Degradation versus content theft?

One way to defeat copy theft of textures and prims could be build into the system a minor amount of degradation in the upload/download process.

Textures could be watermarked with perhaps a dozen pixels in certain locations having their r, g, or b values incremented or decremented by one or two counts, hardly noticable. I a thief ripped this texture and uploaded it, it would again be watermarked but this time the watermark effect would be doubled and the effect more pronoounced. Hopefully it would be more noticable after this second cycle through the upload/download process and be recognized as a stolen copy. The actual bits affected could be a secret function of the bit pattern of the texture and would of course only be applied on the server, not by the client. The tricky bit would be for the function to find the same pixels it calculated the first time through the cycle.

Similarly, prim dimensions and positions could be slightly degraded on the upload/download cycle. To a degree SL already suffers from this in prim drift that most of us have learned to live with for moving parts. Again the idea would be to randomly degrade the dimension where it would not be noticable the first time through, but where it definitely would be noticable the second time through, making it easier to spot copies.

The son of copybot could then be used by content creators to back up their own creations on their own hard disks and they would have the fully accurate copy of their creations before they were uploaded for use in their other creations.

Except for LL's turgidity in dealing with this, do you see any problems?

Ed

I'm moving this Feature Suggestion forum.

Yes you could degrade images, nobody would like it. If you can watermark an image someone will try to remove the watermark. The software you use to identify the watermark can be reverse-engineered and new software that works on the same principal can be built that strips/destroys the watermark.

Degrading prim attributes would have a significant noticeable effect and would annoy the heck out of builders (myself included).

I do not support this feature suggestion.

LL could use a slightly different approach though basically we are looking at some sort of one way algorithm to break the upload/download cycle on stolen prims.

One reason pdf files are so prevalent on the web is that they are not so easy to reverse into their original components and to produce a slightly different .document would be somewhat harder. I have tried many times to convert a pdf document into a different domain (html or word) but never quite got it right. The principle here is that uploads would be in one format, downloads by a different format and that it would not be a trivial task for a program to reverse the formats.

Degrading prim attributes would not matter so much if the upload accuracy was much higher than the download accuracy. The current download accuracy is less than 8 decimal digits (a float has a mantissa between 0 and 16,000,000 and that is only 7 1/2 digits.). If SL were to use doubles in the uploads (mantissa accurate to 15 decimal digits), and the rounding off always went in a particular way, they would compound on articles passed through the upload/download process a second time. The original would be as good as it is now, the copy would be slightly distorted.

Strife wrote:"The software you use to identify the watermark can be reverse-engineered and new software that works on the same principal can be built that strips/destroys the watermark"

My answer would be that since that software runs on their server that it would not be available for reverse engineering. LL are certainly not giving anything away on their Supply Linden software and are very coy about its functionality, so I guess they think it unlikely for someone to reverse engineer that.

I guess there is not much chance of LL taking any measures like this to reduce copy theft. They say it is a resident's responsibility!

Ed