User:Zero Linden/Office Hours/2008 Jan 15

From Second Life Wiki
< User:Zero Linden/Office Hours
Revision as of 10:23, 24 January 2008 by Zero Linden (talk | contribs) (New page: Transcript of Zero Linden's office hours: {| |- style="vertical-align:top;background-color:#FFFFFF;" | [13:08] | Tao Takashi: | style="white-space:normal;"|TP ...)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

Transcript of Zero Linden's office hours:

[13:08] Tao Takashi: TP is borked
[13:08] Teravus Ousley: Hello :D
[13:08] There is no suitable surface to sit on, try another spot.
[13:08] Tao Takashi: or was the sim down?
[13:09] Teravus Ousley: I was having an invigorating conversation with Zha about trust :D
[13:09] Teravus Ousley: .. then the sim went down
[13:09] Zero Linden: hmmm... what was that all about?
[13:09] Tao Takashi: you should not trust SL ;-)
[13:09] Jarod Godel: Virtual worlds aren't a place, they're a medium!
[13:09] Jarod Godel: sorry, had to get that out
[13:09] Zero Linden: see - and I take a large!
[13:09] Teravus Ousley: well, we were discussing the fact that regardless of how the system is implemented.. you eventaully have to trust the system
[13:10] Entering god mode, level 200
[13:10] Tao Takashi: I am not sure if there is so much to discuss about this topic, isn't this obvious? ;-)
[13:10] Teravus Ousley: .. so it needs to be looked at from multiple perspectives
[13:11] Teravus Ousley: at least from the perspective of the grid manager
[13:11] Teravus Ousley: .. and from the user
[13:11] Tao Takashi: I was wondering today what actually happens if agent A from Agent Domain X wants to give something to agent B from Agent Domain Y
[13:11] Tao Takashi: without any region interaction
[13:11] Tao Takashi: I guess there also needs to be trust somewhere
[13:11] Tao Takashi: and how will agent A actually notice that agent B gets online if he on a different domain
[13:12] Jarod Godel: hahaha
[13:12] Teravus Ousley: yes, good points.. there will need to be some 'middleware'
[13:12] Harleen Gretzky: everyone get logged out of the region?
[13:12] Zha Ewry: Yep
[13:12] Zero Linden: No, I don't think that it is middle wear in there
[13:13] Tao Takashi: so presence informatzion actually should only be transmitted if both the contact is from both side
[13:13] Zero Linden: it is strickly AD to AD, via some form of Pub/Sub
[13:13] Zero Linden: Hello all
[13:13] Tao Takashi: and if that#s the case they know each others URL
[13:13] Teravus Ousley: Hello
[13:13] Tao Takashi: so it's indeed pub/sub
[13:13] Tao Takashi: just with which protocol
[13:13] Zero Linden: Welcome to my office hours - and as most of you know - it's *on* the record - speak freely and speak in public
[13:13] Tao Takashi: might XMPP be one option? never looked at that closely but DiSo wants to use it for profile updates
[13:13] Kirsten Kiergarten wonders where my zero copy objects went after my crash
[13:14] Tao Takashi: but still there is the trust issue for object transfers
[13:14] Zero Linden: I am not a fan of XMPP, for reasons stated in past meetings
[13:14] Tao Takashi: so not only the region domain migth define who is allowed to connect but there also seems to be somethind needed between ADs
[13:14] Zero Linden: as for use of XMPP for "friends on line", which seems like just what it was made for
[13:14] Zero Linden: I don't know if it has been tested at the scale needed -
[13:15] Zero Linden: most people's IM buddies lists are much smaller than their SL friends list
[13:15] Tao Takashi: well, just sending a HTTP request with whatever content to all of your friend's ADs might be somehow possible anyway
[13:15] Zero Linden: And IM lists therefore demonstrate better clustering
[13:15] Zha Ewry: In general, unless there is a really good case for it, I think that each new base protocol you add, needs to be looked at with a fair bit of skepticism
[13:16] Zero Linden: Agreed
[13:16] Zha Ewry: You can end up with a ton of overlapping, similar XML based protocols rattling around your world
[13:16] Tao Takashi: but it's maybe better to add an existing protocol instead of inventing a new one if you want adoption
[13:16] Saijanai Kuhn: and tested with scripted avatars in massive numbers
[13:16] Jarod Godel: How about stabilizing existing protocols first?
