Automated Object Indexing and Privacy Concerns

Figured I try and get this restarted here.

That burglary thread MUST have been a record.. both in thread length and civility, considering how long it went on. It had its moments, and I contributed to some of it, but overall, it's been "hot" yet civil.

Since Linden Lab is very keen on residents rolling out these types of services, rather than us relying on them doing it all for us, it really is a discussion worthy of forum space, and thanks Strife for encouraging it to continue.

With over a thousand posts, I only noticed a few supporters of the sheepbots activities. For those who somehow missed this monster of a thread..

/327/96/178132/1.html

Just prepare for a long read.

*laughs* it got locked for being too long.

We didnt even make it 100 pages.


---------------------
After all that thread and various other bot threads -

What Id like to see is Linden Labs have some sort of plan in place for Bots.

I do not see unrestricted Bot development of any capability Users can come up with and automate as Ideal for Second Life as long as it remains a virtual world where we are "Residents".

If it becomes an open-source "3d Internet" perhaps that Idea needs to change.

But for right now its the Linden's World - they can at least protect users from the more dangerous aspects of what Bots can do.

If the Linden's stance on (lack of) Privacy is the same as Electric Sheep's intreptation then It would be nice if they clarified that. For a World which is based on the simulation of land ownership the amount of privacy availible seems a little sparse.

Now if the Linden's do come up with a Plan, it wouldnt have to meet my ideals of a plan at all - but SOME plan would go a long way towards showing us that Resident's concerns matter as much as Bot Owner's Concerns.

Here is the principle which I think the creators of the sheepbot failed to consider:

I own myself, my inventory, my objects. The default rule is that I am the only person with access to myself, my inventory, my objects. Someone who wants access to myself, my inventory, or my objects must obtain my permission first in some form.

That is the problem with an invasive system that allows me to opt-out. If a system has access to me, my inventory, my objects before I give permission for that access, then that system has asserted ownership over me, my inventory, my objects. But as we have estabilished, I am in the one who owns myself, my inventory, my objects.

Now, an opt-in system respects the fact that I own myself, my objects, my inventory. An opt-in system takes the proper route to obtaining my permission for access to me- it asks, and waits for my answer.

If developers of information systems understand those basic concepts- that they have no right to access anything for which they have not been first given permission to access- then those developers will avoid invasion of privacy problems.

I think that is really the main point of contention in the whole debate. If they didn't make it "opt-out," there wouldn't have been the outcry over invasion of privacy.

Plus, if the system were "opt-in", there wouldn't have been the problems of things being sold that weren't intended to be sold. Someone who gets to "opt-in" gets a chance to be on notice that his or her inventory could be sold the second it is put up for sale, and can plan for that accordingly. "Opt-in" would eliminate some of those nasty surprises.

All this comes from the principle that a person has ownership over herself or himself, and her or his possessions. Anyone who recognizes that, and they must obtain consent beforehand to access a person, is going to be able to figure out how to develop their projects in a way that doesn't invade privacy. I think it really is that simple.

It should probably get said here on page 1 that there's not much they can do to prevent people from making bots. I know many people don't like hearing this and some don't believe it but it's still true.

Now we can add a bot tag to our profiles. "Don't wanna verify you're not a bot eh, buddy?"

They didn't have to consider it; neither do they challenge it.

The default rule in SL is this: if you or your objects are in-world, you and they are visible to everyone. The server sends presence and other metadata about you and your objects to each and every client within viewing distance of you or your objects. It has to do this because, otherwise, it could not render you or them, and there would be no point. They don't require anyone to opt-in for this, because simply creating an account and agreeing to the ToS IS the prima facie acceptance of opt-in.

The sheepbot does not change ownership of anything about you. It simply catalogs what it sees and shares it, which is the exact same thing that everyone else can see and has access to.

The problem with totally opt-in systems is that they simply do not work for many things. For example, a Name2Key database is useless as opt-in. Who is going to bother giving permission to list their name and key on a third-party website or ten? There is nothing special about the information; it is easily obtainable by anyone, and its knowledge hurts no one. It's not an invasion of privacy, because LL doesn't, by default, guarantee privacy of what you do in SL. In fact, regarding ownership, you really don't even "own" yourself or your objects, according to the ToS. The only thing you do own is the IP rights, not the representations themselves, nor the data used to represent you or your objects.



