|
|
|
"special client software" - - - that is the key.
30 to 70 avatars in a skybox above 500m, on ONE SIM, hours and hours, run by "special software", logging in from ONE computer, using ONE IP-#... why are they mostly ruth? there should be a technical solution. i am not a technical expert, but i know enough about SL to be sure: this is possible. Detect special client software - LL obviously has difficulties even in detecting malware-clients like ShoopedLife. So it seems if the client creator doesn't want his client to be recognized, it's not yet possible.
The other things you propose like "lot of avatars with ruthed shaped in a certain heigth using one IP" seems to be a bit fuzzy .And with these criterias you cannot detect not ONE bot alone, only a group of bots. And if a "bot" detected to be a member of a "bot group" logs in the next day as an normal avatar controlled by an human with a normal LL client, is he still marked as a bot? All these criterias you propose seem to be a bit fuzzy and would need to implement some kind of artificial intelligence in SL. These algorithms could fail in certain situations and don't seem to be failsafe. Bots are often "Ruth" because their Master just has created a new Basic Account and did not do anything to change that default-Avatar.
And there simply is no way to distinguish between bot and nobot because you can use any avatar with that bot software. You even can think of a viewer that can switch between viewer and bot - that viewer software and the necessary protocolls are open. Why not keep it simple?:
Every client connection (regardless of the client type) is logged with the source IP address. IMO this a good approach to distinguish between residents and bots. Bot farms are usually running on the same host and all bot connections (of the same master account) share the same source IP address. A possible solution could be: At login, check whether there already exists more than lets say 2 to 5 connections with the same source address. If so, flag the new agent/avatar login as a bot connection and disable traffic accounting for it. I know this isn't a 100% solution but I think this way you can easily identify the vast majority of existing bot farms. Then all the families running SL on multiple computers on their home network are flagged as bots. Also viewers such as ShoopedLife use a proxy to change their IP address.
Lots, it is not unusual nowadays for every family member to have their own PC. Husband, wife and 18 year old child all having our own PCs and SL accounts, doesn't sound unusual to me. I personally always have my main account and alt account logged on to two separate PCs. And this would not distinguish between running multiple clients on the same PC, which lots of residents also do.
Besides doing this would only make the bot programmers use a proxy to avoid detection and bot farming would continue. How about the MAC-address? This solves the problem with families or neighbours sharing one WLAN.
You don't have a good reason to run multiple accounts on the same computer at the same time, do you? A Proxy would produce 30 different IPs for 30 different clients running on the same machine at a time? /me is really non-geek, sorry for simple questions I don't because I use two PCs, but if I did not, I would run my alt and my main everyday in multiple instances. One of the main reasons is the group limitation of 25 is too limiting for me, so my alt is part of groups I have no room for.
Pretty sure a proxy can give both a different IP and MAC address every time you use it, but could be wrong. But if you are talking more then 5 connections, then this would probably handle most cases of families and normal users running multiple instances, I just think 2 or 3 connections is too low. Unless you are using bot software too many instances of the normal SL client running on the same PC would bring it to a crawl, if not crash it. The way networking works, the MAC address is not actually exposed to LL as part of the communication channel, like the IP address is. The only way LL was able to ban by MAC address in the past was because the client voluntarily sends the MAC address to the SL servers. It's utterly trivial to send a fake MAC address; just use ShoopedLife.
"i am not a technical expert, but i know enough about SL to be sure: this is possible." The thing is that I am a technical expert, and I think Harleen is too, and we're going to tell you: it's not possible to consistently and reliably detect robots. Either there will be a loophole that every bot owner will learn to exploit very quickly, or you'll end up calling a lot of real people "robots"... probably both. Just checked this ShoopThing - - - NOBODY with an IQ >30 would trust this guy and give Account Data over to him
 Seen his definition of SL? "Second LifeTM (A trademark of Linden Labs etc) is a shitty MMORPG put together by a company made entirely of monkeys" And really, multiple clients simultaneously running on one computer will get multiple IPs when using a Proxy? ShoopedLife is just an example of what I was saying. Anyone can download the source code of the SL client, modify the part that sends off their MAC address, and there you go: you can't rely on MAC addresses anymore.
