LL-Sponsored Preview "Stress Tests"

As LL receives more resources for use, this might be a wise thing to do.

Case in point: Many of the bugs that were solvable in Preview went unsolved, simply because not enough data was taken on the stress levels of 1.6.

Solution: Have LL sponsor several "stress tests," similar to what companies hold with beta applications, for testing the wide effects of radical patch changes. For this, I don't mean a few sims on older hardware - I'm talking a "mini grid" on the current hardware implementations and huge user load there.

Runner - How to Get Users to Cooperate with a "Stress Test": Offer real-account incentives (example: L$500 credited to real account for participation). The economic ramifications of this would need to be weighed, but the benefits to the population as a whole would be better in the form of more stable patches.

This is a proactive, "for next time" solution to some of the problems seen with this patch.

They did a stress test.

Sometimes, though, you just have to release.

Maybe instead of the $L, LL could award special "preview/stress test/bug hunter" trophies. That you could show off in your house/land. Maybe even a mention about it in your profile.

Also the testing could be segmented. Like 3 days where you encourage people to test scripts, and 3 other days where people have to make crazy builds just to see if everything works.

Also send a newsletter about it to users, for the users that don't visit the forum much.

They did not have to release anything - the state that 1.6 has been released in is shameful. What was the point of a preview release if you were going to release with unresolved reported NEW bugs from the preview. It's very sloppy and pointless.

Incentives are great, but only if the company then actually acts on the information the testers provide.

First of all, I'm all in favor of the "When it's done" mentality as opposed to "patch and fix." Since LL commonly favors the former tactic, I'm trying to speak to that.

blaze is correct, however, in that sometimes it's more opportune to "release early" for the PR. This is a common tactic for MMOs and stuff in general, particularly around Christmas. However, when you have an installed userbase, you don't want to jeopardize that, either. This is far from the worst that can happen, but it does give us a glimpse of where we as a community might improve.

As to "they did a stress test," I feel the tests that were done were inadequate; indeed, a lot of the "test" phase is still in play. What I'm proposing is not a test phase on older hardware implementations and settings - it's a test using a significant portion of existing hardware, perhaps so far as to propose a "gradual" transition of the grid if it would provide better data (though, given the problems that brings, it would be a last resort).

At any rate, I think this all depends on the economic resources that LL has and the willpower to simulate high player load with a higher degree of accuracy. It's certainly not an easy thing to ask, but it is an area that I feel could be improved "for the next time."

I don't think more testing will help.

The fundamental problem is that the LL team is too small for such a big project. I gave up really bothering to post feature requests because I realised something.

They're not ignoring them, they just don't have the time to even think about them.

I believe the fundamental solution is to develop an architecture which the community can contribute better in terms of development and QA.

Some have suggested a client-api as a start. I dunno. Maybe we, the technical community, should put our heads together and really think about how they can architect to let us contribute in a way that lets them be profitable but gets us some satisfaction as well.

But, no matter what, LL is overloaded. They need to leverage the community.

Needs more cowbell.

Oh, and the use of 'fundamental'.

I disagree on this point. More testing always helps as far as uncovering problems are concerned. The Linden Hotline really drives that home; any company with the gall to allow users to have all of their concerns answered in a timely fashion is doing a pretty good job at addressing facts presented.


However, I will agree partially with this - and I frankly don't think the Lindens are oblivious to this, either. The only problem in matters like these is of the userbase, only a very small percentage can even grasp what problems LL must overcome, much less address them.

I believe that, ultimately, sims will extend to user hardware. But that's a long ways of. We're focusing on now, with a concern that is possible to address with current hardware options and, furthermore, could use our collective help as a community. If we're not prepared to take more simple steps like these, are we truly fit to rule Second Life?

Fundamentally (just for you maxx), I think that more testing will simply result in more bugs that don't get fixed.

At least we'd know why we keep crashing when we try to log on to SL. I have to say there is some advantage to that.

However, we also don't want to bog down the LL team with our testing if they're not going to fix the bugs. The reward (peace of mind knowing why we can't do things) don't really match up to the risk (time used to tract down the bugs).

What we need to do, I guess, is start having community leaders who can run the preview grid. Maybe you and Strife and a couple of other people could be responsible for answering questions, verifying bugs, that kind of thing.

Obviously you're both pretty busy, but if you share the duties amongst a large enough crew it might work.

Personally, I think a client API would have been the smartest. Perhaps in 1.7