Demise of SL

I've experienced what Eboni is talking about.

On my alts, SL runs a lot smoother for me than my main. My main has about 10K items...

OMG!
Is SL on it's death bed, again??!!

/me freaks out....again

Wasn't this entire point just refuted by a study we discussed here recently? That over the course of a week, a sizable chunk of the existing membership does actually log in (maybe as high as 70%, I can't remember the number).

Anybody remember the study in question?

Hmmm, well if your alts are anything like mine they aren't tooled up anything like my primary.
So, it's not surprising that an alt without: prim hair, fancy textured outfits, hoochie shoes, bling stuff, prim attachments, fancy scripted hud thingies, etc, etc is faster than an AV with all the 'goodies'...
Not saying it's definitely the case for you, but it definitely is for me! And I think we're roughly the same age too.

You are right about the numbers though - it's largely smoke and mirrors - but I don't think you are right about the reason why so many sign up, come in a few times and never return.

Firstly the hardware requirements are way to steep for the mass market. I happen to have a pretty high spec PC and SL is (bloody stupid updates like 1.7 aside - and how many noobs joined during that and never returned?) good. I get high enough frame rates and I'm not running into hardware limitations - but...

I also happened to have a PC that marginally exceeds the minimum system recommendations (not requirements) and logging on with that is not a pleasant experience at all... What it must be like with a minimum system requirement PC doesn't really bear thinking about!

Secondly there are the ridiculous default graphic settings. They might have been designed to make every noobs initial experience with SL as negative as possible. I don't know about you, but it took me some experimenting to get a decent frame rate and the first couple of hours were really horrible. OK, I'm reasonably competent and managed to figure out what was wrong and fix it. How many noobs are?

Thirdly there is the UI... Ye Gods! I mean, how unfriendly can you get? IF LL had employed someone to design it (which I don't believe they did, it's so bad it has all the hallmarks of something that 'just happened'), the individual would be better employed as a lavatory attendant!

Fourthly there is the really poor software design. Where to start? Well, how about software that streams mini-map data when the minimap is off? I mean, it might only seem a small load, but multiply it by the number of logged on users who aren't using their mini-map and then consider the load. Anyone who doubts the inefficiency of the SL software only need take a look at all the redundant data that is continually downloaded to the client and winds up in your graphics card RAM... I'm sure there are other examples, but that'll do for now.

There are four good reasons why noobs show up, download and install the software, log in a few times and vanish. And making free basic accounts hasn't helped. If it's free and it doesn't appear to work, you just shrug you shoulders and uninstall the junk...
Maybe I was lucky - it cost me $9.95 to sign up - so I persisted. Would I have if I hadn't already invested a bit of money? Maybe not...

Here's the blurb from slashdot:

http://games.slashdot.org/article.pl?sid=06/01/11/088242&tid=209

Terra Nova has an interesting discussion going, talking about what really matters when we talk about a virtual world's population. Total registered accounts? Accounts logged in since last month? Concurrent users? Interesting stuff. From the article: "In a similar vein we discussed Second Life's 100K+ members, a figure which I and others have questioned here on TN. Cory Ondrejka said that SL's 'concurrency numbers are rapidly approaching 4500, about 17,000 residents were in SL in the last 24 hours, and 50,000 in the last 30 days... If you go back even 90 days you get about 90% of the accounts having logged in.'"

Yes, most of them noobs struggling a bit before giving up for good... Statistics can pretty much be used to 'prove' what you want. My 'figures' are the number of noobs I meet who, after a week or so, never show up again.

I'm sure you meet plenty of noobs who feel plenty of ways, and I'm sure you're reporting that correctly, but the plural of anecdote is not data. I'm a scientist, and personally I do trust stats more than anecdotes myself.

I certainly hope that Second Life does not become broken into several worlds, though.

They just have to solve this asset server problem. And if they limit inventory, people will just divy upt the invetory between alts and have the same amount of stuff.

I wish that you didn't have to load the same textures over and over again--leave a place and come back one minute later, all the textures have gone gray again.

SNAP!

However, data isn't perfect. Statistics can easily be misleading, especially if the method of obtaining the data is flawed, either deliberately for reasons of proving one's point, or because they are being 'interpreted' incorrectly.