Proxies allow people to send their SL traffic through third parties elsewhere on the internet. As far as SL's servers can tell, the traffic comes from the proxy. There're plenty of proxies around, so yes, every bot will have a different IP address, even if it's actually being run on the same computer. In short: any rule you can come up with to detect robots can be easily counteracted by people who don't want you to think their robots are robots. That, or the rule will be so general that it'll get a significant number of non-robots. If technically possible, I'm all for banning bots from appearing in the map. But reading the posts, I get the impression that it's not so easy. Are bots run from a special client software? I cannot run two SL clients on my computer at the same time. And if so, bots still must have a registration... so does that mean bots really pay that monthly fee everybody has to pay for an ALT avatar? Or do they register their bots under false user names? Maybe if Linden would introduce some kind of verification, then only a real person could register an account, and running bots would either involve a lot of real people, or make it quite costly.
@moni, eventthe term "bot" is undefined, that means first somebody should give a technically definition what has to be regarded as bot.
So called "bots" normally run on special clients. But for sure, you can run normal SL clients multiple on your computer, and if you do so and your AV and some alts hanging loose somewhere while you are at work, then one could say, they are bots, too. AFAIK it's till some month not necessary to pay for ALT accounts. And to be honest, i don't have the impression that LL is interested in doing any efforts to avoid any bots. I am sure that this jira thread will never see a Linden participating in the discussion, they try to avoid the discussion about this topic. Citing from "Bot tagging", a technical draft about tagging bot software in VW:
============= Anyone who runs a bot should provide a bot tag. It is only fair to Virtual presence avatars are more engaging than nicknames in a chat Disguising a bot as a person is plain fraud. It can be regarded as I am aware that the userAgent string of the current libsecondlife implementations is easy to manipulate, and I'm not sure if there is any durable, reliable way to detect dishonest bots. Society's Rules and Laws are usually definitions on how one should act. It's not the role of the lawmakers to decide if a rule can be broken, and how hard it is to do so. Saying: "This law could be broken by criminal elements, so it's useless, therefore we won't even enact it." shows a serious lack of moral maturity and understanding, but it's not an reason to keep important things unregulated. Off-topic, but related: If all users would have to have payment info enabled, or any other way to track their RL identity, each additional alt would cost the shop owner USD9.95, so a typical traffic bot herd would be USD 388.05. Actually it does cost that amount anyway, but it's not easy to track nowadays. ("If we find that you've been creating an army of alts, we may charge a small fee of US$9.95 for the creation of each additional Basic account as a way to recoup some of the cost of maintaining the accounts for you." https://support.secondlife.com/ics/support/default.asp?deptID=4417 Now, I think Torrid's onto something here. My argument above is this: to rely on a purely technical means of detecting and marking alts is infeasible and will result in both false positives and false negatives. On the other hand, policy-based regulation with human enforcement is a much more realistic way to deal with a situation like this. I bet a good portion of bot creators will gladly cooperate with a self-tagging system to mark themselves as bots. If the ToS or Community Standards said that doing so was a requirement, then LL could enforce this manually through the AR system. It's not possible for software to reliably detect bots, but humans can often do it.
I am very sorry about the change of the virtual Wolrd. I am sure that mass of bots are not a good base for a growing second life. I hope you will through them out, equal the method of cleaning the world.
It is possible now to detect bots that are on a non-standard viewer. Those owners that want the bots off thier land are able to use this method to detect anyone who isn't on a standard client.
