Alphaville Herald story on Viewer 2.0

I just saw an Alphaville Herald story on Viewer 2.0 that I think might be of interest to some:

http://foo.secondlifeherald.com/slh/2009/11/alexa-lindens-prehalloween-treat.html

Apparently some feedback from the testers got into a notecard inworld, or something like that.

It's too long to snip excepts from; I'll just say the comments aren't very flattering to Viewer 2.0.

It is a work in progress and this is the reason for heavy NDA's so please take anything leaked with that in mind. Hey, I have no big confidence in LL's abilities but even so, please do not let alpha testing comments be much of an argument that forms your opinion.

"Transforming the Second Life Experience" (https://blogs.secondlife.com/community/features/blog/2008/11/03/transforming-the-second-life-experience) back in November 2008 announced LL hiring "Big Spaceship" to do their UI for them.

If after a full year they're still in the alpha/design stage then we won't see a new LL viewer for another year at least which just isn't realistic at all.

It's far more likely that they're at the "what's there is what we'll get" stage and we can look forward to a mangled bug-ridden viewer that'll take them 2 years to get as functional as what we have today.

All I know is I want your viewer with the ability to have embeddable links in notecards and notices Kitty!

Oh, I am right there with you. LL has plenty of examples of poor performance and, dare I say, flat incompetence, without having comments leaked in the middle of development.

How about a baseball analogy. If someone was going to signed on Daryl Strawberry and news was leaked about his current physical fitness level ... well, he will get in shape with training and practice, but that is really not the point nor the reason for concern, is it?

This part of that link bothers me:

What are they testing if not the UI? Core viewer functionality testing, independent of the UI, calls for automated unit or integration testing, not pseudo-beta testing at the system level by what I presume to be a group of volunteers.

Or, if this really means not reporting usability issues, then why are they wasting time testing things that might be totally redone due to usability issues?

Any sports analogy will probably just have me nod like I know what you're talking about .

I honestly wouldn't really care how far or how little they're along if LL hadn't put all their eggs in the "2.0" basket though.

We're not getting bug fixes for bugs that have been reported and have attached patch fixes for months and months now and we likely won't see any of those fixed for many months still.

It also looks like they made the wrong call a year ago: growth has sputtered and all but flattened out so wasting time and money on the "new user experience" is probably misdirected.

It's far better and far more useful to complete the things we already have by getting rid of all the little idiosyncrasies we do to work around things that don't work quite right which automatically helps new residents because they won't have to find out about those and learn how to compensate.

If you look at any little thing we already have there will be some rather big gaps:
- we all have several 10k worth of inventory but not really any good way of finding anything in it (and that's assuming you're not one of the seemingly increasing crowd of unlucky people whose inventory doesn't fetch all the way)
- reorganizing inventory is a pain as well; things like forcing an automatic filter refresh whenever someone logs on or off or when you're given inventory makes it even worse
- we can rez and return prims on/from land but there's no way to actually see what is rezzed and where it's located
- we can uncheck "Show online" but it doesn't really actually work
- you can kick someone from a group but you can't prevent them from rejoining unless you make it invite-only
- I think half of the IMs I sent from offline by replying to an email didn't actually make it through and some offlines from other people didn't make it to me in return (either in email or when I logged on)
- some clever Linden at some point differentiated between "objects" (that are to be rezzed) and "attachments" (that are to be worn) but it's only half hooked up to anything. If a newbie could tell the difference between something they can wear and something they should be rezzing that's a bigger step forward than anything I've read about 2.0 so far

And dozens upon dozens of other little things that don't work "quite right". It's really boring stuff to fix for the most part but what I would expect from "2.0" is a *finished* viewer, not fluff for fluff's sake or making things even harder still.

I read the emphasis there on "weekly"; it appeared to me that the feedback solicited for that particular testing sprint was to be focused on the communications tools (IMs and chat, I guess) not on other UI functionality. Of course "communications" is part of the "UI"--hell, the whole viewer is UI--but maybe previous week(s) had generated a lot of feedback about buttons or other UI stuff and development was busy fixing that stuff, so the testers might have been asked to focus on some other aspect of the viewer.

The feedback in that mail is all about usability, but within a fairly narrow range of viewer functionality. It's mostly not the sort of testing readily automated (which is not to say that LL does as much automated testing as it should).

