Guest Editorial by Galen: A Tale of Two Sansars

It was the best of times, it was the worst of times, it was the age of wisdom, it was the age of foolishness, it was the epoch of belief, it was the epoch of incredulity, it was the season of Light, it was the season of Darkness, it was the spring of hope, it was the winter of despair, we had everything before us, we had nothing before us, we were all going direct to Heaven, we were all going direct the other way – in short, the period was so far like the present period, that some of its noisiest authorities insisted on its being received, for good or for evil, in the superlative degree of comparison only.

Charles Dickens – A Tale of Two Cities

This is what it feels like sometimes to be a creator in Sansar, the Social VR platform being built by Linden Lab (LL), creator of Second Life (SL). It seems like everyone considering Sansar is at a polar extreme about its prospects for eventual success. The only thing it seems we can all agree on right now is that Sansar is pretty cool, but nowhere near “done” enough to grow its nascent community of residents.

Not that LL hasn’t tried. LL has fostered several deals to tie existing popular media properties into Sansar in hopes of drawing people in. The prime example was the combined Intel CES and Ready Player One project. Also noteworthy are the popular Twitch streamer UmiNoKaiju, the Art of Drew Struzan gallery, and Mission Log, complete with a reproduction of the original Star Trek Enterprise bridge.

And not that we residents haven’t tried. My colleagues and I have worked very hard to create and foster the HoverDerby team sport. Alfy has been working hard on his live music events and new Voices of Sansar live competition. Longtime SL bloggers Draxtor Despres and Strawberry Singh have teamed up to bring us their weekly Atlas Hopping YouTube show and more than a few other broadcasts about Sansar. The nearly 1000 experiences listed in Sansar’s Atlas speak to the attempts of many of us to draw people in.

New World Notes blogger W. James Au recently broke the story about Sansar’s low concurrency rates using data collected by Sansar resident and scripter Gindipple, creator of The Combat Zone paintball experience. Gindipple started collecting data from Sansar’s own API in mid-February. The most prominent conclusion one can draw from his data is that the number of people visiting publicly listed experiences in Sansar rarely exceeds 50 people at any one time. And that the per-day peak has not been growing in the past 3 months that Gindipple has collected this data.

This is a sensational conclusion, you must admit. It leads more than a few people within Sansar and outside to draw very pessimistic conclusions. Maybe Sansar will never grow. Or maybe it will be overtaken by other social VR platforms like High Fidelity or VRChat before it ever gets off the ground. Maybe the poor stats of all the social VR platforms means that the world isn’t ready for social VR yet. Maybe it never will be.

But I’m an optimist. I think it’s too soon to sound the death knell for social VR, and certainly for Sansar or any of its other promising competitors. I’ve been collecting data from the same source as Gindipple since March. When I study it more closely I see a different picture.

First, a word about data. I get one very small three-dimensional lens to look through: head-count in each listed experience at this current moment in time. I take a snapshot every ten minutes of all this right-now data and add it to my database. Looked at over time, you start to get a very rich picture of where Sansar is today. Let me give some examples of what I see.

Experiences

This first graph is striking:

Figure 1 – Number of listed experiences since the start

Starting from my first day of recording (3/25/2018), I have kept track of how many experiences are listed each day. Thankfully, every experience listing comes with the date it was first created, so I was able to back-fill my data with an estimate of how many there were on each day going back to the beginning of Sansar. Please note that any experiences that were deleted or delisted along the way would not be represented in this historical view; hence the early history is a slight underestimate.

The very first experience added to Sansar and still around when I started collecting data was Midgar, created 12/20/2016. The data says it has not been updated since 3/2/2018, but I suspect that the owner hasn’t meaningfully worked on it since last year. And since each resident gets to create and maintain up to 3 experiences without paying a subscription fee, it probably will be around and unchanging forever.

Since then, you can see an explosion in the number of experiences listed. But the growth doesn’t follow a simple exponential or linear pattern as you might expect. There are pronounced upticks in growth along the way. One big one starts around 6/30/2017. At a product meetup that day, then Community Manager Jenn announced an experience building contest was beginning with a top cash prize of $10k and other awards. That jump in experiences tapers off just after the 7/25 submission deadline.

The second big jump begins around 7/29/2017, right as Sansar finally opened to the public. Looking past that jump, from 8/22/2017 to 5/22/2018, at least 380 new experiences have been created and listed at a fairly steady rate of about 1.4 new experiences each day.

