gravity7

3,500 movies w browse, fuzzy search, recommendations, mini film courses, personal taste reader, sorting by facets, tone, mood.

6,000+ inquiring lines across LLM research whitepapers w prompts for extended discovery. 6,000 topics, 1,800 hand-curated LLM whitepapers. Researchers' goldmine.

I've added the ability to paste an Arxiv url & view related research, topics, questions from my hand-curated LLM whitepaper excerpt Arxiv archive of 1400 papers. Just launched. Try it!

My Obsidian vault of ~1400 LLM whitepapers online - built w Claude to surface related concepts & questions across AI research. Work in progress.

Analyze articles & posts for reasoning, rhetoric, & target personas. Score them on the LLM smell test. Proof of concept.

Select nodes and view related research topics that cut across papers; view topics; see arxiv papers w urls. Built on my Obsidian vault of 2,500 white paper excerpts.

Interested in how LLMs write, what reasoning they expose, and what rhetorical styles they ab/use? (I am). Copy paste an example to analyze it.

Curious for feedback. Limited proof of concept recommender that uses semantic search on audience reviews only, no genres, dbase, hierarchy. Shows that LLMs work to extract relationships & match.

Right. And I think the right features will help to produce the conversations in which people will more likely say the things that the algorithms need ;-)

They're working on it and purchased Fridge to help get there -- as far as I understand.

Not often you see a Lacanian reference in this industry. It's more than data loss -- or rather, data is as often the product of interaction and a form of communication as it is just data. So preserving interestingness and relevance by sustaining attention, which is easily accomplished by action (not data), is I think possible. Features simply have to be designed to privilege communication over raw information.

Whether topical circles would enhance or dilute G+ is a matter of opinion of course. You might be right -- I tend to enjoy deep dives more than most.

But I see no reason, either in functional integrity or in social practices, that topical circles would diminish G+. Those not interested would simply not use them -- this is true of FB groups and lists.

It is incredibly difficult to achieve the network density and connectedness required to make a social network stick. Attention is in short supply -- witness Ning's failure to become the social network of interest groups. Perhaps it's not feasible for vertical and niche networks to match the success of the generic social nets. For this reason strikes me as in the ballpark to enable sustained topical conversations where the audience already exists, and where shared topical interests could augment individual profiles and deepen engagement.

I would never have thought of that! Now, if you could grab your content circle and drag that into an audience circle... But as convenient as that would for sharing, I suppose it would create a lot of noise!

But I wonder if unstructured forum isn't precisely the model for sustained conversations best suited to feed/stream interaction?

My sense is that time-based content consumption (streams) beg for solutions that allow activity to preserve and sustain the good stuff. This is essentially how FB updates work. But in FB there's no navigation back to active posts. Which is where I think G+ could provide a real solution.