(Full disclosure: I'm a member of BSI and got Alexa's misdirected email, but I'm not a tester for Viewer 2.0.)

Hmm seems like they are breaking all the good stuff.

Ifwish they would have just stayed on the old viewer or Snowglobe path, why can't they put as much effort into real features like avatar mesh upgrades , more attachment points or puppeteering or even finishing windlight or the age verification system.

Ditto. Especially the avatar mesh.

I understand that Viewer 2.0 is a work in progress but the comments in that email suggest they have started off on the wrong foot altogether - most probably by having the thing redesigned by a team with eyes so fresh they never spent any significant amount of time in-world.

The CEO himself doesn't appear to be much interested in communications if this comment on the thread about a recent interview with him is anything to go by: . The unwieldy interface of his beloved Viewer2.0 would go a long way towards explaining why.

This is the new compulsary viewer that Blondin promised all us "Adult refugees" would be released by the end of this year after LL backtracked on the promise of 1.23 being compulsary....

Looking at that email, i'll be suprised if it's compulsary by the end of next year...


Gomez

That's the problem: the time spent on this Viewer 2.0 thing is just a killer--and not only for adult content providers still awaiting 1.23 to become mandatory.

1. It's killing open source development of Linden-distributed viewers. The Viewer 2.0 source is hidden from outside developers, with no documentation of what parts have changed or been replaced. Who the hell wants to develop on a branch when you're prevented from knowing what parts of that code base have already been replaced on what will become the main branch?

2. It's killing acceptance of all Linden-distributed viewers--itself included. As the calendar slips by with no updates to the current standard viewer, more and more people are using Emerald and other third-party viewers. By the time Viewer 2.0 is in GA, nobody is going to want it (nor Snowglobe, nor anything LL will be able to pull together) because everyone will expect the features and functionality in their choice of third-party viewer--features that could have been in Linden viewers were it not for the opportunity cost of 2.0.

3. It's killing Linden development. Practically every warm body who can spell "viewer" is now pulled off on this thing. They can't talk about what they're doing, they can't get any feedback from the general public, and they can't see any of their work in general use. It's not a happy time in the code cave.

It's time to pull the plug on this thing and regroup on some strategy not chained to sunk cost. One approach would be to swing development over to getting a rock-solid Snowglobe version out as a 1.24 release candidate this month and as the standard download by year-end.

Maybe they've learned something useful in the 2.0 effort that they can resurrect at some point as features for the standard viewer. In other words, maybe they can recoup some of that sunk cost. But just watching one's entire dev team become irrelevant and demoralized is no way to run a tech company.

Maybe not pull the plug as much as pick-and-choose which things have been refactored and finished and plug all that into the 1.23 code along with the variety of bugs that have been fixed over the past few months and keep on releasing a 1.24 and onwards.

SL-2.0 can keep being developped at its current snails pace and they get the benefit of live testing the new code under the old look/behaviour.

SL-1.23 RC0 was released in April 2009 and was then rushed through RC to make the "adult only" deadline; 7 months without any kind of viewer update, not even bug fixes, really is just getting to be ridiculous.

At this point I'm almost convinced that the "cone of silence" around 2.0 isn't because they're trying to have a spectacular release but out of embarassement that nothing about it works. If they mangled up chat *that* badly as the email suggests then everything else probably isn't in much better shape.

Edited to add: although... maybe if everyone starts publishing outrageous rumous about 2.0's new features then LL would have to start talking about what's actually in there or people will hate it for everything they didn't put in that others made up .

when i hear 2.0 all i say to myself is..will i need a new computer.
is it just gonna show up after like 1.50 or 1.75..
do we have 150 versions before we get to 2.0...every version we get like 2 or 3 versions..1.23.4..1.23.550..
if we stick to this rate of releases i'll be dead before it ever gets here anyways..

2.0 is LL's magical cure to the low retention rate of new residents (under 10% which *includes* newbies who are actually alts) so probably not.

Past "easier to use" enhancements:
- instead of having a separate friends/groups floater we now have the lovely communicate window which can't be sized down as much as most people want it to be
- the Tools menu playing hide-and-seek for a while (they reverted that but it's slated to make a new appearance in 2.0)
- changing the pie menu around, kicking off options we all use and replacing them with options few people ever use (not that we'll have pie menus anymore)
- enhancing the Edit window by making it so that you can't collapse half of it anymore

One of their favourite things seems to be coming up with new ways to make sure there's less and less available screen space to actually see the world so if we'll have to buy anything new I'd guess it would be bigger monitors .