[13:16] Tao Takashi: shouldn't the AD part maybe not simply act as some sort of IM system?
[13:16] Saijanai Kuhn: AD...?
[13:16] Tao Takashi: if I just login to the AD I maybe simply can use it as an IM service
[13:16] Tao Takashi: Agent Domain
[13:17] Zha Ewry: I would actually love to see someone find some hard statsitics about large chat group memeberhsup behavior
[13:17] Zha Ewry: The numbers and patterns in things like Fahsion Consolidtaed, have to be pretty unsual in the chat world
[13:17] Teravus Ousley: well, must IMs pass through an arbitrator? or can IMs be passed from beginning sim to end sim directly.. it could know about the end sim via the arbitrator, parhaps.
[13:17] Saijanai Kuhn: if I can ever get that python chat bot done, I'll make a bunch of avatars and run tests
[13:18] Zero Linden: So would I - I had one vendor of IM systems talk with me a year ago and tellng me about their large load tests
[13:18] Jarod Godel: in what regard, Zha?
[13:18] Zha Ewry: Mulitple thousands of members, and a lot of steady turnover in the active on line list
[13:18] Tao Takashi: I don't even know the Linden Lab numbers regarding chat ;-) Might be interesting to know how many people actually participate in group IM conversations
[13:18] Zero Linden: they weren't close to my back-of-the-envelope needs for SL
[13:18] Teravus Ousley listens
[13:19] Tao Takashi: and nevertheless is might be cool to be able to login to SL with my chat client
[13:19] Tao Takashi: so I can participate e.g. via cellphone
[13:19] Tao Takashi: or Eee ;-)
[13:19] Zero Linden: Tao - when we talked about this a few months ago,
[13:19] Saijanai Kuhn: again, if I can get the python chat, you should be able to turn that into a javascdript chat
[13:20] Zero Linden: the group discussed a model where if you did that, your avatar would show up in world at some
[13:20] Zero Linden: place, like your home location or your "out of world" location
[13:20] Tao Takashi: so a region domain is always needed?
[13:20] Zero Linden: and though you couldn't control your movememnt, you'd be there in proximal chat,
[13:20] Saijanai Kuhn: right now, you either get a "ghost" avatar, or a ruthed avatar if you just want chat
[13:20] Zero Linden: and of course have access to private IM
[13:20] Tao Takashi: well, I am mostly talking about IM
[13:20] Tao Takashi: not in-world chat
[13:21] Zha Ewry: Right. Once you get the agent domain split out, one could imagine only getting caps for that, and never actually being in chat at all
[13:21] Saijanai Kuhn: private IM vs group IM?
[13:21] Zero Linden: Tao - well, for you, your AV can be in a skybox! :-)
[13:21] Zha Ewry: But.. that would requires some tweaks to the protoclls
[13:21] Saijanai Kuhn: Would require a major change to the protocols
[13:21] Zha Ewry: I think, at the moment, you can't get IMs sent or back without a circuit
[13:21] Jarod Godel: stick a login on the im<->email gateway and you don't needs caps
[13:21] Saijanai Kuhn: right now, the packet handles all forms of chat
[13:21] Tao Takashi: I am talking not about now apparently but about the AWG future
[13:22] Zero Linden: Actually, I think in the new protocols, an AD only login is a natural fall out
[13:22] Saijanai Kuhn: well, we need to redesign the chat IM packets for that to happen
[13:22] Zha Ewry: I think so too, but only if we change the IM as well
[13:22] Tao Takashi: so I wonder why you'd need a region then.. if I am just in a skybox for doing private IMs better save these ressources for other avs
[13:22] Tao Takashi: same for object transfers, group management etc.