Beyond the fact that there is nothing being discussed which challenges your self-ownership, I definitely am not going to ask you permission to see you or your objects and wait for your answer before you rez on my screen; to me, that is what this notion is tantamount to. Yes, it is an extreme interpretation, but what we are talking about here is limits and boundaries, and they can arbitrarily be set anywhere from lax (as we have now) to extremely rigid (which would make SL totally useless to anyone). Personally, I don't want to go to LL-mandated opt-in requirements, because it would necessarily proscribe things that I actually would like to see get working well and advance the utility of SL (and, yes, that includes things like search engines).



Except that there has not been any bona-fide "invasion of privacy" as a result of the existence of the sheepbot. Even the original poster's situation in the other thread was not an issue of "privacy". He had employees set out objects publicly for sale in a publicly-accessible sim and they were that way for at least the better part of a day, if not longer; long enough for the sheepbot to happen by, catalog them, people to see them on the search engine, teleport in, find them, and buy them.

LL provides the capability to obtain some modicum of privacy; its price can vary from free (though not from LL) to the cost of a private island. I know the "Let them eat cake" argument will come up again, but it is irrelevant, because it has little weight in the case of SL. We don't HAVE to be here; it is not important for anyone's survival. SL itself is a luxury, not a necessity, and trying to equate it to one is quite ludicrous. Privacy exists and is there, for anyone willing to expend the money, time, and/or effort to obtain it. In many cases, it can be had for relatively cheap.



I personally don't see this as a problem. The "For Sale" controls were meant and intended to "put things up for sale to anyone". There's no confusion about that, even when people put themselves at risk using them as a way to transfer ownership of an in-world item to someone else. Engaging in risky behavior, using a function in a way it wasn't meant to be used, with full knowledge that someone could appear at any moment to buy it out from under the participants in the transaction, is on no one's head but the seller's.

I know people use the function in a risky way quite often, and there IS an actual feature request on JIRA to give people the ability to sell to a specific avatar to make this process less risky (or to eliminate the risk altogether; I also note a striking lack of votes for it from some of the most vocal opponents of the sheepbot; go figure). However, as the length of the other thread proved, the issue is bigger than just this instance. I severely doubt a middle ground will be found, because everyone's argument seems perfectly reasonable to them. However, there is too much overlap in the disagreements for a middle ground to form. No matter what, though, LL will do (or not) whatever they will do, and some folks will be unhappy with it.

I didn't see this thread when I made another new thread for this:

/327/3b/184573/1.html

coco

Im not sure middle ground is really even the right term. Becuase the Lindens are already on "your side" of the issue. In that any statements they have made publicly have been in favor of Sheepbot, and the TOS/CS does not have any provision that would deny its control, much less use.

Furthermore thier approach towards all automated client programs is to ignore any abuses until such time as they cant avoid it.

Landbots being the perfect example of a serious problem being ignored. Copybots being the case where they couldnt ignore it any longer.

I think this approach by the Lindens is the wrong one. But it seems obvious its the course they will be taking.

So its not a search for a middle ground, Its more that -
Theres some of us whod like to change the status quo, and others who really would prefer the status quo remained.

Its more a question is will the Lindens even bend a little bit.

They didn't really take much action against CopyBot. They simply reiterated what was already stated in the ToS, and that they were going to actively enforce gross violations of it with regards to CopyBot. The changes they made to the protocol which "broke" the existing CopyBot were already in the pipe, and needed to be done, whether or not CopyBot ever existed. Perhaps they moved up the timetable for deployment a bit, but that's not really a huge reaction to the issue.

The second point is that CopyBot had the potential to affect a LOT more people. How many people have stuff set out for sale that they don't mean to sell, compared to the number of people who have stuff set out for sale that they really do want to sell? What is the percentage of people who could be negatively impacted by SheepBot versus the number who could be negatively impacted by CopyBot? I think that also drives the decision process. Numbers do matter.