What can we conclude from this one graph? Sansar’s user-generated content is steadily growing now and showing no sign of slowing yet. At this rate we should hit the 1000 listed experiences mark by the time this post gets published. Second, LL’s first big content creation contest worked very well. With about $36k in prizes, Sansar’s contest triggered the creation of up to 192 new experiences for an average cost to LL of $188 per experience. LL clearly could not have created that much new content by paying its own staff or outside contractors that rate. Third, the new experiences were perfectly timed to greet the flood of newcomers to Sansar when it opened up. Fourth, it’s clear that the opening did bring in a bunch of new talent, given the steady growth of new content since then.

Overall concurrency

Take a look at our next series of graphs:

Figure 2 – Minimum, maximum, and average daily concurrency

The top line represents the peak concurrency (number of people visiting experiences at one time) recorded that day. I take snapshots every 10 minutes of concurrency, so these are approximate. The bottom line represents the minimum concurrency. Not surprisingly, this is nearly zero most days. The real surprise is that there are some days when it is not, a reflection of the fact that Sansar’s residents are global. And the middle line represents what I’ll call “traffic” from now on. I compute this by averaging the concurrency in each 10-minute snapshot over one whole day.

Figure 2 shows a fairly clear pattern: no real growth of traffic in the past 2 months. Figure 3 shows the same thing smoothed out by week instead of by day:

Figure 3 – Minimum, maximum, and average weekly concurrency

It’s difficult to glean much from this, other than that how many people visit and how long they stay on average have stayed steady each day. Right now Sansar’s traffic hovers around 10, which is equivalent to having 10 people logged into Sansar all day with no variation. The daily (and weekly) peaks reflect the events that occur each day.

Events

Let’s dig a little deeper. Let’s pick one single day —  Tuesday, 5/15 — and analyze it. Here’s what the concurrency was during each 10-minute snapshot across all experiences:

Figure 4 – Concurrency during all of 5/15/2018

Here’s where it becomes apparent that the traffic (average concurrency) value of 9 for the day does little justice to understanding this particular day. The real question is: what was happening on 5/15? Was everyone at one place? Was there a big event that day for over half the day?

In fact, the data lets us find out what was going on along the way.

Figure 5 – HoverDerby’s concurrency during 5/15/2018

Figure 5 shows the same top line (blue) with Sansar’s total concurrency, but also shows the concurrency specifically for HoverDerby, one of my own projects. Here are concurrency numbers from some other experiences from that day:

Figure 6 – The Beach (by C3rb3rus) during 5/15/2018

Figure 7 – eSports Hangout (by Aleks) during 5/15/2018

Figure 8 – RPO: Aech’s Garage (by Sansar Studios) during 5/15/2018

When we combine all 4 of the above experiences together, it becomes apparent that they explain most of the day’s traffic:

Figure 9 – All 4 above experiences’ concurrencies combined during 5/15/2018

Yes, there were other experiences that had events and visitors that day. 33 of them had at least 2 simultaneous visitors at least once that day and 93 of them had at least 1 visitor. Here’s the top ten popular experiences for that day:

Figure 10 – Top 10 experiences during 5/15/2018 sorted by average concurrency

Note the occupancy rates, which indicates what percent of the day that experience had at least one visitor present. And the number of people who favorited each experience. The “At” column represents the (first) time when the experience saw its peak concurrency for the day, which typically represents when some event was in full swing. The “Pct of Total” column reflects how much of the total traffic for all of Sansar went to that experience that day. The Beach, for example, gobbled up 31% of Sansarians’ online time that day. And these top 10 experiences represent nearly 80% of all visitors’ time spent in Sansar that day.

30% of the day’s traffic went to experiences with peaks of 1 or 2 visitors. Arguably, this was mostly individuals and couples exploring 80 of Sansar’s roughly thousand listed experiences.

Looking back at figures 5 – 8, you can see the events that occurred in each experience. HoverDerby had its two daily practice sessions. The Beach had an all-day party. The eSports Hangout hosted the daily Community Meetup event. I don’t believe Aech’s Garage had any particular event, but it had a 6-hour bump in visitorship. I suspect two people were there greeting visitors, who are almost always Sansar newbies.

