Jacek Antonelli

Cap'n, reality/rhetoric divergence is at 70% and rising! The baloney sensor readings are clear off the charts! It's shifted into hyperbole-space, Cap'n!

The Japanese themed houses are very well made, and are definitely my favorite of the bunch. They are also the only area that actually had paths/walkways among the houses, and likewise the only area that felt to me like an actual neighborhood, rather than a jigsaw puzzle of houses.

Mind you, the other areas are not awful, but they feel very artificial, especially the modern one. The modern and cabin area need big trees to break up the monotony, like the fantasy area has. It's a pity SL doesn't support megaprims.

Be aware that "inflating prices on Xstreet SL compared to in-world or other e-commerce sites" is not allowed on Xstreet. In other words, selling items in-world for less than you do on Xstreet can result in your Xstreet listing being removed.

https://www.xstreetsl.com/modules.php?name=Content&pa=showpage&pid=22

Excellent interview and a great read. It's refreshing to see something communicative from the top tier of Linden Lab about what they're thinking and where they're headed. And surprising to see something that hasn't had all the substance and meaning boiled out of it by PR!

@Nalates: I agree that it would be a monumental and expensive task to scan the entire collection of existing textures, which is why I didn't suggest that.

What I suggested, as one possible method of detection, is that new textures could be compared at the time they are uploaded to the fingerprints of all registered textures (not the entire collection of all textures). The task of comparing an individual image to a large set of image fingerprints is feasible, as the TinEye service demonstrates.

Your claims of the amount of processing that would be necessary to provide a useful service are grossly exaggerated. "Hundreds of thousands of variations of rotation" is just nonsense. Even if the system needed to be robust against rotation, such a high degree of precision would be unnecessary. TinEye is not fooled by small amounts of rotation, so at most a few hundred rotations would be needed, and more likely only a few dozen.

But speaking practically, it could catch all lazy rips (textures re-uploaded without any modification) without any variants, and many other rips with only three rotational variants (90°, 180°, and 270°). TinEye appears to be fairly robust against scaling, cropping, some perspective shift, color changes, and many types of visual filters, so it would not require checking a huge number of variations of the image, as you suggest it would.

But even though it can be fooled in some ways, it would still be a useful method of detection. And more importantly, the trade-off for it would be a measurable investment of material resources, whereas most methods of prevention would require a drastic reduction in the quality of the service and user experience, and would still be less effective at deterring or punishing infringement.

Speaking as both a third party viewer developer and as a content creator, I would not make any additional trade-offs to try to prevent copyright infringement in Second Life. Such trade-offs are way off balance: you have to give up a lot of freedom and useful features just to get a tiny bit of extra security.

For example, Linden Lab could try to block all third party viewers from connecting to the grid. That might stop bad viewers from connecting for a little while, maybe a week or two (if LL was really clever about it), before the baddies figured out how to fool the system to get around the block.

In exchange for that fleeting bit of protection, you would have to give up all the extra features in all the third party viewers. And not just avatar radar, rainbow beams, and jiggly boobs -- you'd lose all the third party tools that help content creators, too. You can say goodbye to free temp uploads, the asset browser, backing up your creations, and other tools that might be created in the future. Personally, as a content creator, that's not a trade-off I would make.

And even if no third party viewer could connect to the grid, that would only make infringement slightly less convenient, not stop it. Even the normal Second Life viewer downloads (caches) all the textures you see onto your computer, and they are available to anyone who knows where to look. But, in theory, Linden Lab could get rid of the texture cache to make it even less convenient to rip textures (but still not stop it entirely).

In exchange for getting rid of the cache, you would have to re-download all the textures you see in SL, every time you log in. That means content in Second Life would be a lot slower to rez, and Second Life users with bandwidth quotas (a very common situation around the world) would run up their meters a lot faster. Linden Lab would also have higher bandwidth bills, so they might increase the upload cost or other fees to make up for it.

So, nearly every action Linden Lab could take to prevent infringement would have negative consequences much greater than the protection they would offer.

But, there's a much more effective way to deal with copyright infringement, with much less severe consequences. Instead of prevention, the focus should be on detection and enforcement.

For example, Linden Lab could conceivably employ a system like TinEye ( http://www.tineye.com/faq ) to compare uploaded textures to existing assets. Of course, there's a trade-off there, too: the system would cost money for Linden Lab to set up and run. But, they could perhaps recoup their investment by charging a small per-texture fee (say, L$10-L$50) to register a texture in the system, for content creators that want to do that.

A few extra L$ to instantly find out when someone rips off your texture anywhere in the grid? Now *that's* a trade-off many content creators would be willing to make.

Surely it would not have been very hard to _license_ the trademark for use to the SL Education wiki, for free but with conditions about how it could be used. It would have done just as much to protect their trademark, and sure would have saved LL from a lot of bad will from the community.

That said, I suspect LL wasn't directly involved in this decision. They probably hired a legal group to "set the hounds loose" to find and kill sites that were using their trademark.

If LL's higher-ups have any sense, they'll apologize for the gaff, explicitly permit the SL Education wiki to continue, and reconsider their legal strategy. A company doesn't have to become evil and heartless to protect its trademarks.

"Badass" describes it pretty well. Also, "shakey". Their Steady-Cam is severely malfunctioning! ;-)

Well, most if not all of the data necessary for this study could come from the transaction history and friends list, which we all know LL stores. The data would still be useful with the names changed to "Avatar #13841", etc.

Of course, there's lots of other data that LL tracks about us, some of it disturbingly personal, but there's no reason to think they'd go sharing that around without making it non-personally-identifying.

It's pretty cool that LL is willing to provide data for research (as long as the data isn't personally identifying, of course).

I'm just disappointed the researchers didn't take the opportunity to coin "the 'Hoooooo!' phenomenon" as a serious academic term. Oh well, maybe next time.