Hi Folks, I found this video a useful pointer to potential new features in Tinderbox (not just Roam). I’d find these capabilities useful if implemented in Tinderbox and wonder if others feel the same. If so, might they be candidates to include in the future roadmap please? Meet Roam-I and Roam-E, A Roam Research Concept Demo - YouTube Regards, JR
Do you mind explaining your proposal?
Watching the 30 minute video would probably not give us the right clue.
Hi Paul, although it is not short, I think watching the video is probably the most efficient way to understand the proposed ideas. I am not in any way affiliated with the video maker, by the way. In essence, there are 2 ideas. The first is probably easier to progress and involves recognising and presenting notes (and tags etc) that would be relevant or helpful in the context of a new note being created. The idea is about tapping the value of the accumulated notes already in the kitbag. I suppose the special sauce is about how relevant notes are recognised to avoid false positives. The second idea is about leveraging the broader Tinderbox community and tapping the power of the hive mind if or when Tinderbox users happen to be working on similar areas. It is about how other people’s content - with version control - can be linked to or reused. By the way, I noticed someone else has posted about the video I referred to already in the Tinderbox forum, so perhaps that person will have something more useful to add. Apologies but unfortunately I don’t have a lot of time to develop this - I’m merely reporting something I happened to come across that I think would be of value to Mark and the Tinderbox community to be aware of and potentially leverage. Cheers, J
1 Like
I watched some of the video but got confused and gave up as the commentary is very unclear. It all seems about doing things with blogs, pr blocks made out of other blogs.
It struck me as also a very programmatic/structured form of approach which actually seems a bit alien to Tinderbox’s strength in surfacing unseen structures, though I can see such an approach appealing to programmers and database folk.
Thanks for the précis. I don’t suppose I’ll find time to watch the video, but this suggestion of yours pops out. Isn’t Wikipedia the thing you describe?
(There seems to be a concept of some sort of mysterious universal enlightenment arising from use of the software in discussions among Roam-ists. I haven’t sussed out what’s going on, but maybe this video is part of that concept.)
1 Like
But surely all this fancy new AI is mainly used in suggesting (selling us) stuff we likely don’t need “We see you’ve just bought a [house], you clearly want to buy more [houses], we have these [houses] on offer”. Or, it is telling me things we already know: “you ate a sandwich today”. AI is actually remarkably dumb—outside niches like pattern recognition—without humans [sic] in the loop.
We’re now on AI generation 3.0 (4.0?) and still not reaping the immediate benefits—at least not in terms of knowledge. In fairness we are getting more ads on more channels which is good (somehow?). Machine Learning, a major part of the current crank of AI handle,) has limited understanding of human language, which we humans navigate with relative ease. I’m currently partly involved in a project finding that … ML is pants at understanding the semantic meaning of numbers in text (i.e. let’s use ML to find ‘data’ in social networks). Actually, you mainly find K-pop, data loggers and some pr0n.
What I’d like to see is more Engelbartian augmentation of my understanding of my data rather than black-box algos offering me guesses masquerading as certainty.
In fairness to Tinderbox, here it is 14 years ago† ‘doing AI’.
† I knew keeping old copies of aTbRef online had to be useful somehow. 
2 Likes