Land Metrics Forum Transcript

Introduction

Glenn Linden: Hello, everyone and thank you for joining us today. Today's session is about Land Metrics. Our data team has been working on a project that enables us to caputure and display data without costing cycles on our main database (which is essential for anything to scale). Siobhan Linden from our Data team will be joining us shortly to give you an update on our project and to get feedback on what will be most useful from you.

We have two objectives in this meeting: 1) to get input from you as landowners or as people who work with businesses that buy land on what metrics are important 2) how we can implement these metrics in a way that enables external businesses around providing metrics-related services

As part of this project, we're meeting with you today to get your input prior to initiating the actual work. Many of our landowners expect LL to provide metrics.

Evis Blackflag: interesting. A standard of metrics would be great to level perceptions on effective spaces.

Ancient Shriner: You have three other companies here that have done the same - Mine, Maya, and Second Analytics Kim Anubis: Mine too Ancient Shriner: ok, Four Ancient Shriner: I may speak for all of us that it is my hope that LL doesn't intend to encroach on our collective businesses Neil Robinson: well we dont know what they've done yet because he hasnt told us fully LOL Glenn Linden: (that's why we're meeting with you to get your input) Kim Anubis: Nah, I hope LL will do it, Ancient. Ancient Shriner: I don't and I don't think Hack or Kriz do either

Metrics Requirements

Glenn Linden: While Jennifer can provide more details when she arrives, I can start the first partt of the discussion now. The metrics that we consistently are asked for (and, Ancient, even by some of your current customers) are avatar visits (unique), repeat visits, and time on the region.

Sitearm Madonna: it just so happens there is a service request on this .. several i guess ...including suggesting a feed that owners (and developers) could access - a basic easy report and a feed for fancy metrics Web 261 - Provide a Daily Visitor Tracker Public Data Feed to promote increased Second Life development by providing detailed statistics and proof of traffic for feedback on success of sim designs and benefit clients (investors), developers, visitors and Linden Lab. Glenn Linden: many requests in pJira (Issus Tracker) Sitearm Madonna: what glenn said, was spelled out in that one, for example avatar visits (unique), repeat visits, and time on the region.

Glenn Linden: So, the first question has to do with what metrics are valuable and what frequency should they be provided. Evis Blackflag: It would be sweet to be able to rss real time data Kim Anubis: I would love to see LL do what my company's tool does -- and it captures that data. It also tracks any area of a sim you want tracked, or during certain times. as well, it chases down new visitors and gives them a copy of the privacy policy -- that is vital. Ancient Shriner: Who, What, When, and Where Kim Anubis: I think LL can do this more efficiently than we can. Hackshaven Harford agrees kim

Charlotte Bartlett: Sorry I am not getting the value here- we can track visitors inworld ourselves, heatmaps etc - I use those tools already from third parties. How does this help development? Surely we should first define what is the critera of success for a development and what metrics do we need to measure it?

Evis Blackflag: If I may state, that certain segments view third party tools as overly taxing on sim performance. Kim Anubis: But if you don't include all of the features we've got, then we have all sorts of redundancy and patched-together stuff going on. Charlotte Bartlett: indeed, but me showing x brand 70 concurrent users is hardly going to sell the platform.

Glenn Linden: Region area data is easy, parcel-level may be harder, but exact location we can't do right now.

Neil Robinson: In terms of what I'm asked about by clients = charlotte touched on it, stuff like how many unique visitors is good but things like heatmaps are much more useful, we can get a rough idea already of how many visitors (ctd). What would be rally handy is to be able to see how they are interacting with the content - does user X like this bit more than that bit, do male avies like the rollercoaster more than the carousel, whatever Sitearm Madonna nods.. Neil Robinson: tracking users by frequency is great but if we could capture more data than that to apply it to the actual experience design - then we can go further in terms of ROI to our clients by showing them they're hitting their target audience Glenn Linden: by interacting, do you mean click/select or just location? Neil Robinson: and also refine our projects as they are mostly ongoing Evis Blackflag: UUID touched <object>? Neil Robinson: things like, are we more popular with new acconts or old, etc so we can gauge, are we bringing in a new community here? or dealing with an existing one. I was just saying being able to track deeper who is coming to our projects not just their numbers would be super handy and its what I get asked about a lot

Data & Privacy Issues

Glenn Linden: Back to Neil's question - about being able to track touch events, that will have to wait for Jennifer, I don't know what's possible.