[13:22] Zero Linden: I don't think, once we get around to building IM in the new protocol that it will go through the region at all
[13:22] Saijanai Kuhn: http://wiki.secondlife.com/wiki/ImprovedInstantMessage
[13:23] Tao Takashi: I think quite a lot would be possible then. esp. as there also will be more web services available
[13:23] Zha Ewry: I would hope not
[13:23] Saijanai Kuhn: the table says it all and explains why things can get so crazy when IM gets bad
[13:23] Tao Takashi: the table is scary ;-)
[13:23] Zha Ewry: I would think that you'd not need a circuit to a region at all, once we split it correctly
[13:23] Saijanai Kuhn: I've bounced off it a few times
[13:23] Zha Ewry: Indeed, ideally, no circuit at all
[13:23] Zha Ewry: Just a queue (Hopefully cleaned up)
[13:23] Tao Takashi: actually you could the probably also do land management without being really logged in if region domains would have the rigth web services
[13:24] Zha Ewry: Some of those, would need a way to say "Here" in terms of space, tho, tao
[13:24] Saijanai Kuhn: more than one queue total. One for IM another for other stuff
[13:24] Tao Takashi: of course, Zha
[13:24] Tao Takashi: but I probably can somehow address my parcels without that
[13:24] Zha Ewry: right now, your Ave position determines which parcel you're mainpuating and all. Lots of entanglement
[13:24] Saijanai Kuhn still doesn't get EventQueue Get
[13:25] Tao Takashi: they probably have an internal (UU)id anyway
[13:25] Zha Ewry: Heh.
[13:25] Tao Takashi: actually that can also be up to the implementation if it's not defined by the standard
[13:25] Zha Ewry: Everything does Tao, until you get down to about particals in partical systems
[13:26] Saijanai Kuhn comapgins for UUIDs for particles
[13:26] Tao Takashi: btw, I bought that RESTful book :)
[13:26] Zha Ewry: So.. looking at the trust model
[13:26] Saijanai Kuhn: particles have rights too, you know
[13:26] Zha Ewry: the only thing I saw, in a first glance which looked different beyond twhat clients get
[13:26] Zha Ewry: is that scripts are much messier
[13:26] Tao Takashi: and reading that section about naming I am not sure they really would like CAPs ;-)
[13:26] Harleen Gretzky: Particles don't have UUIDs now?
[13:26] Zha Ewry: Oh, and linksets
[13:27] Zha Ewry: Clients don't really see scripts, or linksets, as such, I think
[13:27] Zha Ewry: But, all the rest, the client gets told to render
[13:27] Sae Juran: hi, sorry for "intrusion", got surl from groupchat
[13:27] Sae Juran: (darn sooooorryy)
[13:27] Aminom Marvin: wtf grief
[13:27] Zha Ewry: So. while people may wimper, pasing a set of prims and textures to a sim for rendering isn't anything new
[13:27] Tao Takashi: but doesn't the client get the prims and doesn't it need to know about linksets then?
[13:27] Sae Juran: sorry i forgot the hud
[13:28] Tao Takashi: and I can actually edit a linkset so doesn't it need to know about it?
[13:28] Zha Ewry isn't sure
[13:28] Jarod Godel: We call that GET in the real world.
[13:28] Zha Ewry: From rendeing perspective, knowing things are linked isn't all the interesting
[13:28] Saijanai Kuhn: physics needs to know, though
[13:28] Zha Ewry: The joints, in the aves, we need to konw about
[13:28] Tao Takashi: of course not
[13:28] Day Oh: For physics, it is... for rendering, you always know the parent
[13:28] Zha Ewry: Right, but clients, don't do phsyics
[13:28] Tao Takashi: for physics it is
[13:29] Jarod Godel: clients should do physics
[13:29] Saijanai Kuhn: clients do simple avatar stuff, even if its not called "physics"
[13:29] Zha Ewry: So.. when we look at what changes when you go to another sim, not one in your domain,
[13:29] Tao Takashi: but isn't only the object moved? if the prims would be moved wouldn't it somehow flickr or get blurry
[13:29] Silicon Plunkett: the client still interpolates between physics events
[13:29] Jarod Godel: but that's another debate
[13:29] Tao Takashi: one piece moving different than another if a packet is lost?
[13:29] Zha Ewry: the rtights issues which change pretty much seem tobe scripts, and maybe some linkset info
[13:29] Tao Takashi: but we can concentrate on scripts anyway
[13:29] Silicon Plunkett: velocity, damping, angular velocity / momentum
[13:30] Zha Ewry: The client does very, very little phsics
[13:30] Zha Ewry: Some simple interploation of Ave position
[13:30] Silicon Plunkett: yeah, what I just said is about it
[13:30] Saijanai Kuhn: and its corrected by the server.