That could well be, and would be fine. Certainly the fragment I quoted was out of context and vague, so I had to infer a lot.


That's a common, but suboptimal way of gathering usability info. It's still not clear to me whether LL is open to applying that sort of usability feedback, or whether they're just picking and choosing features to prioritize from concepts they've already decided.

I applaud your disclosure, thank you.

It's SOP for software development companies working on projects of this size to have separate teams working on current and next generation development. I doubt there's any public, quantitative data on the relative allocation of resources. Thus we couldn't say for sure that the 2.0 basket is unfairly disrupting the 1.* bug fixing and improvement.

Of course, there's the indirect, motivational effect, which is even harder to quantify. If the 1.* team keeps getting told that 2.0 is just over the horizon, they're not going to be interested in fixing problems that would be time consuming for them.


I don't follow this. Surely a dramatic improvement on the new user experience could result in a major growth boost. Whether they're current approach will succeed at that is a different question, but investing resources into the new user experience seems like exactly the right thing to do to improve growth.


If the general goal is growth, and the specific problem is getting people to stick around past point A, where point A might be the first couple of hours on Help Island, then fixing things that new users won't see until they get to point B isn't going to solve that problem. Sure, it's a worthwhile thing to do, and one what we'd all appreciate, but how well does it serve a business goal for LL?


I doubt these are problems affecting retention of new users.


I don't understand this. When I rez something, I see it right there - assuming it's visible and wasn't actually rezzed in a way that causes it to be some other place than I expect.


Again, these aren't retention problems.

That's a great idea, but it would require content developers to manually identify objects intended to be attached. Perhaps there could be some automation, so that if a creator wears something, it gets marked as such, but there would have to be a way to override it. There's a huge number of existing products that would need to be updated as well.


Software development simply doesn't work like that. I wish it did. But it's extraordinarily difficult to make a business case for fixing those things.

What gives you the idea that there's any 1.* team?

It's been 7 months since 1.23; "transparency layers" (to replace invisiprim attachments) is one of the things that was partially added to 1.23 and has since be completed but they're not allowed to talk about a release roadmap to name one example.

No bugs are fixed, no new features are released. 1.X development is dead.

Again, you're assuming there is anyone still working on a 1.X.

If that was the case then 1.X would be the main development branch and 2.0 would occasionally get the changes merged into it and the problem you're talking about wouldn't be an issue.

Retention is low because of the nature of SL, its hardware requirements and a whole bunch of other things that are actually server side problems or need the necessary sim-side support in place first.

If someone tries SL and always like a fluffy cloud to themselves and their tp's fail and they can't get anywhere or search is so horrible they can't find something they're interested in then a flashy UI isn't going to make them stay.

Type to chat, arrow keys to move. If someone can't figure that out then they're simply beyond help.

My actual point was though that any improvement of the "new user experience" should have been done in mid-2006 when growth was exponential and you had millions flocking over and giving it a try. Growth is flat now so they should be focusing on making things better for the people they already have.

I never said it would affect new users; I actually said the opposite.

Rez a prim and type in a random number for Z.

How are you going to find it? Spend 5 minutes flying up and down hoping you spot it?
How will you know it's one you sent flying off and not one you accidentally sent up in the air in the past?

What if you're dealing with a full sim and you want to get rid of the "junk" that inevitably ends up in the air?

It doesn't require anyone (outside of LL) to do anything.

Any prim that someone picked "Wear" on while it was rezzed will be an IT_ATTACHMENT (IT = inventory type) already. Rez it back and take that same linkset into inventory the normal way and it'll be an IT_OBJECT. Both are AT_OBJECT (AT = asset type) which makes them interchangeable (just like snapshots are really just the same as textures aside for the viewer displaying a different icon for them).

You most likely have hundreds of "attachments" in your inventory already but there's simply no way to tell them apart from object; the part of making them interchangeable was all done.

So all that's needed is for the sim to "auto-update" an item's type from object into attachment when it's worn from inventory (and a "Revert to object" would be nice although that's possible already). "Wear" on rezzed items and llAttachToAvatar already have the code in place.