Epistemic status: exploratory.

I am using "advanced media projects" as sort of a catch-all umbrella for people working on better ways to present and explore information. I suppose the obvious hub for this kind of thing is places like the MIT Media Lab, but here are a few others I have been thinking about:


Alan Kay gave a couple of talks at Stanford about the time at PARC that I like. The title is How To Invent the Future, Part I and Part II. He talks a lot about the importance of tools and the under-utilization of computers for thinking purposes. I went to school for engineering, and I would particularly like to call out Ivan Sutherland's demonstration at ~26:30 in Part I because when people were still using punch-cards and tape readouts this guy threw together the first computer aided drafting program as a demo for a monitor and a light-pen and my life would have been much easier at several junctures if only we could have rolled back to the 60s. My mind remains blown.


I just discovered Bret Victor, and his blog has a section in it called Kill Math. I am watching the presentation called Media for Thinking the Unthinkable, which seems directly relevant to the problems this community is largely concerned with. His central argument is that all of our progress hinges on people with a knack for manipulating abstract symbols with pencil and paper; this both loses value that might be added by people without such talents, and fails to leverage abstract symbol skills in more than two dimensions.

In the Wild

This is exactly the pitch of Distill, which I discovered from LessWrong. In fact, the demos for Distill look very similar to the first network theory paper example in the Media for Thinking the Unthinkable talk...and following the last link of their examples is that very talk.

But the question is, does anyone here actually use or produce such tools?

It seems like superintelligence is the kind of motivating example which requires the ability to think unthinkable thoughts, but I haven't seen very much discussion about how to do so. Secondarily, we have a strong interest in the pedagogical possibilities from a rationality perspective. What if we could flip the former around, and build a tool to render previously thinkable-but-wrong thoughts unthinkable, like a kind of rationality filter?

New Comment
5 comments, sorted by Click to highlight new comments since:

I am extremely interested in this area and would love to work with other interested individuals in this area, or be directed to any relevant resources.

(Bret Victor and Alan Kay are both, as far as I know, responsible in part for this project, which I've visited and seems pretty interesting: https://dynamicland.org/ . I am not really object-level convinced that what they are currently building is a fruitful direction, but it's at least quite cool, and it's the right kind of new direction/idea.)

my life would have been much easier at several junctures if only we could have rolled back to the 60s

Bret Victor gave another talk along those lines: slides+vimeo, youtube.

(Wikipedia's portrait of him is from this talk, in ... costume.)

I've clearly stumbled into a memeplex to which I am favorably disposed. He quotes David Hestenes, whose textbook New Foundations for Classical Mechanics I just picked up a week ago in the name of filling in my math knowledge along lines more favorable to my intuition.

For the curious, this is the guy who is trying to popularize Geometric Algebra - you may have seen posts around the internet for Clifford Algebra as well, which is sort of the core insight. The lecture he gave when he won the Oersted Medal is here. I considered using this as an example in my other comment on pedagogical tools.

Quick and dirty visualizations have been used everywhere for decades. For example, here's how you plot a differential equation in Excel using only arithmetic and cells. I think the frontier of usefulness between visualizations and text is roughly where it should be, and Bret Victor's work hasn't moved it much.

Tangentially related: my guts suspect that when a conceptual tool is targeted to pedagogy, I should consider it for exploring something completely new. It seems like the increased focus on clarity and comprehensibility are more valuable in contexts where we can't rely on mastery or previous generations of tools to improve on.

It would probably also have an outsized impact when working on a problem that lies at the juncture of several fields who do not share tooling.