Originally published on my Patreon here
Just before Linden Lab test launched "Project Zero", its code name for streaming Second Life on a browser, I had a deep dive conversation with CTO Philip Rosedale on how Zero fits into the overall strategy of growing the SL user base.
One point he mentions (that I initially missed, coming back from the holidays): Starting last week, they're letting a select number of new sign-ups try out Project Zero, to see if that helps retention.
"Basically, their first session will be just completely over the browser," as Philip told me last month.
I'll follow up on that soon, but for this week, here's part one of my Q&A with Philip on Project Zero's overall strategy -- including price considerations, why they're launching it now, and much more:
Philip Rosedale: We're gonna see if we can get streaming to work well enough to be a mainstream way of accessing Second Life. What we're doing, one, is we're using a streaming infrastructure to transmit the actual rendered view of the world. And then the second thing is, in a manner a little bit similar to the mobile app, we're gonna rebuild the UI from the ground up using modern HTML components…
For right now, there's going to be limited capacity and time limits on that, but it's going to be like, probably 10,000 people a day or more should be able to roll through and just try it out a little bit, so everybody can kind of see what we're doing and get a feel for it.
And then a few days after that, probably on Monday the 6th, we'll start factoring as an A/B test, a good fraction of the user signups through this thing. Basically, their first session will be just completely over the browser.
Our feeling is there's a lot more people that are similar to the people using Second Life today that could be using it than the number of people we currently have.
So our mindset there -- and this affects how we're approaching marketing and branding and everything else -- is that what we what we should do first, is remove the two big barriers that doubtless keep people, like all people that are in Second Life, from using it: One, basically just the inability to run it on their computer, or the fact that it just runs way too slowly on their computer to get access to Second Life.
As you know, different groups of people that are using [SL], there's lots that clearly would like to be able to use it with lower end machines. And in fact, we have a large cohort of people -- which we've really seen over the last couple of quarters, because we've been challenged by the updates to PBR and rendering, the rendering core and everything that went out with Firestorm in June. We can tell that there's lots of people that don't quite have fast enough computers to run this stuff anymore, and so moving directionally worse for them, not a good thing.
So we basically started working on [streaming] a couple of months ago. and we're ready to start testing it.
Right now, the way we're doing it stage one, is we are just streaming the existing Second Line viewer UI as it is, no changes to it. But then during January, you'll see us start turning off most of the elements of the UI and then rewriting, in the most important order, basically rewriting the UI as modern HTML React [code], in the same way that we've been able to do with the mobile client.
Wagner James Au: On any browser?
PR: We've been obviously mostly testing on Chrome because that's the most common target, but it works on Mac and Windows, no dependence on GPU. And the thing we're gonna say to people, is it's basically just, If you can watch Netflix on your computer, you’ll be able to [stream Second Life].
Next up: Latency tests, pricing considerations, and riding the back of failed AI experiments.
PRPR: Our testing so far -- I mean, we'll test from different endpoints, different cities and stuff as we scale -- but you know, our initial testing on it, for ourselves and for the Lindens has been that it's way beyond adequately fast… You can't tell the latencies there. You know how obsessed I am with all that latency testing and stuff like that; we've been on top of all this, and we've had some pretty good experiences that made us think this is the way to go.
How Failed AI Projects Made Project Zero Possible
PR: [A]ll the AI work that we've been doing at the Lab taught us one thing, which is that there's just this huge change around GPU availability in the cloud, and with the number of failed AI experiments, we expect that the cost of accessing the cloud is going to get lower.
Obviously, the prices' been coming down anyways over time. So you know, there's an intersect point where --- especially for Second Life users who tend to spend more money in the world -- there's a point at which the cost of sustaining but providing that kind of [streaming] access is lower than the average returns to us on an active person that's using Second Life.
And so we're just kind of anticipating that that's coming quite quickly. And so we were like, alright, let's just get this working and turn it on right now, with limited capacity for now for us to do it. But we're hoping that it will show us the way to building an all new experience around that that’ predominantly access via browser.
And as we said before, we view the mobile client as being an adjunct to desktop use. So we don't imagine that people using Second Life in the way that they are now are going to just switch from desktop to mobile. We think more likely that a happy Second Life user will make some use of the mobile app and then some use of desktop access. So what we're trying to optimize for here is obviously the desktop access, which still requires a massive machine.
WJA: But then the browser-based [streaming] will that fall under the bucket of desktop access?
PR: Exactly. Our hope is that we can transfer people that are currently using Second Life on desktop or on laptop full screen today from doing it with their old CPU/GPU system requirements and download and install.
