MOOSE.page

My original post (changing the name to Moose for now– multi-user online operating system environment?):

> I think the 3d object space is really cool. There really needs to be a smoother way to browse through all your files, bookmarks, shortcuts, and more. I think it can be designed not far from how it’s shown to work well with how we think. With improvements in search (e.g. video speech-to-text) as well as ability to relate items together – this will finally allow us to not have to worry about organizing files. I think the Firefox 3 address bar goes very slightly in this direction, and I already use it extensively. In fact, there’s not much reason to separate desktop and internet anymore. I have keyboard + a) a little switch that I turn if I want to keep something private + b) big fucking knob that I crank when I like something. The more I crank it, the slower it drifts away on the z-axis, and the more readily it appears in my friends’ and peers’ 3D worlds when they are browsing + c) of course, a joystick for zooming through the files. Basically, now you have to click a link or save a file – all that is bullshit in the future. potential ‘links’ appear in your peripheral vision and you choose to explore them in they look good, which they probably will because dozens of variables have been analyzed to present you the optimal choice. saving will be irrelevent because high speed internet will be ubiquitous and if you just crank dat knob enough, the file will be virtually right at hand.

Welcome to MMOOSE: the Massive Multiplayer Online Operating System Experience Essentially this means there is a merging of the web and desktop interfaces that incorporates large-scale interaction with others online. Moreover, MMOOSE is a highly complex artificial intelligence agent derived from the mutually recursive relationship of society to mind, mind to society. 

## Why the knob is so important 

The web, as is the world around us, is composed of unfathomably large amounts of data. The human mind is efficient because it focuses only on that which is exciting to us. The effects of chemicals in the brain cause us to form a memory or take action. Thus, MMOOSE must use a similar concept to decide what to present to the user. Any object, universally, should be able to be assigned an ‘interest level’. In fact, I find the concept so important that I propose the ‘knob’ – a physical device for assigning the interest level to an object, just like we would have for controlling the volume. AI can only exist if it has motivation to achieve a goal. Most successful AI programs have a clear goal – chess bots are trying to win a chess game, and they have clear progress along the way. Only by analyzing the resulting interest levels can MMOOSE determine whether it is achieving its goal and how to increase its performance. 

## Limitations of the knob/Beyond the knob 

The initial design is very simple – any object is rated 0-100 in a single dimension by the user. The human brain is undoubtedly more complex, and while it may certainly not be the best possible design, but it is still a very good one for us to model. Multiple knob dimensions may decrease usability without much performance gain until we have a better understanding of the brain, so I’ll instead focus on automatic interest level assignment. In a highly active field of research called recommendation systems. Amazon, Netflix, Pandora, and countless others are using social data to recommend items to a user. The idea here is similar but far more universal – working across inhomogeneous objects will likely prove to be a huge obstacle. The following may be reasonable assumptions for automatic interest levels: - decreased interest over time with no - diminishing marginal returns for repeated viewings of the same object The web as a virtual world The beautiful thing about the world is that it’s yours. You can certainly download precreated themes. You can choose to have a very simple, clean world that looks much like a traditional filesystem. Interactions MMOOSE has just brought you a headline article about a crime in your city. The article was in fact assembled by MMOOSE long before any news companies got to it – a witness to the event posted it via MMOOSE and due to skyrocketing interest levels, it quickly found it’s way to your newsfeed MMOOSE exists/Believe in MMOOSE Pretty much everything I’m talking about exists in primitive forms: recommendation systems, socially aggregated news, 3D virtual worlds. The main problem is that our traditional forms of communication and organization can’t deal with the amount of data.

First I want simplify the bit around recommendations by again making a model from the brain. The saying in neuroscience goes “neurons that fire together wire together”. That is, if you are often feed when you hear a bell, then hearing a bell will make you think food. Similarly, we can form semantic links between objects (web sites, files, etc) that are often used simultaneously or in sequence.

The easiest way is to think of it is linguistically or even just by keywords, and, really, this isn’t more than a standard Google search. But ideally, we want to abstract to a raw brain-like pattern, whatever that looks like. Then Moose can take in words, visual input, speech, and even brain waves as one, and if you think about being burned, for instance, it will return pictures of fire, things with the word “heat”, sounds of sizzling, etc. Actually, better yet, it will return a (virtual?) bucket of water and tell you to stop drop and roll. Basically it’s a complete AGI (artificial general intelligence) project. But instead of being a brain, it’s the aggregate of all participating brains, as well as any computational processing that goes on in the background and submits its own input.

The only ideas that are remotely both implementable and original right now are the interface and idea of universality. Assume universality as I describe some ideas for the interface. As I was saying in the post, the idea of hierarchical filesystems and web results returned as pages of lists should be archaic. The benefit I see of imagining this creation as a virtual world is that it brings computers back to something that humans are more familiar with. Ancient — likely still superior — memory techniques involve imagining placing things into elaborate buildings and rooms and envisioning creative mnemonics, not lists or folders. That means it’s what we’re most comfortable with. But now, with the computer, there’s the power of changing everything instantly with a keystroke.

That’s all very abstract for now, but if I feel like it, I’ll put some more thought into concrete models and examples, and compare it to more existing technology.