[13:30] Zha Ewry: right
[13:30] Zha Ewry: Thus the float through object, then snap back
[13:31] Sae Juran: any news about the new grid architecture "2008" ?
[13:31] Silicon Plunkett: hehe, that isn't "supposed" to happen.. packets dropped
[13:31] Zero Linden: er, I lost the chain of this thread - are we worrying about if the client knows what is linked to what?
[13:31] Zha Ewry: Or frames, Silicon
[13:31] Tao Takashi: Zero: I think so
[13:31] Zha Ewry: Just trying to ID the few things simulators know that clients don't.
[13:31] Tao Takashi: but I think we wanted to talk about trust and get a bit off ;-)
[13:31] Zha Ewry: Scripts, so far is the only certain one
[13:32] Zha Ewry: Well, Tao, there are entnagled, as the client already is given 90% of what we would give a sim
[13:32] Zha Ewry: Trusy or not
[13:32] Zha Ewry: *trust
[13:32] Zero Linden: Physics state - like the momentum of an object
[13:32] Zero Linden: state of the wind simulation (oy vey!)
[13:32] Teravus Ousley: hehe
[13:32] Zha Ewry: the sailor in me, would say "The totally random wind"
[13:33] Zha Ewry: Or 'The remarkably non relalistic simulation"
[13:33] Tao Takashi: I bet 50% of CPU load is just for the wind simulation ;-)
[13:33] Sae Juran: about script... memory size, any hope to increase heap ?
[13:34] Zero Linden: Or 'Totally goofy simluation put to awesome use by Dutch composer'
[13:34] Tao Takashi: "since we switched off wind we can get 500 avatars on every sim"
[13:34] Zha Ewry: LOL
[13:34] Zero Linden: right - but then we didn't get Dutch avant-garde music
[13:34] Zero Linden: so we turned it back on
[13:34] Zha Ewry: Seems to be just a slightly history based random number
[13:35] Zha Ewry: Not terribly expensive
[13:35] Zha Ewry: now.. if you simulated gusts flowing down the landscape, around the buildings, and such
[13:35] Zha Ewry: Ok.. that would be expensive
[13:35] Tao Takashi: ok, maybe back to trust ;-)
[13:35] Teravus Ousley: fluid dynamics anyone?
[13:35] Silicon Plunkett: agreed.. just a variable for clients and scripts to snag
[13:35] Zha Ewry: Anyway
[13:35] Aminom Marvin: fluid dynamics... now there's something that would be very useful in SL's largest industry
[13:35] Zha Ewry: Besides scripts, is there *anything* the sims see, that clients dont?
[13:36] Teravus Ousley: talking about the object domain, right?
[13:36] Leffard Lassard: Zha, why ist trust with scripts so messy?
[13:36] Zha Ewry: Two reasons, really
[13:37] Zha Ewry: one, do I want to run other people's un trusted scripts on my boxes (in one direction))
[13:37] Zha Ewry: and two.. at the moment, you can't steal scripts int he client, since they never get there
[13:37] Jarod Godel: aren't scripts sandboxed?
[13:37] Zero Linden: Sure you do - they are coded in an interpreted language and you can (and do) sandbox 'em
[13:37] Zha Ewry: Even sandboxed, you may be reluctant to rub them. But yes, you can, and od
[13:37] Zha Ewry: *(do
[13:38] Sae Juran: about (c#) that is mono fully sopported?
[13:38] Zha Ewry: *run
[13:38] Zero Linden: And your library implementation should never trust the script - but check authorization on each call
[13:38] Jarod Godel: Zero, ala usergroups/permissions?
[13:38] Zha Ewry: OK. So maybe we're down to "right now people like that you can't copy scripts with the client?"
[13:39] Zha Ewry: The simpiler the trust story gets, the happier I am :-)
[13:39] Zero Linden: Wel, for example, if a script makes call like llGiveInventory(), the library routine that implements that call is going
[13:39] Zero Linden: to have to check all the preconditions - not just assume the script is valid
[13:39] Finrod Meriman: even sandboxed scripts can do "interesting things" (http request to a adcounter??)
[13:39] Tao Takashi: I was actually wondering if I should bring this topic more up in the Data Portability Work Group but I think it is very special
[13:39] Jarod Godel: Why do we care about who sees the code, if what they're seeing is bytecode?