Ancient Shriner: you need numbers first, then comes analysis

Kim Anubis: You can code that right into your content

Neil Robinson: really? avatar age and gender and stuff?

Kim Anubis: age, I think

Glenn Linden: Neil, age and gender are possible, as long as they can be aggregated to more than 5 accounts due to European privacy laws.

Hack Richard: Before we decide on the set of data items, we should have a mechanism for setting the level of privacy on a parcel

Hackshaven Harford: right then, so in response to Glenn's statement that things like positional data cannot be tracked, is the reason technical or legal, IE to privacy implications have to be worked out first, or is this more a too many other bugs to squish issue. Same goes for object interactions, etc.

Glenn Linden: The privacy issue is that if only one avatar comes to your region, you can track gender and sex to that avatar.

Kim Anubis: Only if you capture the avatar name or someone is there watching

Hackshaven Harford: so correct me if I'm wrong but until these legal issues are address linden lab will not be providing this data?

Kim Anubis: in which case they'd know.

Evis Blackflag: with 3rd party tools and medium reporting you could probably determine that anyway

Glenn Linden: And we can't prevent someone from capturing av name when they come to a region.

Evis Blackflag: Seems like only average data would satisfy requirements

Hackshaven Harford: so I know my time is about up, but Glenn, it seems a lot of the data we need is ropped off from LL providing no?

Glenn Linden: Hacks- first we want to find out what you'd like to know. Certainly, country is a high request.

Ways to Obtain Land Metrics Data

Glenn Linden: I'd like to introduce Siobhan Linden, who is working on the project in our Data Warehouse team. Siobhan, can you provide a brief overview - we've spent some time talking about requirements.

Annie Milestone: Sorry, I missed the first part, maybe smeone already said - but what would be important for many, I think, is to be able to find out, which country a visitor is from

Ancient Shriner: you can do that with IP address geolocation and media playback logging Annie

Annie Milestone: Yes I know, but that would mean that someone would have to collide with a prim

Ancient Shriner: yes, you only get about 80% of the traffic that way

Kim Anubis: Don't need to collide with a prim

Siobhan Linden: So on Evis point, about only needing average data, would that be daily averages or hourly averages?

Siobhan Linden: And is real time data of value for monitoring region activity

Evis Blackflag: I've love to have the option Siobhan, I could then break down which event(s) had relevant visitors

Evis Blackflag: I'd*

Siobhan Linden: Would a web page filterable by various parameters be sufficient, and a csv feed possibly,or would folks desire an api to capture a stream of data from for

Sitearm Madonna: "yes" (all three would rock) (and give developers options for services)

Evis Blackflag: yes all 3 :)

Hackshaven Harford: API please :)

Siobhan Linden: sorry for just jumping in here late, I am probably asking these questions out of context a bit

Kim Anubis: I'd like to see LL take it as far as you're willing, so that we don't have to do a bunch of redundant development

Sitearm Madonna: to me the issue is, "LL can provide more accurate and less laggy data collection systems than external developers having to build (sense) on top of the platform"

Kim Anubis: Yes, Sitearm

Sitearm Madonna: give us the data so we can analyze/metricize/etc

Hackshaven Harford: agreed sitearm. I'm willing to bet that LL can design a more efficient system than any of us

Neil Robinson feels they could too

Ancient Shriner: I don't

Glenn Linden: Ancient, our data provision would not have lag or stress on the servers.

Ancient Shriner: you are putting a read in the stream Glenn - that read will cost cpu cycles

Glenn Linden: We don't want to be in the analysis business - many of you already do that.

Kim Anubis: I also think about the privacy issues, particularly working in the Teen Grid. LL is in a better position to maintain privacy than we are.

Annie Milestone: What I also find very sad is the limit to 16 avatars at a time

Sitearm Madonna: Siob, where you came in, we were on the topic of whether privacy issues prohibited LL from providing the data

Ancient Shriner: I don't see any way around that

Siobhan Linden: we would be taking feeds off of the backend of the systems, more like logging feeds, and plumb them into making that data available.

Ancient Shriner: ahh, then you WON'T have the date I have. That's fine. I have much more than that . . .