If I were the Lindens, I would be expecting to see a more universal outcry, and at least some amount of middle ground between all sides to come down into and take action on behalf of. Until one forms, there is still much to be considered and too much conflict between the sides to make a good decision over. Thus, I kinda expect them to maintain the status quo (they DO have many bigger fish to fry, IMO) until it does, and rightfully so. Rash, half-thought-out responses to community issues serve no one.

Considering most people dont even know the searchbot exists - Universal outcry isnt going to happen. A large amount of people would have to know about it in order for a large number to complain about it.

Most people dont even know Landbots exist.

A lot heard about Copybots becuase Sellers paniced, which considering their IP was at stake I dont blame them.

JIRA has proved very cumbersome and difficult to use for many people. I had to get educated about it too. The number of votes on it is more an indication of the obstacles to using it than anything else. And there is a report elsewhere on here that it is borked altogether atm.

A lot of people complain about JIRA being hard to use. It's as simple as:

1) Going to the http://jira.secondlife.com website.
2) Clicking on the log in link in the upper-right corner.
3) Entering your user name and password (yes, it is the same as your SL user name and password; just like the forums, just like the Wiki, etc).
4) Going to the specific bug/feature request.
5) Clicking the Vote link.

If you can navigate the forums, let alone SL itself, you can EASILY navigate JIRA.

..and no, it is not borked right now. It was the other day when the server was acting up, but has been working fine since.

For the record, here is a post to the previous thread intended to sum up the main issues pretty well:

Originally Posted by Kyrah Abattoir
Well to take their defense I do not feel like i have to ask permission either to make data mining.


Hi, Kyrah, welcome to the thread. Actually, your observation about whether you "have to ask permission" is not so much a defense of ESC as a partial explanation of why there is a problem in the first place. A problem, by the way, that reaches far beyond ESC and its malignant little searchbot.

If a would-be data miner did "have to ask permission" there would be no problem here. Any problems would be strictly between the miner and the people who gave permission.

In the present case ESC released a searchbot in-world essentially scanning avery Resident-made object in it and collecting those where the "for-sale" box was checked. ESC then posted these with landmarks set right in front of each object on a website separate from SL and LL, claiming it was a search service for in-world buyers. ESC did all this not only without asking anyone's "permission" but also WITHOUT NOTIFYING THE OWNERS OF THE OBJECTS. As a result, owners found themselves relieved of valuable objects from inside their homes and properties for nominal or zero prices by unscrupulous people when the owners were innocently unaware that the for-sale box was checked, or were builders using a workaround that had been safe common practice for years and had suddenly and without warning made them vulnerable. ESC's response to complaints about this effect was profound contempt and condescension - and for quite some time not the slightest inclination to fix or remove this bot. (And by the way, the bot cannot read the contents of vendor devices, so misses maybe 90% of the objects legitimately for sale in SL; it is almost useless for its stated function.) Whether by accident or design, the result was not merely data-mining, but predatory data-mining, and from the moment ESC knew about this effect, consciously predatory. It is evident that from their point of view, The Electric Sheep Company are the meatpackers, and we Residents are their sheep. One ESC officer has been quoted in a publication as speculating about "life-logging," meaning basically tracking every move an avatar makes as well as every object it owns or disposes of. I will leave it to your imagination what other harmful (to Residents) uses spybots could be put to by exploitive or malicious people. I have thought about it, and some of the possibilities would have a deeply chilling effect on interest in SL by individuals and businesses alike. Some of the possibilities are just plain chilling.

For us, the questions now are, what can or should LL do about the problem ESC's actions have exposed, and what can or should the Community Standards be to deal with the possibly predatory use of spybots, and what can or should Residents do to deal with what many see an unwarranted and harmful intrusion into their SL lives and sense of privacy, and what can or should "privacy" consist of in SL, and how can such privacy be guarded?

For LL, there are other questions, such as, should they be allowing outside entities to be making such dubious use of data on LL servers which LL, after all, legally owns, and which might someday actually have monetary value to marketers and others? And, at what point might LL lose the legal right to control their own data through failing to take action to assert and protect control?