[13:39] Zha Ewry: Nothing more interesting than they would do on thier home grid
[13:39] Zero Linden: Jarod - bytecode is easily decompilable
[13:39] Tao Takashi: and does not really fit into the web like thinking as there is no need to really copy data from host A to B
[13:39] Silicon Plunkett: yep
[13:39] Jarod Godel: I know that.
[13:40] Zero Linden: and larger security schemes can be encoded in the code if the code doesn't go to the clients
[13:40] Teravus Ousley: :D right, just as it must check and fix quaternions that are declared by the user to be <0,0,0,0>
[13:40] Finrod Meriman: if executed multiple times on each client, the effect would be different than once on a grid
[13:40] Jarod Godel: Just...copyright your code.
[13:40] Jarod Godel: Same as a texture. You can't stop people from stealing a texture, so you copyright it
[13:40] Sae Juran wonder where this discusson bring
[13:40] Tao Takashi: wasn't the idea to let scripts of attachments run on the AD?
[13:40] Jarod Godel: what's the difference?
[13:40] Zero Linden: Right- but there isn't any serious advantage to a script running in the client (for scripts in objects)
[13:40] Tao Takashi: and I guess for in-world objects they run on the RD
[13:41] Zha Ewry: Actually, serious disadvantages
[13:41] Zero Linden: in fact, since scripts in objects (including attachements) affect global state of the sim,
[13:41] Zero Linden: it is a disadvantage to being in the client
[13:41] Jarod Godel: Oh, I thought Zha was talking about shipping your script's bytecode to someone else's server.
[13:41] Zero Linden: since all would execute it, and then some/all/one would send state back to the sim
[13:41] Zha Ewry: That's one of the questions
[13:41] Zero Linden: and how would the sim know that the change was valid?
[13:42] Zha Ewry: right now, my scripts run on the region simulator I'm on
[13:42] Tao Takashi: ah, you are talking about in-world objects or attachments?
[13:42] Zha Ewry: Both
[13:42] Teravus Ousley: yes, :D both :P
[13:42] Jarod Godel: what's the difference? They're both executed on the sim you're in.
[13:42] Zha Ewry: In fact, counting huds, all three ;-)
[13:42] Zha Ewry: .
[13:42] Tao Takashi: I meant Zero because the statement about "all" would only make sense for inworld objects
[13:42] Zero Linden: We could define an interface to scritps that enables you to optionally run the scripts on a "script server" you trust
[13:42] Zero Linden: with the addd overhead of the communication costs
[13:43] Zero Linden: this probably makes sense for attachments
[13:43] Zha Ewry: Right. And lower boundary/tp crossing costs
[13:43] Zha Ewry: A lot lower, in fact
[13:43] Jarod Godel: just put JavaScript in the client.
[13:43] Jarod Godel: there. problem solved. ;-)
[13:43] Tao Takashi: thanks, Jarod
[13:43] Zha Ewry: That makes sim state management a nightmare
[13:43] Tao Takashi: :-)
[13:43] Zha Ewry: (Which I think Zero just mentioned)
[13:44] Zero Linden: yup
[13:44] Tao Takashi: if possible I really would put in-world scripts into the RD
[13:44] Zha Ewry: Which isn't to say one couldn't (and shouldn't) do good things with client side scripting, just that
[13:44] Kurt Stringer: That was my point last week
[13:44] Zha Ewry: they are very different from sim side ones
[13:44] Zero Linden: I would too, Tao
[13:44] Tao Takashi: because if there are many there might really be much communication going on
[13:44] Zha Ewry: yes
[13:45] Zero Linden: I think ifyou are going to put a vendor down in a region, you are already expressing trust in that region, for example
[13:45] Zha Ewry: I think we end up with scripts in more places
[13:45] Zha Ewry: which doesn't seem bad at all
[13:45] Sae Juran: right
[13:45] Silicon Plunkett: yeah, and you can't really verify client-side scripts
[13:45] Tao Takashi: for attachments it could be the AD or some scritp server
[13:45] Jarod Godel: I agree, Zero.
[13:45] Zha Ewry: The only one I am not sure of, is where you run scripts on moving vehicals
[13:45] Tao Takashi: client-side scripts might also be problematic regarding HTTP requests etc.