Kriz Janus: I think the thing is not whether offering an API or not but WHAT is made out of it. Its the user interface which has to be easy

Siobhan Linden: in the end, adding instrumentation could track most anything. But in reality, we have to be careful of how far we go and how we do it.

Sitearm Madonna: from one of those service utility requests, Sio, could your back feeds provide this? quotes

Sitearm Madonna: "Provide public data feed for resident and developer use which contains: . For each 24 hour period . For each parcel: + Visiting Avatar Name (Agent ID) . For each visitor avatar name: + Date, time and duration of stay . Linden Lab would provide the data feed and a rudimentary parcel report similar to the economic reports. . Residents (parcel owners, sim owners) would be able to access the reports . Developers would be able to access the data feed and run custom reports and/or make salable visitor tracking and reporting products for residents to purchase. "

Sitearm Madonna: ok.. the only thing is.. for the volume of development business we need in sl, i think we need something common from sl for all sims

Glenn Linden: Siobhan, Neil asked earlier if we can track touch events

Siobhan Linden: we could yes. But note, we are not necesarily looking at adding metrics to put any in world businesses in jeapardy. We would prefer to offer enahnced data where it makes sense for us to do that.

Types of Land Metrics SPs are Interested In Obtaining from LL

Hackshaven Harford: perhaps it would help if we first list the types of information we are interested in and then LL could tell us what of that they legally have the ability to collect say a year down the road. I'd be happy to take a stab at the list, unless someone else would like to

Annie Milestone: good idea

Siobhan Linden: A list like that would be very very useful.

Hackshaven Harford: fair enough, I'll start rattling -- keep in mind this is data type not analysis Position x,y,z, rotation x,y,z, camera position x,y,z, timestamp, avatar status (idle, busy, sitting, etc), sim health (FPS, time dialation, etc), profile information, groups, birthday, pay status, etc., object interactions, sit, touch, pay, etc., client IP address. Ah yes, avatar name of course

Ariadne Recreant: I'd like to keep track of websites guests are using when they are on the island

Evis Blackflag: 1. # of Unique Avatars 2. # of visits per Avatar 3. Country of Origin 4. Total stay time per Avatar 5. Avg stay time per Avatar 6. Most visited parcel per estate by Avatar 7. Daily breakdown of the above data. 8. < Parcel level breakdown of above data>

Ancient Shriner: and when you have pay, how much as additional data

Hackshaven Harford: sure ancient, all the object interactions come with associated data

Ancient Shriner: also mapping of avatar/alts common locations

IP Address/Geolocation

Neil Robinson: ip address?

Annie Milestone: IP address (not on media stream) over lsl

Ashmuel Gould: absolutely not

Evis Blackflag: what would we do with IP address?

Hackshaven Harford: country of origin

Kriz Janus: geolocation

Hackshaven Harford: and ... firewalling to specific networks

Evis Blackflag: Better handled by LL - IP would be related to privacy no?

Annie Milestone: Gelocation is so important to the customer!

Hackshaven Harford: evis, it's already being done :)

Charlotte Bartlett: I agree with IP Address...I can get it as a webmaster....surely no difference here.

Neil Robinson: there'd be issues with that I think - but if LL would provide us with country of origin, we wouldnt need ip address. I think if word got out that we were getting peoples IP addresses there would be murder, people would be up in arms

Ancient Shriner: yes, it also allows you to do context switching of content based on country of origin

Neil Robinson: pr nightmare

Hackshaven Harford: sure metrics lead to personalization of experience - think amazon.com

Evis Blackflag: You know how guests are - they'd denounce "omg you have my ip addy"

Neil Robinson: if that was an opt out thing - people would just go nuts, the media would pick up on the 'invasion of privacy'

Evis Blackflag: if it was opt out we'd have handicapped data

Siobhan Linden: (Note sim health and real time fps info are something we could provide probably better than anyone externally)

Sitearm Madonna: yep

Glenn Linden: any additions to that list?

Ancient Shriner: current pose . . .

Korvel Noh: OS, client

Kriz Janus: I have a question: when LL would provide an API for Infos, would it also provide a database for a lifetime view on this data?

