OpenSpaces: A Bad Product Gets Worse

It doesn't really take a lot of effort to see that the price “jack-up” on the Voids that became OpenSpaces means anyone who keeps that simulator product will be paying a lot more for a lot less. And not just less prim density--but less peformance. Aren't hosting+server costs supposed to go down over time? Isn't performance meant to improve over time? Isn't efficiency meant to improve over time? LL seems to have it all backwards (...again).

The offering from M Linden (= Mark Kingdon, CEO Linden Labs) does little to make this mess right and adds a new layer of confusion and technical nonsense to the already stressed Second Life user-base and the SL virtual world platform as well.

Anyway, here are some comments and more detail on why the OpenSpace offering looks so terrible. I am not sure this commentary is new, but some of the ideas explaining what the issue really is might be interesting enuff.  * Warning danger of boringness is high and I show a spreadsheet*  (Better to go watch a fun machinima here!)

Linden Lab basically has two product offerings: 1) a full CPU-based simulator = a private island, also called a Region or 2) managed land sharing of a CPU-based simulator which is a parcel or an OpenSpace. The original Void product was still a full CPU-based product but configured the simulator into 4 Regions instead of 1. The original Void still was sold to one owner and had to be kept together as a block of 4 Regions. The scandal of the OpenSpaces began when Linden Lab started to allow these Voids to be split up to different owners effectively making it blind CPU sharing. Blind CPU sharing on the Second Life grid is very bad as it makes it impossible to manage sim performance, plus the possible sim loading is significantly higher than normal, further resulting in owners getting what appears to be randomly serious lag and even sim crashes. At least on a Mainland Region you can look at the map and see where the possible lag-inducing activity may be coming from—with an OpenSpace this is impossible (although LL could actually figure that out).

At the same time the OpenSpaces were a huge part of the claim that Second Life was growing—really the only good news about the future of SL that we have been fed in 2007 (ok, so Mono and H4 are actually good things and should not get forgotten). But how does Linden fumble the execution on such a popular offering—twice even? There seem to be many options for dealing with this mess, but LL chose ambush and misinformation over a real solution. But why did LL even start selling OpenSpaces in the first place? Any technical LL staffer who thought about it for 5 minutes would have immediately realized the potential performance issues. Seems like another internal disconnect between the people with the revenue targets and the people with the technical geek-speak, babble. It's about the money and LL maybe was trying to squeeze the landbarons. Greed sounds like enuff.

Let's drill a little deeper to the core of the OpenSpace issue. Think about a full sim (=Region) running on a new class 5 server (not such a blazing hot server btw). LL says they host dual-core dual-cpu machines that run 2 Regions each on its own CPU. That type of server btw with 10mps of unmetered bandwidth is about $USD 200-250/month when we priced it recently. An Island=Region=Full Sim (like Greenies for example) can handle about 50 avatars max under active conditions before it get lagged-out beyond most people's patience threshold. We see that type of sim crash regularly under club conditions if it goes over 60 avatars.

So let's take a look at some of the costs on a Region like Greenies: 15,000 prims for $USD295 per month. It also would cost $1,000 to buy it now (actually we paid more in 2007 before LL screwed er...revised the pricing to lower acquisition costs and raise monthly tier). So if you look at this cost over the course of a year it comes out to $25.22 per thousand prims (CTP) per month. The old void cost was a little higher as the prims were a touch lower and that CTP = $25.56. The new pricing for the void with a paltry 700 prims comes out to $136.90 CTP and the 3750 version comes to $41.67 CTP. There is no clear economic sense to paying some much for so little.

But of course there are other factors involved more than cost. There is prim-to-land Density. So actually there is more space on a sim than just the 2D view of a sim. You now have up to 4000m vertical space. This used to be 750m but H4 and the newer viewers give us a huge amount more vertical space. So cubic km3 on a sim is 262,144km3 or even more if you travel beyond 4000m. So if we look again at Greenies we can see a prim/1000km3 density at 57.22 and the old void at 14.31. Now this density topic is more interesting on parcels where there is a lot of variance. It will also be interesting to see how/if LL decides to start pricing vertical space limits.

More interesting though than Density are two other factors that I call Interference and Action Rating.

Interference is an index that tries to capture potential lag from sharing that one and only CPU. So if you run something with 50 avatars max and you have no contention for CPU processing, you would expect to have no interference. Now if you push it to 60 or go nuts to 100 you would also expect to see increasing sim processing interference and have a diminished user experience resulting in lag and/or crash. But at least in that case you are interfering with yourself and could stop.

* Weird formula alert *

Interference = (Max sim loading / Avatars not under control) / Avatar benchmark max

You can see quickly that because each void was allowed same avatar loading as a full sim that the potential loading would have been 200 even though each owner was only aiming for standard usage. LL could have just told everyone and/or capped the avatar allowance.  Also this could have been fixed by finding a way to communicate capacity across the voids sharing that CPU (ie like a percentage of total). But it is clear also that LL did not want to solve the problem, but instead extract more money to ignore it at the expense of the users not in their interests as Kingdon claims.

The Interference on the new void would potentially be low, but LL left it uncommitted what the actual CPU sharing would be...seems like they will go for 8 voids on a CPU? In which case the interference rating would be unacceptable.

The Action Rating is what you can do on with your simulator. On your own Island you still can do what you want (except have under 18s) so the Action Rating is 100. For the new OpenSpace product there is usage and scripting limitations (not yet fully described). Now these low Action Rating are probably meant to offset the Interference, but looks like another fine mess.

How is Linden Lab going to police their usage terms? Wtf is “habitation” is that a house—even if no one is in it? Is it a tree? what about dancing around on the beach--for days on end as campers?? Is it doing something like virtual sex (oops... that is virtual copulation and Linden still let's you do that). Linden has made a massive mistake in even speaking about land usage in this way. They will certainly end-up in court with a major challenge if they try to ban an owner for using his sim. So this is an empty threat—or is Linden about to change its style completely and start monitoring sim usage? Start giving incentives for people to rat on each other?

Anyway, if you wanna take a more detailed look at my weird formulas and try to see if your parcel is a good deal or not download it here. Drop me any comments or feedback as always appreciated.

Net net..the new pricing ruins the benefits of having an OpenSpace. Keep an eye on the Linden Island statistics to see how many people drop them or convert them. There are something like 13,000 on the grid now--so it could be a lot of shrinking the SL Land Mass and a big virtual reality check for the braintrust over in Lindenworld.

related posts:

• What do people do on OpenSpaces/Voids anyway?
• M Speaks About Homesteads
• More Price Hikes Coming??
• Giant List of all metaverse posts here

Disclaimer: We had 6 void sims. 4 have been converted back in to a single full region. 1 is deleted and another will be deleted at the end of November. We also have deleted 2 fulls sims and plan to delete 4 more at the end of November.