff have to go through hoops at all? - geminitojanus, on 10/11/2007, -2/+2"why should people who legally uploaded copyrighted stuff have to go through hoops at all?"
Because the world's not fair little Timmy. For the same reason there are speed limits, safety rules. The "hoops" as you would call it are really no big deal, no more so than writing an email, calling someone, hell if the companies really wanted to they could just have a little button that says "Have a human review this decision".
For the most part, if one of these systems is going to flag your content, you shouldn't be uploading it, it's probably not yours to upload. That isn't to say it will always be true, just as a spam catcher won't catch 100% of spam, just as internet content filters don't work 100% of the time. Digital Fingerprinting systems are simply cheaper to run than all of the required manpower of dealing with thousands of DMCA take down requests.
Don't like it? Call your congressman and get him to repeal the DMCA... In the mean time, this is the best solution. - slapthemonkey, on 10/11/2007, -3/+3Fingerprinting field is still problematic. It doesn't work.
- Travelsonic, on 10/11/2007, -2/+1To amend my last post: Speedbumps impeed an illegal activity, you propose to impeed upon a legal activity on the assumption that it will hurt only the potential of illegal use.
- Error601, on 10/11/2007, -4/+3This is nothing new or particularly advanced. It works fine. Someone just screwed up their version of it.
- geminitojanus, on 10/11/2007, -2/+1"1) the rare instance in which the copyright holder gives prmission to upload the works I.E indie music, or in the instance where the copyright holder is uploading the music he made."
The system has this built in with content registration. If the copyright holder gives you permission, then explain it to the site runner; it's likely they've got some kind of override for uploading protected content from legit sources, yours is no exception.
"2) Very time consuming."
The average metadata extraction rate varies by implementation but for very simple downsampling, transforms and storage, it can go really, really fast. One company I worked with advertised a "11 millisecond or less" extraction rate (as tested on an Opteron 260, a fairly average server processor), and the key-sizes are around 3k per document, so having a database to store the information isn't a problem either. Even better, you don't necessarily have to have the whole document to start your keysearch, so if half way through the upload enough frames are identified as being copyrighted, you can simply drop the connection and explain to the user what's going on and why you're not going to waste your bandwidth on it.
"3) Workarounds... people are smarter than you think."
Sure, they are. But when the system is essentially a black box, all they can do is poke and prod until they get something right. Like I've said before, specific implementations can be defeated, but the technology as a whole cannot. It'd be possible to run multiple implementations, different algorithms based on different data analysis systems, and to do this all within one server. And if one $2000 server can save your business from a nasty $500,000 copyright violation lawsuit, I'm pretty sure you'd pay for the server. This technology isn't rocket science, it's not even hard, but it is exceptionally hard to defeat if properly tuned because you can't ever change anything enough to get it pass the detector and have it remain the same information. All published attacks verses the algorithms (not specific implementations) have required data injection, and once you're adding noise, you're defeating the purpose (who wants to listen to a terribly noisy, further compressed source?). - Travelsonic, on 10/11/2007, -2/+1The world isn't fair often, that's true, but that does not mean that, in the eyes of any logical person imo, this is a good excuse for laziness, or an excuse to make things non-fair as often as possible deliberately.
"For the same reason there are speed limits, safety rules."
Non-sequitur, speed bumps and whatnot do not act the same in prevention of actions. - geminitojanus, on 10/11/2007, -4/+2It works *extremely* well if you've got your ducks in a row. Apparently they don't. Go read any number of papers or get a real demo from a better company (I won't name names because of pending contracts).
- cjhowe, on 10/11/2007, -4/+1...
- geminitojanus, on 10/11/2007, -5/+1"Instead of forcing/relying on sites enforcing fingerprint checks (like some kind of Soviet bastard child) it would be better to take the far more likely-to-work and more capitalistic method of having people or private investigators go around websites looking for fingerprinted material and tracing that material back to the person who bought it so they can be held responsible."
Why instead of doing all of that, just check the files when they're uploaded. If it's copyright someone else, delete the file and ban any further uploading of it. The process works, this company's implementation apparently doesn't.
"Fingerprinting can be done properly and it can be a very effective anti-piracy tool. If you know that your personal information is attached to an audio or video file in a way that cant be removed without completely destroying the footage, you will think twice about uploading it to sharing networks."
This isn't about attaching information to the file other than simply identifying it as copyrighted. No fault, personal or otherwise, is applicable to identification of media, it simply prevents the infringement from happening, saving both you and the company from the grief of having to deal with the MP/RI/Intellectual-property-guild-of-your-choice-AA.
The Digg Toolbar for Firefox lets you Digg, submit content, and keep track of Digg even when you're not on the Digg site. Download the official 
© Digg Inc. 2009
— Content created and posted by Digg users is