Kriz Janus: or just a stream of current data

Sensors & Concurrency on a Sim

Annie Milestone: Ok, the problem I see is the limit to 16 avatars that can only be screend at a time

Ancient Shriner: that can be worked around Annie

Annie Milestone: So if I have an event with 50 avatars dancing. I only get 16 of them

Ancient Shriner: but I would like sensor sweeps to grab more than 16

Sitearm Madonna: yep

Ancient Shriner: my system grabs 64, but I would like more 4 sure

Charlotte Bartlett: I reiterate what company wants to see stats on 50 avatars on a sim - it actually damages the platform we can't have 10K avatars....

Annie Milestone: But why not as much as the sim can take, 100

Sitearm Madonna nods at annie yep

Ancient Shriner: well, that's a good point Charlotte . . many people don't want to know the truth

Kim Anubis: There are ways around the sensor limit.

Charlotte Bartlett: If I am putting an event to x brand and say hey I can show you 50 avatars on the sim, it works against us...can't see the point of it tbh.

Kriz Janus: true but important to know

Neil Robinson: yeah, low concurrency isnt something we need to draw attention to, instead we need to be able to show how over a month how people connected with the brand, to show it is a useful engagement

Charlotte Bartlett: Now if I can show same company here is a group of 1000 avatars who have HUD/ Notice interaction and it was event driven by x y c and then a b c occured - it helps. The rest isn't going to increase our ability to bring business / brands to SL.

Neil Robinson: totally

Sitearm Madonna: the data i need is simpler - a) total number of individual avis on a parcel in a day, and b) their names

Charlotte Bartlett: So I would prefer so we not waste development time on metrics that do not give considered benefit to the onboarding stategy we may have. Just my thoughts of course.

Issue Tracker

Siobhan Linden: I would start from a wish list as opposed to trying to worry about implementation details in this discussion

Hackshaven Harford hopes we can come up with a list and submit bug requests to JIRA later

Siobhan Linden: I would agree with Hackshaven, if we could jirafy the request list it would be a nice starting point for sorting this out

Glenn Linden: We'll create an Issue Tracker for this and send it to all of you to add your ideas to.

Sitearm Madonna: will you link all the existing jira requests to this one new one glenn?

Siobhan Linden: Good point Sitearm

Charlotte Bartlett: Are we going to have cost/benefit - if it takes 3 months to create something that only benefits a project that will onboard 10 avatars, it's not going to work for the larger companies here.

Ancient Shriner: Charlotte, all of my customers want to know the truth . . . avatar interaction isn't the only way to get ROI out of SL

Glenn Linden: Charlotte - that's good data to add to your Issue Tracker list - you're the one who knows how these metrics benefit your clients.

Siobhan Linden: Agreed Charlotte, I think it has to have value for larger developments as well as for smaller ones, and be additive to what exists today

Charlotte Bartlett: Really? Truth is fine, however non value added development will further make this platform unviable. Let's ensure we have the right focus.

Sitearm Madonna: agrees , where "truth" = actual numbers individual avatars :). Cumulative number individual avatars in one day helps a lot - number of exposures to the content in the sim, blah blah .. works

Annie Milestone: One more thing I have: Well, the 96 meter limit. It's ok, we can work with it, but much better would be one sime wide scan

Kim Anubis: We already do that, Annie

Annie Milestone: ok

Sitearm Madonna: and NAMES .. even better for followup w visitors

Hackshaven Harford: it seems like many of these are types of analysis which are built off the basic data types we were listing before

Sitearm Madonna: yes hacks.. except for that rotation bit ;)

Timeline for Release of Land Metrics Data

Hackshaven Harford: how soon SIobhan?

Sitearm Madonna: how about a test feed in May :)

Siobhan Linden: depends... ;-)

Kim Anubis: Wow, great

Hackshaven Harford: May? serious?

Kriz Janus: cool

Siobhan Linden: we have a prototype up of some real time and aggregate data feeds, so the issue is producticing that and matching it fit the needs of the community

Glenn Linden: (Of course, that won't be all the data on Hack's list!)

Ancient Shriner: camera, need camera

Siobhan Linden: I would suggest it be a phased approach, and rotation, well, maybe in a um later later phase maybe

Siobhan Linden: May, um, lets just say we might be able to provide _something_ in may for early review ;-)

Kriz Janus: What aggregated feeds will there be?