[13:45] Sae Juran: and XML-RPC too
[13:45] Jarod Godel: I think if you walk into a sim, you're expressing trust in that region
[13:45] Zha Ewry: I suppose, they move from sim to sim, and expensive
[13:46] Zero Linden: Vechicals are always the exception, eh?
[13:46] Zha Ewry: They are tricky
[13:46] Tao Takashi: I'd think that XML-RPC will be gone at some point but of course this is the same issue. any outgoing communication actually
[13:46] Zha Ewry nods
[13:46] Zero Linden: They are like quasi-objects, quasi-attachements
[13:46] Silicon Plunkett: can treat the vehicle issue the same way you treat can-a-script-run-here
[13:46] Silicon Plunkett: if it takes controls
[13:47] Tao Takashi will draw some sketches soon to get the trust problematic maybe a little clearer for him
[13:47] Bell Boyd: Crystal Ice Mirrors!
[13:47] Bell Boyd: Water Prison Jutsu!
[13:47] Zero Linden: I find it interesting that while attachements, objects in region, and vechicals are really all the same thing: prims with scripts
[13:48] Bell Boyd: Water Prison Jutsu!
[13:48] Tao Takashi: like who needs to trust whom and which components etc.
[13:48] Zha Ewry: https://wiki.secondlife.com/wiki/AWG:_Cross_Domain_Trust_issues#The_core_problems
[13:48] Zero Linden: we have different expectations of them based, in part, on our real world expectations of the things theymodel
[13:48] Zha Ewry: feel free to peek here and steal at will. If you want the .ppt from that, let me know
[13:48] Jarod Godel: Zero, so it's not a code problem, it's a people problem?
[13:48] Sae Juran: y it's funny, 0
[13:48] Saijanai Kuhn: and new issues arise if you wnat to provide ways for HUDs to talk to clients directly
[13:48] Saijanai Kuhn: for more sophisticated interfaces
[13:49] Leaving god mode, level 200
[13:49] Tao Takashi: the problem is simply that our real world is not divided in regions ;-)
[13:49] Tao Takashi: at least not as small as SL ones
[13:49] Silicon Plunkett: lol
[13:49] Saijanai Kuhn: smaller regions, broader boundries
[13:50] Sae Juran: llEmail cheat lol
[13:50] Jarod Godel: The core problem with trust is people think they can trust computers.
[13:50] Zero Linden: well, that, and there are physical barriers to copying and state inspection.... Isn't that right, Mr. Heisenberg?
[13:50] Jarod Godel: Not bloody likely.
[13:50] Zha Ewry: Dr. Heisnberg is uncertain abour that Zero :-)
[13:50] Zero Linden: Oh, and that helpful exclusion principle, eh Mr. Pauli?
[13:50] Tao Takashi: only if we build nano vehicles ;-)
[13:51] Tao Takashi: but there are physical boundaries in SL which prevent doing that ;-)
[13:51] Jarod Godel: Trust issues just seem to be a divergent branch of DRM issues.
[13:51] Jarod Godel: "People can copy my bits! Oh, noes!"
[13:51] Arawn Spitteler wonders that scripts don't capture client input beyond control
[13:51] 3LKatana_Drawn whispers: Sending.. [REGISTER]
[13:51] Zha Ewry: I think that's probably true Jarod
[13:51] Tao Takashi: more client input would definitely be great
[13:51] Zero Linden: (and boundary issues and trust issues sound like relationship councling...)
[13:51] Saijanai Kuhn would join in the learned discourse, but his brain is hopelessly entangled
[13:51] Jarod Godel: Then as technologists, let's all agree DRM will nevr work, and move on.
[13:52] Saijanai Kuhn: so forget about permissions in SL?
[13:52] Sae Juran: i agree
[13:52] Tao Takashi: no
[13:52] Gibson Willis: Isn't it more than just an IP protection issue?
[13:52] Arawn Spitteler wonders if Trust and Boundary Issues may be likened to physical state, where somethings are dissolved in Glass, while others are dissolved in fluid forms.
[13:52] Tao Takashi: but permissions are more a description of your intent
[13:52] Zero Linden: Well.. there are those that say we should just throw up our hands and say... on the open grid, permissions are not available
[13:52] Silicon Plunkett: eh, put it all up to good faith? don't know about that
[13:52] Gibson Willis: It's also an issue of what scripts people want running in their environment
[13:53] Zero Linden: Well, the web does a good job of that...
