Bots and Second Life

This should give everyone alarm and reason to pause to think for a bit.

“When calculating the online now number,” Jim told me, “we do not filter out users/bots who are logged in through thin clients like libsecondlife, but we know through closely monitoring the user data, that they represent a very small percentage of the overall number.”

“By design,” studio director Jeff Linden added, “the exact version of the viewer is not integral to the login process, just as different web browsers should still allow you to login to your bank’s online site. Viewers do voluntarily self-identify (like browsers do), but this is not authoritative, so we can’t be 100% sure which versions are connecting. Because of this, it’s difficult to provide an estimate of what percentage of the ‘online now’ number may be bots logged in… That said, anecdotally, we’ve noted that connections made with clients that self-ID as libsecondlife are generally very brief, and very few of these clients are logged in at any given time.”

Interstingly enough linden Labs posted, per Zee Linden, some new data.

Concurrency. Peak concurrent users, shown in the red line on the chart below, grew 12.5% in the fourth quarter to more then 58,000 - up more than 210% for the full year. We’ve seen a growth in concurrency almost every week since the beginning of September. Growth continued last weekend when concurrency grew another 5.3% to 61,500.

Data is Data

The problem with this data is, and I commend Linden Labs for releasing it, that it really doesn’t tell me what I need to know. Actually in December I need metrics from post Novemeber to end of December if I am to make meaningful comparisons to virtual and web-based shopping behaviors.

Zombies Walking Around in Sims

Timeless Prototype notes the SL Polymorphism Proxy with Imaginary Friends for SL. The developers plan on including support for animating the proxy, saving the local users appearance preferences, tweaking controls, etc. From Timeless’ site…

In the true sense of the original meaning of hacking (not malicious), Thoys Pan has achieved over 100 avatars in a simulator. But there’s something ghostly about it all. Concept by Fw0rd Utorid. Thoys very kindly demonstrated to the technology to me, see screenshot attached to this post (credit to Thoys). The code is now open sourced, and the project is here

(http://www.secondlife.com). The product enhances the user experience by providing user control over avatar appearances, and further by allowing the placement of avatars anywhere in the virtual environment. Both of these events occur only to the user running the software.

and here you go: Scene from the living dead or that Shaun movie thing.

Metrics & Web

Jeff took the topic to a broader perspective, noting that this is a problem with measuring general Internet usage. “It seems the bigger question is: how do we know that Second Life isn’t filled with bots and devoid of real human activity?” As he puts it. “Turns out this is a statistic problem that’s well solved by the web world– we can determine this in much the same way that a website would determine that visitors are real users rather than automated webcrawlers.

I would not go so far as to say it is a “statistic problem” nor well solved- I would have to say it is quite the inverse- ask the IAB. Especially if you want to get down to the nitty gritty and get people to understand the difference between say- crawlers, bots, botnets, agents, virtual botnets and some of the really damaging things some of these things can do. Of course not all agents are bad- it is all in the eyes of the beholder. For example is Cyveillance’s agents good or bad? It all depends on the relationship, need, and value delivered.

The Obvious Lesson

I have tried to make this post as short and concise as possible. Second Life leads us to consider the death of old metrics and more meaningful new metrics- good- it needs to go there. I am less concerned about “traffic” and more concerned about “engagement”. Thanks to Sam for pointing out my annual rants about “clicks”. Not that bots cannot be engaging…but concurrent logins simply does not cut it as a meaningful stand-alone metric. Plenty of conclusions can be drawn, but without looking at the type of agent logging in, its activity and posting this data- one really cannot draw a meaningful “anything”.

Much more could be said, but I am trying to keep it brief…

Popularity: 3% [?]

Share This