As you see, Kyrah, there is a lot here for intelligent people to work on; a lot of questions about SL that need intelligent answers. I hope you will help work toward those answers.

Or, to sum it up in fewer words:

No one needs our permission to LOOK at our in-world stuff, (unless
we've put it on a restricted access private island.)

So, ESC's searchbot looks at stuff, remembers what it's seen and
tells people where to find things.

ESC seems careful not to cause unnecessary load on secondlife's
resources, they seem to respect the DMCA and they seem to be
complying with the ToS and CS, which are the laws and ethics that
linden lab has set down for residents.

Whether or not search & listing internal content on an external site
is an intended use of secondlife is something that linden lab should
take a stand on. Likewise, some position on the nature of our
in-world rights to privacy would from linden lab would be helpful.

well - Its looks at ALL of Everyone's Stuff it can get to.

And it will Systematically and periodically look at that stuff.

Thats the scale and scope difference I meant. Theres a difference between that and People wandering by and looking at your stuff.

And Electric Sheep hasnt 100% decided what it will do with all that recorded information. Though Forsetti is working on that- It sounded like through an offical post on ESC website.

So, everyone who wants to maintain a modicum of privacy against ESC - change the descriptions of aqll your stuff to something as misleading as possible (have fun with this : D), get your sim owner to ban any ESC bot possible, tell your friends to do the same - and post ideas on further disabling this sort of intrusion on these threads.

(nod to Colette above) And I'm keenly interested in what policies ESC announces that they will be operating under.

My secondary interest in this issue is watching certain residents acting as if they have the right to set policy and demand compliance, as if linden lab had abdicated their role as policy-makers and enforcers to them.

The day may come, where democracy establishes our rights in secondlife not the whim of linden lab's upper eschelon. What a glorious mess that will be.

Here's what's going to happen with the various scam-bots [ESC sheep bot included].

LL will continue to ignore the problem.

Sooner later one of these scam bot will hit the WRONG victim - a relative of a powerful politician or someone with inside access to the mainstream media.

A firestorm of negative press will insue.

LL will back pedal like crazy, saying this is CLEARLY a violation of TOS, that LL never had ANY idea such things were going on, take draconian and over kill steps to solve the problem, and proceed merrily on their way.

Just be patient my friends and stock up on popcorn. This particular train wreck will be WAY too good tomiss!!!

On the web, you would either : throw a .htaccess file restrincting access to a user or a group of users; or : move your data outside the realm of the server.

The second case is buying a Private Island.

The first case is implementing a kind of banline which would restrict streaming any content inside some bounds to an access list. For other users, everything inside these bounds would be an empty hole.

Technically, it may be more easy to sell "mini-PI" or low-prims PI (aka OpenSpaces), running many of them on a same server

Susanne you and Talarus basically are basically thinking the response will be along the same lines. However he sees it as a possitive and you see it as a negative.

Its obvious that this is the approach they are using. Basically wait till theres a real problem then fix it after the fact.

Putting a proactive structure in place to anticipate some obvious problems would be a more organized way to do things.



A former Employer of mine used to love to point out the Capability Maturity Model and apply it to all sorts of things - not just Software development. http://aelinik.free.fr/cmm/tr25_o2.html#D241 for those interested.

He used to like to say, in a derogatory way,
- "Disorganization is an Organization style, too"

Well, so far the only "problem" that has become evident is the paranoia of a fringe group of residents that don't seem to understand the concept of making something "for sale" or "NOT for sale".

This isn't copybot. You get from the sale the amount you set when you put the item up for sale. The system was designed this way. No one is stealing anything and the information being obtained about you is pretty much inconsequential.

If anything I think LL just sees a bunch of whiners that will never be happy with anything other than paranoia and discontent.

One of the only reasons its a fringe group of residents - is becuase only a fringe group knows about it. I could say that only a Fringe group of residents supports the searchbot too.

The Bot problem is bigger then ESC's search bot. Linden Labs has no process in place to anticipate problems cuased by Bots except waiting until they are big problem.

Many of these points were raised and shot down in "that other thread" btw.