there seems to be a recent surge in the buzz from the virtual world blogosphere that Second Life is dead. i’m not sure how that ethereal vibe started but if Linden Lab is truly a $100 billion dollar a year company with under 300 employees, then it is doing pretty well in comparison with many businesses in the US
from second hand accounts, Rod Humble’s speech at this year’s SLCC was a bit neutral with this statement “The fact that Second Life still enjoys that growth is stunning” and then a continued mantra of “fixing bugs” (makes me think of a doctor being surprised that a patient they operated on is still alive! and still fixing search? cripes, that’s a long standing one) =D
Second Life is pretty nifty as it is and has not drastically changed since its start. it’s better for sure – less laggy, fewer crashes, and other things are better too (like sim crossings) but it has not really changed much – still 15,000 prims, still 50-60 peeps per sim, and still $295 a month
to be fair, Linden Lab has an offering that is fairly unique and rich in content. plus, they were the first to be big in this space and don’t have a road to follow, so it’s easy to be a backseat driver and second guess their decisions
being a “blank canvas” limits Second Life to being a niche offering especially since it is more of an application than a true social thing (my distinction being that social things are highly accessible by the masses, somewhat limited in what you can do, and as web pages without the need for anything but a browser)
SL (and OpenSim) is a bit like Microsoft Word – it’s not so much about the thing as what you do with it. you use Word to create a document and you use SL to create a virtual space and both start out as clean slates. but Second Life is not exciting without exciting content and LL talking about how they’re going to address lag is like harping on how italics will be better in the next version of Word
things like mesh are big news, but only for some existing users. for the slowly changing SL demographic, it’s not that big a deal (like the stoopid ribbon in Word – big whoop for most people). Rod did address that this new demographic is being offered “curated areas” which means that new people are
looking to be entertained rather than creating entertaining builds
making the first hour user experience better is not doing anything to satisfy the hunger that creators have and does not keep creators excited about building in Second Life. creators made SL what it is today and creators have been on a slow exodus over the last few years for a variety of reasons
- is this “death” of SL just the realization that SL has peaked? (don’t forget the 11 million user hour decline in 2010)
- is keeping a creator excited to build more entertaining stuff more of an after thought now?
- is the death of SL just the slow down in new builds plus the loss of old ones?
Rod acknowledges that SL is built by its residents but does not seem to be doing much to fan the fires of creativity or to empower creators or to retain creators (some would go so far as to say they sometimes discourage creators! *raises hand*)
with a dwindling creator pool and with nice sims disappearing, is SL destined to eventually become a vanilla virtual world to visit for events but not one to innovate in?
vanilla world on a stick (patent pending)
vanilla world on a stick (patent pending)
vanilla world on a stick (patent pending)
Running monopoly pricing and vendor lock-in of user builds and creations is not sustainable in my mind… its the voodoo death doll of any monopoly…
Eurominuteman
15 Aug 11 at 8:25 pm
Second Life is Dead – Prepare for an OpenSim World
http://sloanconsortium.org/conferences/2011/aln/second-life-dead-prepare-opensim-world
Eurominuteman
15 Aug 11 at 8:49 pm
<3 vanilla world on a stick
Hiro Protagonist
15 Aug 11 at 8:56 pm
While you’re filing a patent for “Vanilla World on a Stick,” you should also trademark “Vanilla World on a Stick” and copyright the build.
LOL
Maria Korolov
15 Aug 11 at 9:05 pm
I doubt SL will die. Runescape, Everquest and others are still going, due to loyal followers. But SL is shrinking. http://gridsurvey.com/charts/historicalconcurrency.png
I think that they will end up reducing their land (to save the server costs) and their tier (to stay competitive). They will no longer be the top grid, if they still are. As OpenSim develops and becomes more stable, someone will set up OpenSim as a familiar style of installable binary so that the console is no longer needed. Just download, install, choose default or custom (which would open a window to enter your settings) and away you go.
Sarge Misfit
15 Aug 11 at 9:33 pm
Sarge —
Well, nobody has done that with Apache yet, and websites are everywhere.
I think what’s more likely is that OpenSim hosting prices will get close to zero. With on-demand regions, Intel’s scalability improvements, and super cheap storage, I believe grids will soon be able to offer residential land for free. Think GeoCities or Blogger — but in 3D. Everyone can have their own free little world, and teleport around anywhere else via hypergrid. Grids can still make money by charging for central commercial space, selling stuff, and by charging for prominent placement of hypergates to other grids.