[13:53] Tao Takashi: and SL is so nice to disable those functions which are not allowed by the permission settings
[13:53] Saijanai Kuhn: in the end all copyprotection schemes are declarations of intent.
[13:53] Teravus Ousley: well, I doubt if permissions were dropped that a healthy economy would even be possible
[13:53] Zero Linden: and the only thing the web protects are server side scripts
[13:53] Sae Juran: on an open grid, security is the biggest issue, that's sure
[13:53] Jarod Godel: Permissions work, to a functional degree within the boundaries of the grid
[13:53] Zero Linden: So - seems like a viable model.... say some
[13:53] Teravus Ousley: even if they were slightly less then effective
[13:53] Jarod Godel: permissions cannot stop anyone from stealing a texture, a model, etc.
[13:53] Zero Linden: Teravus - that is the issue
[13:53] Tao Takashi: but then again there are also advocates who want to get "Show source" removed from browsers
[13:53] Saijanai Kuhn: there have been sucessful RL lawsuits about SL theft
[13:53] Nemesis Exodus: why do people lock their car doors?
[13:53] Zero Linden: but remember, if your viewer can see it, you can copy it..... to a degree
[13:53] Tao Takashi: or aren't there actually browser which this disabled?
[13:54] Zero Linden: and that degree is probably enough to have it
[13:54] Gibson Willis: right, Zero, but I can't push my server-side web script onto your web server
[13:54] Zero Linden: (of course, if you can see a Prada handback, you can copy that too....)
[13:54] Silicon Plunkett: well, even on the web, you have protected code, javascript being an exception
[13:54] Tao Takashi: and there was some discussion on flickr when they changed from flash disaply to ajax display and people suddenly could right-click and save the image
[13:54] Saijanai Kuhn: alll cars have a mechanims to allow towtruck operators to unlock the door without breaking the window
[13:54] Sae Juran: well u can limit some reproduction having a trust server net but u cant really avoid it
[13:54] Tao Takashi: big discussion about this change
[13:54] Saijanai Kuhn: so why bother locking the car door
[13:54] Zha Ewry: As a technologist, I think that we basically can say "We'll mark intent, we'll try to be good and get that into the protocls" and then we'll try to enable people to use existign mechansisms when they get theft at a level which justifies it
[13:54] Ariensky Noel: It can not prevent someone from stealing it, but it takes more skill to steal it now (except textures) than if it was all full perms.. the harder it is, the fewer will do it :)
[13:55] Zero Linden: True - so I don't think scripts go where they can't trust the place they are going
[13:55] Saijanai Kuhn: and those that might casusally steal without even realize they are stealing might not
[13:55] Tao Takashi: and objects can?
[13:55] Gibson Willis: and vice versa - scripts don't go where the place doesn't trust them
[13:55] Saijanai Kuhn: hence the desire to block certain things from the viewer like sculpt maps
[13:55] Day Oh: Permissions do work great for assets... like, I can take your texture, but not the item in your inventory
[13:55] Tao Takashi: of course they need to but nevertheless this seems unfair ;-)
[13:56] Sae Juran: i dont see all that issues with scripts, it's the framework they run on to be trust and secure, never the scripts
[13:56] Sae Juran: just like javascripts
[13:56] Saijanai Kuhn: javacripts are run clientside
[13:56] Zha Ewry: Well, when it's Joe's server on his box, in his garage, it gets trickyier
[13:56] Sae Juran: ok
[13:56] Tao Takashi: the thing is that SL scripts are sold but JS scripts are not really
[13:56] Silicon Plunkett: yeah, lsl is more like php for that analogy
[13:57] Sae Juran: let's say AJAX
[13:57] Zero Linden: Right - so right now, if someone has a clever web too kit that does cooool AJAX stuff, and you use it in your web site
[13:57] Tao Takashi: although you can copy the gmail JS code it's the service which is important not so much the script
[13:57] Zero Linden: the tool kit author just accepts that the scripts will all get downloaded to everyone's client
[13:57] Sae Juran: y if i trust the producer
[13:57] Sae Juran: just like ssl certificates or any software
[13:57] Zero Linden: they dont' worry about permisssions -- they let the legal framework protect that
[13:58] Zha Ewry: I think that's the basic model
[13:58] Zha Ewry: We make sure intent is marked
[13:58] Zero Linden: I'm not saying we should do that - just pointing out that in an AJAXy world, this problem DOES come up for websites
[13:58] Sae Juran: exacly (i totally agree with u and cant see the problem with scripts in fact )
[13:58] Zha Ewry: and do what we can
[13:58] Saijanai Kuhn: but few webbrowsers have easialy accessable javascript IDEs
[13:58] Zero Linden: ? Don't they all, Saij?