Anyway, time will tell.

How many avatar name/key pairs do you have in your database now? You posted that there was around 250,000 accounts overall at the time we hit 100,000; what is it now and can you link the source for that information?

I don't get the connection between free accounts and loggong in once and leaving. I would think it would be more logical to argue that free accounts increase the number of people who stay past 7 days since that was the point in time that the free trial was up and one had to pay $10.

General question to anyone:
Are Cory's numbers in the related link (thank you Luciftias) more or less credible/believable to you than any other numerical report found in an outside source as opposed to being provided here on this website? Say the numbers associated with earning potential which Philip has put out there, as an example.

As a scientist, what is your confidence level in the opinions which are put forth and based upon data analysis of the meager, imo, amount of information that is verifiable?

I log in every 30 days to decline inventory and keep my account "active". I stay logged in about 2 minutes. Nice to know it counts for something.

How are these numbers calculated? They seem to be very different from numbers given by LL.

So the 2 weeks after 1.8 and 1.7 when nothing was rezing, do think newbies thought "Wow, this is the best thing since sliced bread! Sign me up!". I guess we aren't using common sense, just L33t sense.

All of my alts are Hucci'd out. They have the same attire/attachment points used as my main, they just have small inventories.

Server - as in software, not hardware.

Sims are independed. A sim that crashs will not take the other 3 with it. Additionaly, a server (hardware) machine that runs 4 sims is usually doing it because it can handle 4 sims without a hitch. So, as far as the system concerns - each sim is its own unique server to and fro.

I graduated from a top ranked Computer Science program with honors. I understand the technical concepts.

One physical machine means shared resources, saying dedicated server is misleading. Running sims without a hitch, LOL. I almost had diet coke come through my nose.

Here's an interesting and relevant article at Joel on Software that I read several years ago. Good Software Takes Ten Years. Get Used To it.
People might argue with the "get used to it" part but don't let that put you off. The whole site is worth perusing if you're interested in software development and usability.

Heh, I don't see where I dismissed anyone's problems. I can only share my experience, however. Through my experience, I relayed what I found to be slowed on my main account: inventory. Maybe you could share some details? Saying "new accounts lag" is like saying "my computer doesn't work." Well, does it turn on? Does it boot up? Do you get a blue screen? More details are needed before people can address your specific problems than "it lags!" I never said "It doesn't exist!" and most people don't. What I *have* seen is people scream "OMG! SL sucks! It lags!" without giving any *useful* information and then complain about no one caring. Give as much detail as you can, and people will be willing to try to help.

4% is not a conversion rate. It means that 4 out of 100 people with accounts are on at any given time. People have jobs. People have lives. People do other things with their computers - shopping, browsing, games - SL isn't everything (yet). Where have you seen any numbers on SL's conversion rate? Is it higher than banner click-throughs? Where do you get your data? I'm not denying anything, but you're making up stats here.

So what if Microsoft is profitable with a majority market share? That wasn't what we were addressing. We were addressing the fact that the most successful software package in history was also one of the most unreliable, buggy pieces of crap ever. And an improvement on their former releases like Millenium Edition! Stop the bait and switch, please, and get some numbers for your sources (IE: conversion rate stats) instead of pulling thing out of the air, please.

Regards,

-Flip

The numbers of 250,000 were provided to us by LL at one point; I forget the exact source. Adam Zaius may remember, as he was present (I believe it was on the #secondlife IRC channel). Adam, if you're around, can you help out here?

I'm currently at about 75,000 keys. A few months ago, I moved to a much more passive collection system, however, as the bandwidth of constant emails hitting my server was beginning to cost far too much money. The world has also grown so large that collecting keys is becoming more and more difficult with any kind of regular coverage.

The connection between free accounts logging in once and leaving is this: if you pay $10, you're far more likely to give something a chance and work on it rather than just say "forget it!" and leave, due to the initial investment.

Regards,

-Flip

Flip,

Older acounts with larger invenotries are laggier than newer accounts with smaller inventories. If I spend most of my time building in SL for the past 2 years, should my account be rendered useless because *gasp* I am actually using it? We had this debate a few months ago when they said our inventories were too large. People have paid for with real $$$ these items, built them with time, effort and creativity, why should they have to give them up because the system is poorly designed?