Right now, Kitely is running over 1,000 regions free for people. They’re self-funded, so they can’t do it forever, but another company, with some startup money, could do just that. They can run at a loss for a year or two and basically take over the market and dominate the industry, with the idea that costs would drop fast enough, and other revenues would come in fast enough, so that the company would be cash-flow positive before the funding ran out.
Actually, I’d invest in something like that.
Maria Korolov
15 Aug 11 at 9:47 pm
You may be correct, Maria, time will tell. Thing is, SoaS has all that, so all it needs is to be compiled as a single binary with a user friendly setup window and away it goes. User friendly as in the average end-user, with less knowledge than us gridlings :-)
Sarge Misfit
15 Aug 11 at 11:13 pm
Had another thought, it may well be that we’re both right. Look at operating systems. We could end up with two or three ‘types’. One that is a simple download and install, like most Windows software, and something more detailed and hands-on, which would be the Linux of grids. And variations in between.
Sarge Misfit
15 Aug 11 at 11:16 pm
The part you’re not seeing Maria is that the infrastructure costs to run a large grid are no where near zero and definitely are a lot more expensive than cheap web hosting. You can not keep equating web hosting with virtual world hosting, they have two completely different levels of system demand. As the grid itself grows the amount of “always on” servers will also grow. Any central storage can not just be turned on and off, and making each region it’s own hypergrid makes the problem worse because you need to serve assets off of each “grid”.
Web sites hardly need any CPU or I/O resources. Simulators on the other hand need a ton of both. The comparison is apples to oranges. Like comparing the load of a nintendo game to the load of an xbox game.
Its a nice thought but telling everyone hosting should go to “zero” is really silly. If that happens you’ll just have to deal with the data mining again for the “big company” to pay for it’s servers. If you’re following nymwars you’ll see how well that works. Also if you’ve ever actually used cheap web hosting ($10/mo for example) you know how crappy and unreliable it can be. And this is for http that puts very little demand on a lightly used server. Compare that to the monstrous loads of a simulator running physics, scripting and supporting heavy network I/O.
So the on demand idea works great for regions that need no interconnection and no central asset serving, but once you’re looking at a real social platform there are many more complications that come into play.
Tranquillity
16 Aug 11 at 12:45 am
Also, Intels scalability improvements are achieved by using _more_ servers to handle the loads. This does not equate to cheaper hosting.
Tranquillity
16 Aug 11 at 12:55 am
Using Opensim means that grid operators, as Utility Providers, are in the backseat now, any other aspirations are somewhat frivolous.
So only quality-driven ones count.
Eurominuteman
16 Aug 11 at 1:05 am
I think ignoring the social aspects of OpenSim and other 3d social technology and rendering everything as a “utility” is a good way to be mediocre. Apache httpd doesn’t come with built in ways to interact with others.
Ignoring the social aspects of this tech relegates OpenSim to being just another modeling tool.
SL survives because of the people that log in every day to be with their friends as well as their creations. It survives due to the real, meaningful, and lasting bonds these people have made with each other.
People live here.
Tranquillity
16 Aug 11 at 1:13 am
On Opensim, the backseat Utility Role promotes competition and gets away from the monopoly role as pre-assumed by SL…
The Opensim social outreach function is then part of the agenda of each value-add content creator pointed to out-world newbies… This value-add for out-world newbies is what drives user concurrency…
Whereas, this out-world newbie view is missing in SL, moreover the domestic in-world newbie view prevails…
Eurominuteman
16 Aug 11 at 1:39 am
Hi Maria,
I’m not sure the things you listed can be sufficiently monetized in a completely open metaverse to support the costs or running Kitely but if we had a way to continue giving out the service for free, hire some additional people and start taking home some money to cover our living expenses then we would probably chose to pursue a freemium business model. The problem is that a freemium model only becomes financially viable once you have enough users and we need to be able to grow the company and feed our families.
That said, we plan to continue offering a small amount of free Kitely Credits for some time after we start selling them so that casual users will also be able to visit worlds where the world manager has selected that visitors will cover their own usage costs. Our ability to continue doing so over time will depend on the amount of KC that we actually sell and how quickly our user base grows.
If people help spread the word about Kitely and we see fast active user growth with good conversion into paying customers it will help us maintain a freemium model.