Developing Wish List for Land Metrics Data

Glenn Linden: We'll get the Issue Tracker up and link to those we can find easily and send that to you.

Glenn Linden: We'll summarize the discussion today and post it to the wiki.

Charlotte Bartlett: Glen, once you have this "wish list" how will you strategise the development in terms of content?

Glenn Linden: Our plan was to do a survey of both Solution Providers and holders of multiple regions in SL to get priority.

Glenn Linden: Certainly, information you can provide on how your clients would use the data should be added to the Issue Tracker to assist us with prioritizing.

Doc Tokyo: but what is about the ip adress in general ... will it be possible or is there aready a clear no?

Glenn Linden: Doc - as we prioritize the requests, we'll have to review legal privacy requirements. We can provide country more safely than IP address.

Doc Tokyo: ok .. make sense ...

Ancient Shriner: but how you get the country is important Glenn, geolocating IP address is better than looking up what was filled in on signup

Siobhan Linden: it is more a matter of should we than could we for ip address level type info

Neil Robinson: let THEM geolocate and give us the data.

Hackshaven Harford: LL has the IP address regardless if they give it to us - let them geolocate

Sitearm Madonna: there's a fear that SL interest is lessening in general.. but i've noticed SL interest is increasing for specialized audiences and topics - for projects that have money and realistic understanding of SL - for those projects, this land metrics utility is most timely!

Charlotte Bartlett: Indeed, we'll ensure the community elements are defined in our request versus basic sim traffic of course.

Glenn Linden: Charlotte - great. Site - good to hear.

Neil Robinson: I would strongly vote against the IP tracking, I dont think SL can take the backlash that'd create and it doesnt serve any purpose my clients would care about other than country of origin. If LL gives us country of origin we dont need IP tracking.

Charlotte Bartlett: Well you can get it anyhow from streaming I suspect if they visit off-platform...

Neil Robinson: yeah but thats not LL doing it.

Charlotte Bartlett: agreed so no need for it here.

Neil Robinson: its just how people would react - we dont need that storm

Last Questions

Glenn Linden: If Siobhan can stay a minute, you're welcome to ask any additional questions.

Siobhan Linden: sure, I am having food brought to me :-)

Charlotte Bartlett: Glenn one final thing - are you also going to publish a timeline for this project / milestones etc so we can monitor progress?

Glenn Linden: Siobhan - can you reply to Charlotte's question?

Charlotte Bartlett: Siobhan I am more interested in seeing a breakdown of the milestones rather than just something in May (sorry)....e.g. we are doing Voice of Customer, then I assume the prioritization with cost/benefit and then development lead time and implementation

Siobhan Linden: thankfully we have a working prototype that streams and captures a good chunk of data, so that is a big headstart

Charlotte Bartlett: (Reason for me asking is client expectation externally and what plans we can use to ensure we provide correct and relevant communication to them)

Siobhan Linden: Charlotte, I fully agree, and in the end the deliverables should come out of these discussions matched against the dev costs. So I would think a published milestone list for review is entirely a reasonable thing to expect

Charlotte Bartlett: Thanks Siobhan

Annie Milestone: Oh and one more thing, I forgot, also handable, but could be easier: OverThisParcel would be more helpful that OverMyLand

Hackshaven Harford would also love to see data collection settings in the hands of the land owner... another tab on the about land. LL can wash their hands of the legality....

Ancient Shriner: That's a good idea Hackshaven

Neil Robinson: do we want LL with hands of legal responsiblity tho

Glenn Linden: Hacks - interesting idea, and makes public what's being collected so avs can make choice about staying.

Hackshaven Harford: bingo -- /me refers to the XML P3P implementation...

Doc Tokyo: but as a sim owner you maybe want to monitor the whole sim ... if every parcelowner can make his own desicion about collection data there has to be a overwrite functionality

Kriz Janus: yeah ;)

Hackshaven Harford: and thinks of a land tab with similar options that would affect how LSL works on a given parcel. Estate tools would specify if parcel owner could change the setting...

Glenn Linden: Thank you all very much for coming today - we apologize for the late start

Madhavi Linden: thanks everyone. Great to see so much interest in this

Siobhan Linden: Yes, thank you all, this was very valuable!

Return to SLDev_Forum_Transcripts