Interesting article - and interesting approach to solve that issue also.
I think the basical problem behind all that is that we simply cant enough screen space to keep all the lookups we need in sight together. Even 4,5 or 8 screens like in obsidian is often not enough.
I’m a professional writer and journalist. Connecting (and interpreting) a lot of little facts is my daily bread. And for a single sentence like, let’s say, “elelctric cars are not zero emission in fact because of the carbon footprint of the electricity they use” will need up to seven single facts if you would want to state how much all cars in your country would emit still altogether.
You can do the lookups one after another and jot your findings down. In most cases that will happen like this.
An easier way would be if the right side would be a fully functional clone of your main page you can search in and from where you could just drag all blocks plus the needed linked resources and text resources from the pages over to the left, where they automatically turn into embedded blocks. You can bring them into sensible order then and have everything ready to start - almost painless and very frictionless research.
For myself I’ve switched to rely on my brains capacity though. Reviewing your facts in effective spaced repetition costs me 6.5 seconds (my personal average) × 5 reviews for each item I write, so roughly additional 30 seconds in the following 12 months to rely on my brain.
The speed gain in writing is enourmous then, when you can spare the whole time for typing in your query, scroll and read and then jot down the results. The initially invested 30 seconds pay off multiple times in the end, cause your brain just needs milliseconds to come up with the result, so almost in no time.
Plus you can play the wiseacre in discussions, add new facts to problemsolving, signal competence in interviews and can correct someone trying to sell you bs in interviews on the spot.
That’s why I think a simple spaced repetition function for blocks would be an invaluable gain in efficiency in Logseq.
To use and best support the much higher efficiency of brains is possibly still altogether a lot more efficient than any technical solution that tries to bring all facts in front of your eyes together. IMO that often gets too much neglected.
Plus, when you remembered things, they start to “simmer” in your brain, form connections and insights. When the facts appear on the screen you just start connecting them only the moment they appear. Databases and second brains are backup resources that add a noticeable retrieving time for information pieces - processing only happens in the first brain, and what’s already put into the working memory (RAM) has no retrieving time.
Researching things newly in browser is a totally different scenario though - and for this I find the approach with trails very interesting and definitely a great enhancement when I imagine it in action. I’m not sure though if it’d add so much efficiency gain in logseq. Dragging to embed from one side page seems more frictionless and faster, but that’s just me.