We estimate from our numbers that probably at least half the people that sign up for Second Life are critically negatively impacted by not a great computer, not being able to install [SL] or just run it at a very low frame rate.
And so our rough thinking is, if we can fix that, we're gonna immediately see a significant increase in new residents from that. And then we figure out how to offer [streaming] to existing residents as well, in some fashion.
WJA: I would think people who have low end machines also tend to have low quality broadband.
PR: [W]e've got lots we've got lots of tracking, so we'll know if that's the case. But as I said before, it's like Netflix is a pretty low bar. We really only need enough for a video stream… But yeah, like, we can potentially turn the resolution down to hit lower bandwidth, all that stuff.
Pricing Project Zero for Regular SL Users
WJA: You haven't talked about price yet. So how much does this cost?
PR: Well, our hope in the long term is that it costs nothing, so that it just ultimately becomes, you know, a choice in terms of accessing the world.
In the meantime… we’ll look at ways to provision the access. There's obviously a difference between providing to everybody that's an existing resident that wants to use it in any manner they care to, since many existing residents have, like, multiple atl accounts, so they would like to run 10 avatars at the same time all day.
This is obviously still highly non-zero as a cost. There's these different streaming providers… [and] it's obviously a very good time to buy our streaming services, but still, right now it's relatively expensive. So we're not going to provide 50,000 concurrency access to it right now.
But we're thinking hard about ways if people want to use it more than we can give them access to it; what do we charge them? A pay-as-you-go thing, or a part of a premium subscription. That's more Brad’s thing to think about, focusing on that, [I’m] just making the technology work right, feels good.
WJA: I remember we had OnLive. I helped to launch it, I think the pricing was $3 an hour for streaming, something like that. And they said they were seeing good adoption. Like that was kind of the mystery -- they were seeing good uptick of people subscribing [to OnLive], but then it got canceled; apparently it wasn't scalable right then.
PR: As you can imagine, that's part of what drives my thinking. As you know, at High Fidelity, we actually built a whole streaming infrastructure, just experimentally inside [the company]. We got a better feel for how difficult this is as a project to do, you know, so it gave us a level of confidence coming into this one.
But the quality of this stuff is so much better… Latency [in streaming] is now really pretty imperceptible, unless you're playing like Destiny or something. Google stupidly demonstrated that streaming Destiny is probably not the best idea for launching a streaming service. But for Second Life, for the experiments we've done, the latency is absolutely undetectable.
WJA: No graphic card and fans coughing and wheezing to [display the world]!
PR: I'll be logged into the world chit chatting with people, and I'm actually on our [Project Zero], and my computer’s so cool, there's no sound around me other than the sound of Second Life.
Next week: Improving the UI, adding game mechanics, and Linden head Brad Oberwager joins in. Join my Patreon to read it first!
My only concern with this is that Philip is talking as if it's impossible to get better performance from the viewer and I don't think that is true. I know the ways to do so are unattractive since they amount to a ground up re-write of a lot of code but... it's worrying that they don't seem to have even started.
The streaming stuff does work very well but it cannot be used as a substitute. SL could run a lot more acceptably on a lot more computers - but not all, I accept - with some effort put into re-developing the desktop viewer with better multi-threading and using a modern graphics API.
They've built a new viewer in Unity recently and while it's slow going they do seem to be getting somewhere. This is no small project and I can't help but feel with both mobile and streaming projects ongoing they're neglecting the important and now pretty time critical work of bringing the desktop viewer up to date.
Posted by: Kay | Tuesday, January 14, 2025 at 01:18 PM
Anytime Philip has been involved in SL the last decade it's only been disaster! From his 1st golfing buddy destroying everything LL had built including selling everything even the lab itself near bankruptcy. at the same time he had a competing service trying to suck away SL residents while his golfing buddy talking about SL was a sinking ship and Sansar would rescue everyone then we could all live in a virtual prison with ugly avatars.
Now he has his 2nd golfing buddy pushing mobile and a SLGo replacement I mean what could go wrong? just like a million times before we had a voice like Kay who posted a
voice of reason here about things like other things important being neglected but just ignored.
What else does Philip plan that will in the long run will finally finish SL off?!
Best thing Philip could do is go chase the next shiney he comes across and never look back.
Posted by: Steve E. | Wednesday, January 15, 2025 at 06:36 PM
> SL could run a lot more acceptably on a lot more computers - but not all, I accept - with some effort put into re-developing the desktop viewer with better multi-threading and using a modern graphics API.
They did that. It was called Project Sansar.
Posted by: Arduenn Schwartzman | Wednesday, January 22, 2025 at 02:09 AM