I can look at any particular day and figure out roughly what was going on with Sansar’s community using this same analysis. Almost every day I do this, I find there are several events going on that represent most of the day’s traffic.

Here’s another interesting graph reflective of the community’s daily activities:

Figure 11 – Number of experiences each day with at least N peak visitors

The top line (blue) is the number of experiences that had a peak of exactly two visitors each day. The line below it (red) is those that had a peak of 3 to 4 visitors. And so on down to a peak of 30+ visitors. Let’s smooth the data out a bit:

Figure 12 – Number of experiences each week with at least N peak visitors

This is the same graph, but per week (ignoring the current incomplete week). It still looks a bit like spaghetti, but look more closely. The 5-9 (orange) line shows a clear trend upward. So does the 3-4 (red) line. Even the 10-19 (green) line is generally trending up. Experiences with peaks of 20 or more are generally flat or trending downward over time.

What can we conclude from this? There are more events going on and people are going to them in smaller clusters. If the average concurrency isn’t changing much over time, this means that each person has more event options to choose from. One can conclude that Sansar culture is growing in diversity.

Case study: HoverDerby

There are lots of interesting questions that can be answered with the basic experience concurrency per snapshot time in aggregate, but it helps to consider single cases. I’ll take HoverDerby because it’s of personal interest to me, as one of its owners. And because I have additional data available. Consider a first graph:

Figure 13 – Min / max / average weekly concurrency at HoverDerby

This period starts from the week before HoverDerby’s premiere episode on YouTube. Naturally, the opening saw a lot of traffic. It’s important to point out that this combines traffic from both the main arena experience and the viewing lounge where we prefer non-players to be during our shows. Let’s take a closer look at the traffic (average concurrency) from day to day:

Figure 14 – Min / max / average daily concurrency at HoverDerby

The first thing that jumps out is that every Sunday, our YouTube-broadcast game days, get the highest traffic. The vertical grid lines in the above graph all fall on Sundays.

The second thing that is apparent from these two graphs is that, like Sansar as a whole, HoverDerby isn’t seeing much growth yet in our own traffic. However, it would be a mistake to assume that the people seen in these graphs represent a stable, unchanging population.

Since 5/10 I’ve started collecting data on individual visitors to the main HoverDerby arena. Here’s a first look at unique visitors:

Figure 15 – Unique vs regular visitors to HoverDerby per day

Almost every day lately we are getting 20 – 60 different people visiting. The blue line is the total number of uniques. The red line represents our regulars. That is, people who have visited before. That leaves everyone between the red and blue lines as first-time visitors to HoverDerby, or 10 – 30 first-timers most days, or around 140 first-timers per week. When I attend practices, I almost always personally welcome 2 – 4 newbies to Sansar and help at least one with basic how-to advice. Some of them eventually become regulars.

Conclusion

I don’t want to weave a fiction here. The reality is that Sansar’s concurrency numbers are not really growing. There is a fairly persistent core of active residents. Some fade out while others join to take their place. The concurrency story at High Fidelity is very similar:

Figure 16 – High Fidelity’s peak concurrency per day vs Sansar’s (source: SteamDB)

The blue line is HiFi’s peak and the red line is Sansar’s. To help untangle these, HiFi’s average peak is 13 and Sansar’s is 11. Both are lackluster.

However, I don’t think these numbers tell the whole story of either platform. The data I’ve laid out shows that content is growing, events are becoming a prominent part of daily life, and more experiences are capturing at least small crowds each day.

Perhaps most significantly, Sansar has a steady stream of first-time visitors each day. Concurrency numbers say nothing about this fact. People are finding Sansar and giving it a try in healthy numbers. Clearly, most of them are choosing not to stay. Why they aren’t is a critical question for both LL and Sansar’s residents to try to answer better. My current estimate is that maybe 200 first-timers are showing up each week. We need to convince more of them to stay.

I’d love to see Linden Lab publish some of their own data and summaries. In the meantime, I’m grateful for them sharing the small trickle of very useful data that I’ve been able to harvest and mine for insights. Sunshine is good. Some recent feature enhancements are making it possible to collect and summarize even more information. I have collected only two months of data so far. Expect more insights very soon. And I hope to see more of the same kinds of analyses for other social VR platforms soon, too.

Liked it? Then please consider supporting Ryan Schultz on Patreon! Even as little as US$1 a month unlocks exclusive patron benefits. Thank you!