The most interesting part of this talk was how many dead-ends they hit in the process, and how those dead-ends basically sunk many of the popular notions of how to improve programming.

- Make it like Excel = doesn't pan out

- Make it like English = doesn't pan out

- Make it visual = doesn't pan out

I find the end result (kinda like a Wiki crossed with a Relational DB) intriguing.

This looks incredibly interesting. I'd love to see what a compiler looks like in this system.

It also was interesting to see him mess up the overcharging query I would have liked to see a more realistic debugging situation.

Granger's 'Light Table' was always an experiment in my mind, but I'm kind of sad that he dropped it and moved on to the next 'let's fix' programming thing.

Was it a dead-end?

How does this compare to Logicblox?

This reminds me a lot of Lotus Notes. Which I guess is no bad thing: it was an amazing environment to work in.

Would it be fair to characterize this as a re-invention of that, or are there significant differences I'm overlooking?

Notes wasn't without it's problems though: version control and data typing being two important ones.

I couldn't but thinking that was John Carmack speaking.

I don't like the handwritten messy slides.