PDFS--Of the WotC Court Case

[Nikosandros]

Very true, I can confirm this as my partner works on 'Question Documents' as her daily grind in a Police forensics lab. To answer the question, it only takes a forensic scientist to poinpoint the watermark with the printer in question, which is then tied to the manuafacturer, then the store it was shipped to, then to who it was sold if payed by credit or POS. If by cash, that can be traced to the time of sale and a marriage of surveillance cameras. Of course, this process only works in a perfect scenario; real world physics always throws up extenuating circumstance or five to make it harder. Just remember, it is well within the realms of possiblity to trace it to at least the time and place the printer was initially sold. Extenuating circumstances just make the trail longer.

I'm not disputing that this works in Australia, but I'm fairly certain that in Italy this would be currently impossible. I sincerely doubt that electronic shop keep a database that correlates customer data with inventory data of sold items. As for surveillance cameras, the recordings must be destroyed after a fixed time, IIRC.

 

[avin]

That would never work here in Brasil or other 3rd world countries.

 

[Grazzt]

wow... between the three cases they are estimating that 2993 people downloaded the PHB2 in the span of a few days...

With a cover price of 35 dollars that means that wizards lost 104755 dollars...

Not exactly. Of those 2,993 people, what percentage would buy the PDF instead of the actual book? Also- what percentage of those would NEVER buy either one? That last group is not paying customers....so their dollars wouldn't count in that equation.

 

[jensun]

Finally (!!) the burden of proof in civil cases is 'on the balance of probabilities' or 'on the preponderance of the evidence' - a far lower standard than 'beyond a reasonable doubt'. While a judgment on the civil case might help a prosecutor (and bring more information to light) it isn't a 'slam dunk'.

No, no it isn't. The STANDARD of proof is balance of probabilities as opposed to beyond reasonable doubt.

The BURDEN refers to who is required to prove the particular issue.

 

[JohnRTroy]

Since people are mentioning foreign countries, I wonder if WoTC might decide on a strategy to not allow PDF sales outside the US and Canada, either stopping sales in the EU or at lease delaying sales to those people until a few months have passed.

That's another possibility why they've pulled PDFs. If they can't control the foreign piracy, they might want to keep it to the countries that allow them to prosecute--at least for a little while, to minimize "Day Zero" piracy.

 

[Nikosandros]

Since people are mentioning foreign countries, I wonder if WoTC might decide on a strategy to not allow PDF sales outside the US and Canada, either stopping sales in the EU or at lease delaying sales to those people until a few months have passed.

That's another possibility why they've pulled PDFs. If they can't control the foreign piracy, they might want to keep it to the countries that allow them to prosecute--at least for a little while, to minimize "Day Zero" piracy.

We were actually drifting on a tangent related to printers and scanners, but I guess that you might be right. Amazon, just to cite an example, doesn't sell digital downloads to customers abroad.

 

[avin]

Oh, nice, now you guys are going to stop them cancel my DDI signature as well because I live in Brasil? (this is a hyperbole)

I was talking about the "trace the scan owner" trick. Brasil operates under international law.

 

[Drkfathr1]

And since these lawsuits are specifically about pirating of PH2, it may be hard to determine how much money they lost, when by all accounts, PH2 is SOLD OUT!

But yeah, as others have mentioned, of those @3000 or less downloads, how many were done by folks that already had bought a physical copy?

 

[pedr]

No, no it isn't. The STANDARD of proof is balance of probabilities as opposed to beyond reasonable doubt.

The BURDEN refers to who is required to prove the particular issue.

Oops.

Yeah, one of the problems with teaching law at a pre-professional (undergrad) level is that we don't cover the procedural stuff as well as we might. Still, an inexcusable slip-up!

 

[smug]

Pixels of each page are more than 6 millions. Are PCs robust enough for this? Assuming software exists for the job.

Each pixel would be at least a byte*, right, so are the all-image pages of your pdfs really 6 MB a page or more (even with medium compression, that's 1.2 MB a page)? Your monitor can't display that many pixels anyhow and although print quality is different, the pdfs they sell aren't print quality, surely? How many of us have a top-quality printer anyhow, that it'd be worth selling them, with all the extra hassle and cost that'd entail?

However, the pages aren't all images; it must be more efficient for text.

Anyhow, even if they are using 6 million pixels per page (and they could be; I don't know anything about pdf or publishing or even images, really, other than as binary files; I'm a coder) I don't think that it's a computationally tricky problem to find the differences between two of the files. However, if they are (and I assume they do) changing the location of the naughty pixel, the number of different copies you'd need to be confident, assuming they are being clever and throwing in some other random naughty pixels, is perhaps too large to organise or buy that many copies. That's a different issue.

Given the software, detecting a single-pixel difference between PDFs could be done in a trivial amount of time. Consider the fact that your computer needs to process all those pixels anyway in order to put them on the screen, after all.

However: Even if this particular system can be circumvented easily (and that may not necessarily be the case), I'm sure there are several possible ways to have a steganographic watermarking system that's quite robust - there's a pretty major field of research dedicated to this very sort of thing, after all.

You can, according to a paper I read a while back, encode the watermark in the Fourier ampiltudes; you make a spatial fourier transform of the image, doctor some of the amplitudes a tiny bit, then transform back. This can be undetectable to the naked eye because it turns out that Fourier phases are more important to us, visually. You can then get that information back, should the image turn up somewhere illicitly, by comparing a Fourier transform of the evildoer's image with a Fourier transform of the original.

I am sure that's more hassle than it's worth, mind.

Ironically, I looked into this only after one of my legitimate pdfs ended up somewhere it shouldn't have.

*The pixel storing the customer number would have to be the same size, and you'd need more than a byte to store an appropriate number of customer numbers; of course, you could combine the number stored with the location and that would be big enough for all customer numbers, in combination.