Hi <first name>, I enjoyed our phone talk the other day. Here are a few thoughts before our next call.

First, to review, Tidepool™ is our desktop application (built with Java) that people will use to organize their photos & memories. It's in the same marketing space as Adobe Photoshop Album and Apple iPhoto, though we have no illusions of beating Adobe, Microsoft, and Apple at their own game. We're just hoping to skim a little off the top, meanwhile bringing a more authentic, more human, approach to the task. Memories are precious. They form the substance of our shared humanity. We want to be the company that understands this. We don't want to be the makers of "Microsoft Office for the rest of your life" (Apple tagline for iLife).

Storymill™ is the web-application version of the same technology, built using Java 2 Enterprise Edition, that allows communities to collectively tell their stories using text, photos, sound, and movies. The idea began with Lehigh Valley Storymill, a website dedicated to grassroots publicity that currently uses the initial proof-of-concept version of the Storymill product, created in August 2002. A much expanded version will go live this spring.

You may be interested in what we're doing to try to get our community motivated to annotate their memories. It's of course the other big challenge, since as I've said, most people keep all their email in their inbox. Remember, we're not just talking about 50,000 photos on your own hard drive. Storymill needs to handle potentially many times that: thousands upon thousands of very interconnected, very browsable, photos and stories.

I'd also like to talk more about visualizing "useful interconnectedness" ... how it's essential to the problem of someday having to sift through a refrigerator box full of photographs. Yes, there have been other graph browsers, which till now haven't really caught on. If you like, we can talk more about my current thinking on UI, particularly the interplay between spatial persistence of screen objects (things stay where we put them) and more fluid, machine-controlled, arrangements (nodes and links adjust themselves according to context). Both are necessary ... it's the mixing of them that's a challenge, particularly given that we're building this for non-technical users.

Also key to the UI is the division between public and private information. Semantic arrangements will have to switch between these two modes easily. Imagine you're looking at a cluster of nodes with your family members, with photographs and sound clips connected to each person. When you switch to public mode, everything re-arranges itself to show social relationships on a larger community scale, with your softball pictures now connecting to other softball teams and leagues, and your Poconos photos now connecting to other people's photos in the same locations. The fluidity of this switch and the balance between human-control and machine-suggestion will be essential to the whole endeavor. Yeah, we can make something whizbang for tech-types, but that's not our goal. In fact, we're testing this thing on a close-knit group of 200+ senior citizens, most in their 80s and 90s, most of whom have never used computers at all.

Anyway, give me a call when you get a chance.