Ener – you heard it here first ;-)
Ilan Tochner
16 Aug 11 at 1:40 am
Hi Tranquillity,
I agree with you that the costs of running a virtual world service are much higher than those of running a web hosting service even when using on-demand cloud computing for your infrastructure.
That said, Kitely provides a shared inventory across all the 1050+ worlds on our grid using our limited budget and there are some optimizations we have yet to implement that will reduce our asset management costs significantly.
Please note that Kitely is already serving as a social grid for several disjointed communities and will serve a unified community once we add some missing features and capabilities to Kitely.
Ilan Tochner
16 Aug 11 at 1:53 am
Hi Ilan,
Moreover Im talking about the costs of storing, transmitting, and backing up things like:
- 4 TB of asset data
- 100 GB of set data
These things accumulate over time and tend to get larger and larger rather than smaller as a virtual world grows.
Asset data is especially important to keep an eye on with an on demand system because once you connect together simulations, that asset data needs to be available to all simulators a person may visit. Think inventory assets. Anyone live and running concurrently on an on demand system, regardless of where they have teleported, to will require access to their asset data. If that asset data resides on their personal simulation, that simulation will need to stay active for the duration of their stay in the metaverse so that they don’t lose access to their inventory.
If that asset data is on a central cluster, you’ll have to be prepared to pay s3 or wherever that data is stored from transfers outbound to the user if they’re hypergridding and services like s3 charge per request.
I’m simply trying to point out the economics of cloud hosting don’t automatically mean everything goes to zero. Im not questioning what you’re doing. However there may come a point where hosting your central services on a dedicated platform while running simulations in the cloud might actually be cheaper depending on how you deal with the problems I’ve describe above.
Tranquillity
16 Aug 11 at 10:07 am
For the record over the past 3 days there have been over 15 million requests for assets from our distributed storage system. Requesting that many assets from s3 would be extremely cost prohibitive.
Tranquillity
16 Aug 11 at 10:09 am
Hi Tranquillity,
Thank you for sharing some statistics from your grid. I truly appreciate your openness. Can you please share the average user concurrency during those 3 days? (it will help me make more sense of the load that produces 15 million requests)
I agree with you that the costs are far higher than those of running a web hosting business and you may be right regarding cloud based asset hosting costs but I’d need to have a lot more statistics to decide whether some things will be cheaper to run on dedicated servers.
Ilan Tochner
16 Aug 11 at 11:11 am
wow! outstanding information and openness (don’t patent these ideas SpotON3D, now that they appear here, they are copyright of their respective authors and form their intellectual property – which is becoming the norm, rather than patents, for big companies like pharmaceuticals who find it more secretive to have their formulas as private IP in places like Switzerland rather than public patents)
anyway, (i digress) that is an interesting insight Tranq and really frames the data in a new light (a huge light!)
it would have been WAY cheaper for me to do cloud for our grid over the last year but i worry about what “could” occur if subQuark’s book is a hit
i like having a 24/7 box and that is just my mindset and not based on anything more than out dated beliefs, but the box is certainly cheaper for things that are getting hit 24/7
Ener Hax
16 Aug 11 at 11:41 am
Hi Ener,
That is why we will offer an option for you to select that some people visiting your worlds will cover their own access costs (the $0.20 per hour we charge for visiting a world). This way you could cover the costs for some people while others, less related to you, will cover their own costs. The result should wind up costing you a lot less than using a dedicated server for hosting your worlds.
BTW, if your worlds are accessed 24/7 then they will automatically get their own sim on Kitely 24/7 as well.
Ilan Tochner
16 Aug 11 at 11:54 am
Hi Ilan,
I can provide you with the statistical information you require. I’ll send you an IM on skype later in the day.
Tranquillity
16 Aug 11 at 12:06 pm
Thank you Tranquillity,
I really appreciate your willingness to share this internal data with a competitor. I hope there will come a time when people will be able to hypergrid teleport between our two grids.
Ilan Tochner
16 Aug 11 at 12:27 pm
Hi Ener,
I’m glad to share this information because due to the length of time we’ve been up and running, our concurrent user count and number of simulators combined with free uploads we probably have an amount of data that many other businesses haven’t seen yet.
