The Biomedical Paper

Published: 2 Dec 2012

The scientific paper — the one that has taken me away from populating this site — is done, at least it is so far as we are concerned. It has been reviewed and sent off. This post is going in the Kutachi blog section, but could be in the FilmsFolded area as well, because I think a few words on where it fits in are due.

You can download the paper at this yousendit link for the next few weeks. It is 100 pages long and the pdf download is at about 12 megs. I’ve written it for a general audience, using the Scientific American approach with lots of illustrations (meaning you don’t have to know any math). The paper describes a new approach to what used to be called AI, and demonstrates it on a biomedical problem. You may find that problem by itself interesting; it concerns how your sense of smell refreshes itself, and how a better understanding could lead to better insights into how the body works.

It took so long because I had to learn many new things about biology, medicine and the medical establishment. We also took the time to better design the tools we will be building. Most of you are here because of the movie work. Here is how all this fits together:

Topoiesis

Some time ago, I started work on a new approach to modeling the world in a machine. We call the general approach Topoiesis, a portmanteau word from topology and poiesis. The first part refers to a focus on the form of concepts and their dynamics while the second part is the object: creation and creativity. There are some spooky ideas under the hood, but the overall approach is accessible I think. I gave it a good shot in the paper.

This was sponsored by the intelligence community, because they have tremendous problems and there were other good fits. One good fit was that I expected the work to enter the public sphere and do good. There are more examples of this than you suspect. But I was disappointed.

Some substantial effort in the DARPA years then went went into applying the work to build infrastructure for creative collaborative work. This is quite a bit harder than you might imagine and more than a few large problems in society come from the poor tools we are using instead. I think things are getting worse quickly on this front. I still have something of a presence in this community, but probably have done as much as I can there before I go.

My close collaborators want to work on medical research, so that is where I have gone now.

sHeherazade

This is all about building real tools that work better than the old ones. Creating an elegant philosophical structure is far, far easier and these abound without making their way into reality. So it is all about tools and our slogan is to work on the hardest parts first. An ultimate goal would be an intelligent ‘computer.’ This means one that can make sarcastic jokes and be disappointed in love — one that may not be able to make good art (because few people do) but who can appreciate good art, especially narratives. And perhaps one that can understand why better than the average Joe.

We’ve named this speculative being sHeherazade, or sHe. We’ve shown some designs for intermediate tools in the paper, and one of them, the ‘narrative modeling tool’ is planned to be hosted here in due time.

FilmsFolded

I am imagining the FilmsFolded work being fun in and of itself. It was when it was active before, and I believe it can be more so in the future. I’d continue to do it even if there were not a direct contribution to the larger work.

An understanding of narrative, especially the rich narrative structures of film, is essential in the Topoiesis agenda. The next steps with FilmsFolded is to get the comments database back up, write essays on key ‘folded’ concepts using example films, perhaps characterize some films in depth (presumably the *fours*) as complements to the concept essays, then possibly model some parts of those films formally for the tool.

Kutachi

The Kutachi work may be of less general interest, but be a whole lot more fun to see. The reason behind the project is that we desperately need some user interface ideas. No one has ever modeled what we are. Well, of course they have, which is why we have art. We need something between math and art. Here is a mockup of one screen of one of our tools. It is also in the paper.

GeoKabbalitter

Ah, we won’t talk about this now. But it fits in.

The Dynamics Modeling Tool

(Image) A screenshot of a workstation modeling tool.

The Central Pane

(Image) A mockup of the Dynamics Modeling Tool.

Some Scrubber Ideas

(Image) Three integrated scrubbers.

One Face of the Cube UI

(Image) A tool for discovering new dynamics.

The LHS viewer — This fairly well developed design is the face of the cube designed for dynamics modelers. The structure of the cause of the story so far is projected onto a torus, with the points and connections subject to forces. The modeler will redistribute how they snap by direct manipulation, or moving the center 'column' which is a library of dynamics which govern the forces. The color of the stars indicates the mode in which they can be viewed in logic. The iPad is used here as the input for a larger screen in a darkenend room.