[13:58] Zha Ewry: to make it possible for people to manage when they are stolen from
[13:58] Tao Takashi: but on a website there usually is a bigger part of the code behind the web page (e.g. php, python etc.)
[13:58] Zero Linden: And you don't need an IDE - you just get the URL for the .js and download it!
[13:58] Saijanai Kuhn: well not that people are encouraged to use by the website owners
[13:59] Zha Ewry: And.. the other leverage point, would be what license frameworks you put in place to link into the main meta-grid, as it were
[13:59] wendi Ah: dosent this lead back to uuid's
[13:59] Tao Takashi: if we project that onto SL we are back to have scripts running on separate servers again though
[13:59] Zero Linden: Tao - well often, but not true of all those javascript games....
[13:59] Jarod Godel: In the DIGITAL WORLD this problem is pervasive.
[13:59] Sae Juran: lol with js is worse u can surely always decrypt them
[13:59] Jarod Godel: As Cory Doctorow says, "It'll never get harder to copy bits."
[13:59] Tao Takashi: true, Zero
[14:00] Silicon Plunkett: unless you can't access them =)
[14:00] Tao Takashi: but the SL community will not care about all this anyway, they want what we have today
[14:00] Aminom Marvin: I wonder if the more blatant forms of content theft and reselling can be construed as a CS or TOS violation...
[14:00] Aminom Marvin: they could be seen as "false representation" I suppose :)
[14:00] Sae Juran: well if your browser can then u can
[14:00] Tao Takashi: otherwise, see "The history of Copybot"
[14:00] Saijanai Kuhn: there have been sucessful lawusits over content theft in SL
[14:00] Silicon Plunkett: your browser can't see php =)
[14:00] Zero Linden: Like this one: http://www.kongregate.com/games/SimianLogic/filler (there! I've pwnd your afternoons!)
[14:00] Zero Linden: okay - it is flash, but same difference
[14:01] Jarod Godel: The "SL Cmmunity" can't tell the difference between IP rights and an IP number. forget them
[14:01] Sae Juran: i was taking about js silicon, php is a server side scripting, not client side, veerry different
[14:01] Silicon Plunkett: precisely
[14:01] Sae Juran: but sl scripts are with on foot on both :P
[14:01] Jarod Godel: BUT yahoo can see your PHP code when you host it on their servers
[14:01] Tao Takashi: I think LL cannot forget the SL community nor do they want that ;-)
[14:02] Tao Takashi: but you probably trust Yahoo
[14:02] Zero Linden: yikes - my timestamps say 14:01 -- which is the number of my favorite IBM computer (dating myself)_
[14:02] Sae Juran: y the point was in fact that u need to trust the server and the administration
[14:02] Tao Takashi: and in the case of Yahoo the scripts do not move around
[14:02] Zero Linden: but that also means I have my 2pm meeting.....
[14:02] Tao Takashi: in SL they do
[14:02] Gibson Willis: would some of this be cleared up if the scripts themselves were hosted by the person that created them?
[14:02] Sae Juran: like u trust of LL that they dont steal and resell your latest product
[14:02] Zero Linden: I'm going to run.... wonderful as always
[14:02] Ariensky Noel: also SL is a hobby for many, so aiding them with a permission system will be better for the economy, than relying totally on a legal system. If I put a program or picture on the internet I know it is going to be copied. In SL it should be posible to sell something rather easily and not spend 10 hours in cought each week. On the internet I put security locks in teh programs, as SL scripts are very limited it would take up excess room depending on the implementation. A call to a website or a serial code. How would that be for each item you have? ofcause that is doable though.
[14:02] Saijanai Kuhn: take care Zero
[14:02] Jarod Godel: we're arguing "How do I keep the host from seing my scripts?"
[14:02] Teravus Ousley: Ok, take care Zero, thanks again
[14:02] Sae Juran: bye zero, thank you