If you have a region where the owner is using 15k prims, and on that region they’re using 5k different textures (every 3 prims has a different texture for example), just 10 visitors to that simulator will make 50k texture requests to while walking around the sim. Taking into account the way j2k compressed images get downloaded in parts means you can also have.. say 3 requests for the same texture to grab only the portions that the viewer is interested in at the current discard level. That means one walkaround from a single user could request each of the 5k textures three times to display the proper and final discard level of 0. And this is only textures, we’re not talking about sounds or any other assets being requested from inventory.
Take that into a grid with multiple thousand unique visitors a month and you have some serious requirements. InWorldz at this early stage averages around 60 – 80 Mb/sec (that’s megabits) between all of our servers to the internet. Internal communication traffic (queries, internal asset requests) don’t even factor into these averages.
Tranquillity
16 Aug 11 at 12:34 pm
i personally don’t see InWorldz and Kitely as direct competitors. i mean there are a finite number of people in the world but your offerings speak to differing use cases (from my perspective). InWorldz is really about one strong community and is seemingly headed up by people who are deeply entrenched in virtual worlds in an analogous way to those in Second Life
keep in mind that this is just my perspective and one based on 4-5 years inSL and having had 19-20 sims who then fled SL
i tend to think of InWorldz as servicing people like me as well as new people like Mera who want a “better than SL” experience
i see Kitely as an alternative to what i currently do in OpenSim – a targeted task that serves an outside purpose – in our case, being illustrative of science principles
i think communities can exist in Kitely, just as they do on OSGrid and i see Kitely as more closely being analogous to OSGrid
Ener Hax
16 Aug 11 at 1:00 pm
Hi Ener,
As you correctly stated, people currently use Kitely in disjointed groups, i.e. as many small and separate communities. However we intend to add a few, very often requested, features that should help create a more unified user community. Only time will tell whether that will result in us serving the same needs as those that InWorldz serve. In either case I think both our grids can coexist and have people jump between them – as I hope will eventually happen with many of the commercial and non-commercial grids.
Ilan Tochner
16 Aug 11 at 1:22 pm
Well said, Ener. I’m with you on that. InWorldz and Kitely serve different markets and purposes, and I think the whole alternative grid community will be very well served by growth in both.
Jim Tarber
16 Aug 11 at 1:22 pm
Hi Tranquillity,
Thank you for the additional insight. Can you please let us know the average user concurrency that results in this 60-80 Mbit/sec bandwidth requirement? I’m trying to calculate the average realworld bandwidth requirement per concurrent user (you don’t need to divulge unique active per month for this – just the average number inworld at any given time).
Thanks again :-)
Ilan Tochner
16 Aug 11 at 1:32 pm
Hi Ilan,
160 – 170 concurrent users over the past week look to be generating about 70 megabits of outbound (to the internet) traffic and 30 megabits inbound (from the internet) traffic.
Lows for us are usually around 70 users generating what looks like 30 mbit outbound and 15 mbit inbound
This looks like averages around 400k to around 1/2 a megabit outbound per user. Roughly. These numbers are probably also counting TCP/IP and UDP packet overhead so some downward adjustments should be made.
Tranquillity
16 Aug 11 at 2:03 pm
Thank you Tranquilliy, this info is very helpful.
Does this also include voice traffic?
Do you know what effect viewer caches have on these averages?
Ilan Tochner
16 Aug 11 at 2:09 pm
Hi Ilan,
Those numbers do not include any voice traffic.
I know that residents who use their viewers on multiple grids end up having to clear their cache a lot of times due to caching issues in the inventory system on the viewer. There are other failures of the cache that have plagued the v1 viewers for a long time. McCabe knows more about that side of things than I do.
As far as assets, since a few days ago there have been 14.7 million cache hits during asset requests and about 2.4 million cache misses (total in memory cache size is 32 GB on currently active read servers) on a storage network of just over 3 TB. That means to us that our assumptions during the design of the asset system were correct: There are a number of regions that are frequented and where people collaborate far more often than everyone just sitting on their own regions working.
Tranquillity
16 Aug 11 at 4:47 pm
Thank you Tranquillity for your continued willingness to share statistics.
Ilan Tochner
16 Aug 11 at 4:51 pm
I feel that any honest business owner in this arena who wants to succeed can not afford to be an island. Competitive need not mean cut throat. I’m glad to be sharing this space with other trustworthy and honest players. It is a breath of fresh air at times.
Best of luck as always.
