More Sim Stuff.

[swestrup]

I wanted to talk about a couple of different things. First of all here is a link to an article about a 3D fashion simulator being developed to help folks shop for clothing. THIS is far more like the kind of UI I've been thinking about than what Maxis provides.

I've also been thinking about the simulated object forgery problem that tjernobyl brought up in the discussion here. Until I read _sps_'s response, I hadn't considered the following method of making a forgery (and so I think my response to tjernobyl missed the point):

1. Install a legitimately bought template object.
2. Run a query program on the template that asks for each resource it contains.
3. Feed the resources (suitably altered) into a template compiler
4. Sign and seal the new template using a new nym.
5. Start selling or giving away the new template.

Now, I can see no way to prevent steps 1-4. In fact, if our template editor is any good it should happily do items 1-4 for you with a few keystrokes. (Some artists will have multiple nyms, after all). Making it hard for legit users to create and/or modify templates is counterproductive. (Actually, I expect that one would have to disable some safeguards in the editor for the above to go completely smoothly, if you can't show that you own both nyms, but that will probably not be hard to do.) That only leaves step 5 to try and fix. There we have some leverage.

I have been imagining a brokerage service which helps connect buyers and sellers and which supports the online portion of the template trading protocol. When a legit artist registers a new template with us, we will record the 'fuzzy signatures' (_sps_ and I have done research on these -- its very doable) of all of the contained resources. This will allow us to detect when someone has made changes to a registered resource and is trying to reregister it as if new. We may employ something similar on the servers that host the virtual world that neighborhoods plug into. No unregistered objects will be allowed in the game. (This requires us to make registration free, transferable and without onus on the registerer, but I wanted to do that anyway.)

All of this creates a 'safe haven' where only legit copies of created objects get used. It does NOT prevent a black market with its own servers and brokerage service, but I can see no acceptable way to prevent that (and I'm not sure how such a black market could be profitable in a free economy anyway). We can prevent it from happening 'accidentally', by allowing other legit brokerage services to spawn off from us and for other distributed neighborhoods to be set up. By specifying, in advance, methods of starting up new services and how data will be exchanged between them we prevent anyone having to hack the code out of necessity because we've given ourselves an unfair edge. The only advantage the first system will have is that of first-to-market. As time goes on it may no longer be the biggest or the best of the brokerage services or online envionments, but it should manage to hold its own and continue to be profitable. It may be heresy in a capitalistic society, but filthy rich is good enough for me. I don't need to go for 'richest'. Besides, this is just one of a dozen such schemes I have designs for.

One final point I should make: Forgery and illegal trading of copywrit Sims objects is rife and yet many folks are still making money creating and selling the objects they've made. This is in an environment with NO safeguards other than word of mouth and social pressure to prevent it. So, while I feel we need to address the forgery problem to the best of our abilities, the business plan should work even if we have no forgery protection. After all, no modern currency is easier to forge than US dollars, but that hasn't stopped their use around the world.

So, what do folks think of this?

Comments

Re:

[swestrup]

Keep in mind that we are trying to create a flexible multiple-use person simulation package. We want it to be easily retargetable for research applications in AI, studies of resource use in different environments, crisis planning for events where human decisions play a large part (ie. behavior during floods or spectator effects on firefighting) as well as MMORGs like Everquest and the Sims online. So, the answer to any question about what can happen in-game like the decay of objects will be part of that individual game's logic. We will try and have the engine support of as many such object features as seems reasonable, but in the end individual games will be free to turn features on and off as well as create wrappers to add unanticipated features to objects.

With that in mind, I'll respond to your various comments:

I saw a wrestling game on a console system a while back that used a combination of parameter manipulation and texture-mapping that made it possible to make surprisingly good imitations of actual people.

Its very common in The Sims for artists to make Sims that duplicate either themselves, some loved one, or some celebrity. The only difference is that in OUR game, we'll probably require the permission of the given celebrity. Then again, we may allow 'lookalikes', so long as they are clearly labeled so. I'll have to give this more thought.

As games like EverQuest have shown, anything that happens on the client side can be manipulated. This is especially important on an open-source project; any safeguard that exists only as UI in the client can expect to be patched out by at least some users.

Fear not. I have no intention of putting any sort of security in the code that could be compromised by changing the code. Thats what I meant when I said that we wouldn't build in any special priveledges for the creators of the source. We don't want someone who has to modify the code (to remove the special priviledges) for arguably legitimite reasons (ie fair competition) to end up accidentially helping the black marketeers. People should be allowed, nay encouraged to create new user clients for the online games.

If nothing else, objects will be tracable. It'll be possible and actually fairly easy to trace objects back to their owners. But a legal solution is suboptimal.

Will information on the object's author be visible? Fashion might dictate a better-known designer's work would be preferred. After all, Walmart and Tommy Hilfiger products come from the same sweatshop, but the label is what brings the price...

Objects WILL be traceable, and the creators and modifiers of objects will have links to their identities embedded in their creations. This will be necessary to implement payment methods for objects sold through a broker, and will allow for objects to refer to other objects that they go with. "If you like this chair, check out the entire matching livingroom ensemble!"

Legal solutions ARE suboptimal, but that doesn't mean they aren't useful. Most folks are perfectly happy to pay for things when they think they're being treated fairly, and take pains not to cross any legal lines. Thus, just indicating where the legal line lays is often a useful thing to do, even if you never plan to ever sue anyone. So, we'll enforce fairness with cryptography where possible and the law where not. In the end nothing will prevent someone from starting a 'dark fork', but I predict that, as the good guys, we would retain the vast majority of the paying users.

If objects decay, then adding a modifier based on fuzzy signature might do the trick; a cheap knockoff will usually exhibit poor workmanship!

'Cheap knockoff' may well be an attribute, and may be something we decide to add programattically via wrapper on certain items, but the real purpose of the fuzzy signatures is to prevent someone from creating a new car that LOOKS like a Ferrari and just spelling it differently. Hmm. Now that I think about it more it may just be a policy decision. After all, no one but Ferrari will have the signature for Ferrari so you will only be able to get so far in imitating them.