You are confusing terms. A max 4% of the population is logged on concurrently, many other online worlds have numbers upward of 25%, so 4% in my opinion is not impressive. My point was 4% concurrent users is not impressive in this market. I am not making up the 4%, that is pubicly available information and I came up with the 4% using my graphic calculator. Conversions rates I won't say, because I am not sure if it is public knowledge, and I was told by a Linden.

I am not baiting and switching. You brought up Microsoft and I made a counterpoint. Don't compare apples and oranges, unless you want to continue the apples and oranges debate (and apples suck). The majority of accounts are not premoum account holders, people are buying in. (Also you really can't compare WinME to WinXP because WinXP is built on the NT Kernal and WinME was crap built on crap Win95, WinXP Home jumped the shark fromt hat line)

Good Lord I am bored at work.

To clarify, I was comparing the free account and the 7 day free trial period from before. $10 wasn't committed until after the first week. Why there would be more one night stands with one over the other eludes me.

hehehe, that article is great. I haven't read it in a few years, but its always a good reminder. I second Fallingwater's recommendation; and remember, Joel was writing about applications that were far more simple than Second Life by leaps and bounds.

Regards,

-Flip

hehehehehe, I know the feeling.

As far as older inventories lagging, I think its a twofold problem for older accounts. One, the inventories are huge, so naturally, they'll take longer. Also, they may take longer to load since MySQL can retrieve the most recently inserted rows more quickly than rows inserted previous just by the way it stores data, even when indexed. I wonder how often LL has a chance to run the MySQL "optimize table" command on these tables, which I've found to help quite a bit in my own database experience.

However, I don't really experience any more "lag" per se on my old account, FPS rates seem about the same. I can move around just as well, build just as well, the only thing that seems any different is the inventory name/key load and texture load taking a bit longer.

You can see some of my ideas on improving the local cache here, to cut down on server traffic and store more locally, with more options:

/108/12/81593/1.html

Keeping more locally would cut down on the constant inventory streaming that seems to happen for those of us with bigger inventories. A work around for now? As soon as you log in, type "a" into your search box in inventory which will stream the contents (item names/keys) via the search. Stuff you haven't used in a long time, especially big textures, will always take longer to arrive, as I'm sure there's priority code in the caches at various levels (squid, etc). This means it would have to go through a bunch of steps to get to you (not sure here), something like: asset server --> squid --> sim --> you, after the appropriate location has been found. Again, the overall network topology isn't my strength, but I hope that clears up what I'm talking about with inventory a little bit!

Regards,

-Flip

I have seen you say that a number of times and not heard anyone say yes it is true or no it is not true.

You take so much flack for many of your posts I would think for sure that if you were wrong about this there would be all sorts of comments pointing it out. But the problem you describe is such a bad problem that if you are right I would expect to see all sorts of griping and complaining.

I haven't been in-world much lately, the gray loading textures and impeach bush signs are kind of depressing.

I do seem to see some new problems in the display since the last revision, problems that occur with changes in distance. I could well be seeing the effect you describe and just not have noticed it.

It is hard to believe they could make a general cull small objects and call it a cull small attachments by mistake, or hope to get away with a trick like that.

Sigh, I didn't even start SL up on my lunch break. Oh well, maybe tomorrow's update will fix something unexpected.

I don't see why distributing the asset server is so difficult. I have to admit I don't know the actual architecture behind the system, but it seems to me that it boils down to: you download the geometry data from the sim, it tells you to look up asset ID # so and so.

Your client then requests that asset id from the asset server. The server looks up the asset in a big database (on a huge disk array) and then transmits it back to you.

What if on the client side it looked at the last digit in the asset ID and went to a particular server based on that digit? You could install a second server and have it service all of the requests with an even last digit. Populating the second asset server's database would take some time, but I'd be willing to endure a day of downtime if it didn't mean waiting 5 minutes for the assets to finish fetching when I zone in on a sim.

Or, if they don't want to change the client as much and want to spend a bit more money, they could just replicate the asset server and put a load balancer in front of it. It does require keeping the databases synchronized, which is not an easy problem, but one that has been solved in the past.

All in all, I think the first solution is more elegant and easier to implement. There must be something odd about their setup that prevents the Lindens from implementing it.