Using Second Life Client Version 1.18 a person who changes the advanced configurations outfit composite limit to 0 and approaches an avatar will either see a skin loaded from client side or a Grey skin in about 30 seconds. Anyone found to be using a bot program is more likely to not load textures from their client side catch, and it is this what makes them distinguishable from regular avatars. It is still possible to use this outdated client by putting -channel test in the shortcut. In the next release of second life, they should bring back the ability to force textures to be loaded from client side so that a real method of bot detection, or at least non-standard viewer detection, can be implemented to the public. Ralphie, that would be in incredibly bad and downright idiotic idea. Many residents (myself included) use non standard viewers and never use bots. Nicholaz's and CoolSL are two viewers that each have hundreds if not thousands of users in SL and neither are "bot" clients and have many advantages to the official client.
Do those viewers allow for the texture files to be downloaded from the client side?
If so then re-allowing for composite limit to be set to 0 would not effect these clients, and the method listed above is only listed as a basis it will take more interaction than just that to determine if someone is a bot. I haven't looked into it but I'm guessing that they do by default but so what, there are still text only clients like Sleek that are used for legitimate purposes. You can't illegitimize an entire class of client (3rd party) just because some people use a small number of the clients for nefarious purposes such as texture theft or in this case botting.
I'm not for banning bots outright but I am all for giving people the tools to see what's a bot and what's not and act as they will. If you are going to do that though it has to be done smartly and in a way with no false positives and no overly broad assumptions which is what a small minority here seem to want. IF they res textures from client side they wont' be effected, its directed more towards those bots that are ruining the second life economy by not putting the linden back into the game and extracting it for their own personal gain.
I wonder how much of that 160,000 daily usd is taken out by just those who manipulate the market. I'm a monitor in a combat sim, and there's only one thing I like about camping bot - they never whine when orbited. They also tend to wear a pretty standard suite of protections, including anti-orbits. We like people who chat with us in the sim, so if they don't talk, won't accept a free translator, and seem to be ignoring being orbited, we ban them.
People operating bots in their own sims would seem to be more or less their own business, as long as they're not lagging out a shared server, and that should be covered by a standard "unfair use of region resources" AR. What we really do need is a way to restrict bot use from being taken outside the bot owner's sim. I'm totally with Lex Neva on this one. Combining a "bot tag" with an actual regulation is much more feasible than relying on software alone to distinguish "bot versus person."
This proposal could thus be changed to aim in aiding LL's current policy change decision regarding the bot issue by stating: 1: Have bot-only accounts tagged both on profile (under account type status?) as well as differentiated on the map. 2: Amend the ToS and (possibly) CS to include a requirement of accounts solely used as bots to be identified as such (as well as providing "parent account" information) either during signup or via a "Special Questions" ticket. 3: (Just an idea, but could be practical...) Add a "Bot" tickbox option which would grey-out the list of available surnames on the signup screen. This tickbox option would then trigger a requirement of the above (2) if selected, as well as make it completely clear as to whether or not the avatar in question is indeed a bot by simply looking at its name. Also to avoid confusion in the case of the "greying out" of the surname list isn't enough indication, add a dialog asking if the user is certain they wish to register as a bot. I think that would be a bit easier on all sides of this issue. I would propose to add and option to have an
"bot setting" for theese accounts in depending on the function limiting the impact they have on SL like if it is an bot thought to answer IMs and invite people into group/a;llow changing of group functions then not much graphic input is not needed and the bot self could get a generic semi body shape using low ressources on other people and making it distinguishely "bot", same for shop owners using them as showcases the bot does not need to listen to chat for example, in cases like that having them as "simply rezz an avatar mesh to put clothing on " option would suffice and reduce stress on SL. I am not smart enough to workout the details but voluntarily limiting the ressources you need for your bot would be interesting, it is in the bots use interet too for most are used aqt own place sim so reducing ressurce use would be an used item. |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
New Feature
Open
Major
Make Bots distinguishable on Map (different colour)
If you want bots to be marked on the mini-map, what way would you propose the client viewer or server software should identify avatars/agents as bots?