Tranquillity
16 Aug 11 at 4:58 pm
Thank you Tranquillity the feeling is mutual. I’m happy that we can work together towards a greater good and an open metaverse.
Ilan Tochner
16 Aug 11 at 5:42 pm
Having read Tranq’s info, I can only say
Holy Cow, have I got some thinking to do!
I’ve been trying to get a handle on what kind of data flow I can likely expect when I have Excelsior Station go public as my ISP charges based on that amount. I am NOT expecting anything like InWorldz’s level of traffic, but at just 1%, which would be one or two users at any given time, over a month, that could likely add up to a lot. If I end up with that many people visiting, that is. I know that when I visit InWorldz, I easily run over 100 megabytes of data over two or three hours, but that is from the client end.
Sarge Misfit
16 Aug 11 at 7:31 pm
Sarge,
Since you’re designing the world initially I’m assuming you can control some of these numbers by using standard building best practices.
- Use the same sets of textures as much as you can without making your region mundane.
- Try to reuse sculpts rather than making new, similar ones
- Try to minimize the number of scripts changing primitive properties such as size, color, and texture.
Also realize as there are more people on a region more traffic is generated as their movements and changes are broadcasted. Our traffic profile may not line up with your use case.
Tranquillity
16 Aug 11 at 7:40 pm
Done, done and done, Tranq, though without actually having data flow in mind, its just how the build has gone. A large portion of the build uses blank textures, though coloured, and the majority of the rest of the build uses only about a half a dozen textures. I will make sure to add those textures to the default library, though.
I’ve not had time to script anything, either. For now, that will be how things remain until I have it completely built (take a peek at my wikidot site for some pics and a bit of a write-up on it). Then I will export the whole thing in less-than-256 prim pieces as a permanent backup. Once I do script, the most used will be doors.
I am thinking of setting it public for 24 to 48 hours to get an idea of the data flow, which may be best. Some time after the 1st at the earliest, but that also depends on what I am able to do to set up accounts for people. Maybe set it up by email request? I’ll work it out.
Sarge Misfit
16 Aug 11 at 8:09 pm
[...] post is spurred on by comments to a post on Monday (if you want to understand a bit about the resources used by a user, read Tranquility’s [...]
OpenSim evolution is not analogous to website evolution at i live in science land
17 Aug 11 at 10:19 am
I think one way that grids can offer free land in the future is by severely limiting what the users of that land can do. For example, when you get a free blog on Blogger, you don’t get to design your own template or run your own plugins, etc.. You’ve got to choose from what they offer. Similarly, a grid owner might offer free residential regions — but you have to choose from a set of pre-made templates, where you can choose colors and palettes and landscaping themes, but only within certain parameters. That way, grids can add low-use on-demand residential regions without a corresponding linear increase in the asset database.
In addition, I expect third-party outfits — like hotmail and gmail — to come along to host avatar inventories. Hotmail and gmail lost money for a long time until they figured out how to get revenues — they were planning for the future. The technology is there now — a grid can set up a small land area with plenty of tools available to customize your avatar, get freebies, buy clothes and fancy skins, then hypergrid teleport to other grids for socialization, networking, clubs, business meetings, to manage their virtual pets, or to live in their virtual homes. there’s no reason why a single grid has to be all things to all people.
– Maria
Maria Korolov
17 Aug 11 at 3:20 pm
hmm, blogger has value to Google because they can mine lots of user information and behaviour. so if, like you said, Google and Yahoo had some offering, like Gavatars and Yavatars, just like Gmail and Yahoo Mail, then they could gather that additional data and sell that as well (anyone with Gmail or Yahoo mail should know that their data is mined and sold)
so GWorld and Yahoo World? why not!
Ener Hax
17 Aug 11 at 4:02 pm
Hi Maria,
Someone that can subsidize user growth until economies of scale kick in and an advertizing-based monetization option becomes viable may be able to do what you suggested. However, that is not likely to happen until OpenSim (or some other open source platform) gains a lot more traction.
If a freemium model, instead of an ad-based one, is the target then you need to also consider the cannibalization effect offering such free services can have on that operator’s premium services. Operating and development costs need to be covered somehow.
Ilan Tochner
17 Aug 11 at 5:35 pm
I am an active user of Second Life for more than 4 years. I think that SL is the future developments, there is virtually no analogues
dimm Torok
22 Sep 11 at 5:08 am