an exploration of the Secondlife platform as an immersive environment, a poltical space, and as a generative medium for architectural design.

Thursday, March 23, 2006

An open letter to Ben Linden

Dear Ben Linden,

First and foremost, I would like to thank Linden Lab and yourself on behalf of myself and my collaborator, Nicola Escher, for the opportunity to build our vision for the German Orientation area. We enjoyed working on the project immensely and learned a lot in taking on a project of such a large scale. As per your request, I am writing to provide feedback on our experience of the entire process. Here is a brief of some of the thoughts we had while working on our reinterpretation of Chateau Ulrichsburg.

the ruins at ulrichsburgwilkommen zu secondlife

1. The Request for Proposals

As opposed to other Linden RFPs, we found that with this particular project LL was beginning to take the best approach to the process of selecting a designer. Previous and subsequent RFPs posted on the forum have generally been managed as competitions that offer paltry compensation for work rendered sometimes with unlimited licenses for as little as 5000 L$ (20 USD). This project was promising in that in the proposal, the design team could specify the expected compensation for the project.

Where this process fell short of its promise was that the schedule initially gave several weeks for the proposal and a week for execution. The RFP was framed such that design teams were implicitly expected to provide spec work in their proposal to be able to meet the deadline. This is an unacceptable risk for any artist or designer to commit 80% of the work without any assurance of compensation. While we were very excited to participate in this project, it is for this reason we did not enter a bid in the initial round. This setup benefited us personally as we submitted a very schematic proposal that I put together in an afternoon and worked on the project for weeks open winning the bid. However it did not benefit LL as the RFP failed to gain the interest of other professionals.

2. Building

There are many problems that currently plague large scale projects in SL. While the local and reference plane have been very helpful, there are still many things that are lacking.

Limitations on Camera distances. – This can be a very immersive feature that reinforces the avatar as the anchor for our virtual experience, but can be very frustrating when building projects on a quarter region scale.

Limitations on selection distance – This feature, which also reinforces the avatar as anchor by recognizing its sphere of influence compounds the problem of the camera limits. To select a large grouping of prims one consequently has to physically move the avatar to a closer location.

Camera resets when avatar moves – As soon as the avatar is moved, the camera resets to the default position behind the avatar, losing the required view for building. But more importantly –

Objects deselected when camera resets – This means that one can not physically move the avatar to a new location to select distant objects. To do so requires a scripted propulsion system. Perhaps new systems will use HUD displays and sophisticated telemetry to make this intuitive. However, the most intuitive solution would be to allow greater camera distances and selection distances for builders.

Small objects disappear at a distance – This is a great feature to limit fps lag on the client, but for a builder this can be a drag. It should be a feature that can be toggled on and off in a clear documented manner.

3. Texturing

This is definitely one of the most tiring aspects of building complex builds in SL, particularly a ruined castle that we were attempting. The texture controls constantly require manual input for properties which can be quite time consuming when attempting textures on one prim to the next. The arrow buttons make very imprecise jumps. A good solution would be a radial chose to select the precision of the change the arrows administer.

Texturing would be made much simpler if it too became graphically interfaced as building is. What I suggest is way to stretch and move texture tile on the surface of the prim directly with the pointer, dragging corners, sides on the surface as if it were a 2d workspace as in illustrator or indesign.

4. Blocks

SL needs blocks, clones, or what have you. If a builder were able to identify an object or a group of linked objects as a block that could be updated automatically across the build by redefining the block it would immensely increase productivity and offer the opportunity to spend more time in designing and experimenting and less time in menial production tasks.

5. Layers

If a player could tag a series of objects to client side defined layers, this would give the user the ability to turn on and off objects in his or her own view. While SL builds have not necessarily reached the complexity to require something like this, I feel it is worth considering how this kind of functionality could be added in the future.

6. Terrain Editing

Overall the terrain tools were adequate for this project. It would help immensely to have the ability to save different terrain settings to a list that can be accessed by the client. It would also be beneficial to allow faster rates of change to the terrain tool.

7. Divergent Explorations

It is the norm amongst designers to take multiple stabs at the same problem, each iteration offering different possibilities. The platform is quite capable of accommodating this at a small scale, but when large scale projects are attempted, players hit upon a major snag. It is very inconvenient to package up a build into ones inventory and to then rez it in an identical condition later. It is an imprecise and time consuming endeavour which interrupts the creative process. I think it would be very helpful to allow players to save instances of the sim. These instances could either be downloaded or if that is a concern, saved into ones inventory. If the instance item was dragged onto the estate tools somewhere, it could allow players to reload that sim state. It would be really amazing if these states could be saved into the estate tools for quick changes from one to the other. This would allow developers to work on multiple strains of a project or several entirely different projects without the need to purchase multiple sims. One region would be all any developer needs to work on a multitude of projects.

In closing I’d like to emphasize the positive aspects of our experience. Your approach as a client was very easy to work with. Firstly, you provided us with a clear idea of the theme you envisioned yet it was open-ended enough to allow for us to really have a sense of ownership in fleshing it out. Secondly, you were very responsive to our needs, whether it was managing the estate tools, offering us technical information on the rendering client to determine our texture strategy, or offering suggestions and alternatives from your intimate knowledge of the platform.

the chapel the arcade
waterFALLingwater birds eye
(waterfall by fallingwater :)

We look forward to seeing the opening of the German Orientation Area and the neighbouring Welcoming area. We also look forward to future opportunities to work with Linden Lab in creating memorable institutions for our shared virtual world.


Jauani Wu

note: this letter is a copy of the post-mortem requested by Ben Linden and published here with his consent.